Electronic Noise in Digital and Analog Transition Metal Oxide-Based Memristors

点击翻译按钮获取中文摘要

Memristive systems are currently intensively researched as the next generation of non-volatile memory devices that are needed to overcome limitations in computing. Special interest is given to analog devices due to their ability to mimic synaptic behaviour or establish multi-state data storage. The particular advantage of transition metal oxide-based resistive random-access memories (RRAM) over other emerging memories lies in their excellent scalability, high switching performance and their compatibility with complementary metal-oxide-semiconductor (CMOS) technology. However, the reliability of such devices is limited partially due to electronic fluctuations that may cause false readouts in memory applications and resistance-state broadening-related unreliability in synaptic operations. The goal of this proposal is to lead a comprehensive investigation on the nature of such fluctuations in three material systems, namely undoped and La-doped hafnium oxide and pure yttrium oxide. Within the proposed research, a novel sample series will be produced utilizing defect-engineering via reactive molecular beam epitaxy to induce oxygen vacancies in the functional layer. Oxygen defects in these materials are created in three ways: (i) as a direct result of oxygen-engineering, (ii) through substitutional doping or (iii) arising intrinsically as a result of the crystal structure. The fabricated sample series will be comprehensively studied in terms of the materials and structural properties, as well as the electrical characteristics under DC and transient operation conditions. In the main part of our investigations we will perform fluctuation (noise) spectroscopy measurements over the whole range of current-voltage characteristics using fast data acquisition cards (FDAQs). Moreover, we aim to develop an automatized fluctuation spectroscopy measurement scheme titled Continuous Analysis for high-throughput data acquisition and subsequent processing and analysis. We will apply the framework of Continuous Analysis to the three material systems in order to attain a comprehensive physical picture of the microscopic transport processes during resistive switching with the final goal of correlating materials properties with the arising electronic fluctuations ('noise engineering'). In this first-time noise characterization of analog memristive devices, we will put special emphasis on the effect of stoichiometry, and also on the intriguing phenomenon of conductive filament stabilization during DC and pulsed training. The impact of the proposed work lies in providing valuable information not only for fundamental physics but also for targeted material's engineering of artificial synapses and data storage applications.