Impact of space charge on Perovskite Oxide based non-volatile memories

• Main Author: Zou, Xi.
• Other Authors: Wang Junling
• Format: Theses and Dissertations
• Language: English
• Published: 2013
• Subjects: DRNTU::Engineering::Materials::Microelectronics and semiconductor materials::Thin films
• Online Access: http://hdl.handle.net/10356/51349
• Institution: Nanyang Technological University

A recent study reported that the amount of information stored by humankind in 2007 was 2.9×10^20 bytes, which grows at an annual rate of 23%. Most of the information is stored in magnetic hard disk drives, which are bulky and slow. At present, Flash memory is widely used for portable electronics because of its low cost of production and the well-developed semiconductor fabrication technology. However, Flash memory has very low operation speed with ~10 µs programming and ~10 ms erasing time. Furthermore, it can only withstand ~10^5 rewriting cycles. These drawbacks prevent it from becoming the primary storage device in personal computers. To meet the ever-increasing demand for data storage, several non-volatile memories are being developed now, e.g. magnetic random access memory, phase change memory, resistive switching random access memory and ferroelectric random access memory. Our study focuses on resistive switching random access memory and ferroelectric random access memory, both of which involve oxides as the active materials. In resistive switching random access memory, the resistance of a metal-oxide-metal capacitor changes between two stable values, which are used to represent “0” and “1”. The switching is usually activated by an external electric field/current. In ferroelectric random access memory, the spontaneous polarization of ferroelectric materials is used to store information. An external field can switch the polarization between two stable directions. In both cases, the performance of the device is strongly affected by space charge activities in the oxides under electric field. We aim to clarify this relationship in this study. The uniqueness of our work lies in the planar device structure used, which allows us to conduct in-situ imaging of space charge activities in the active layer while electrical properties of the device are being studied. For the resistive switching random access memory study, we used Pt-(Ba0.7,Sr0.3)TiO3-Pt planar capacitors. The IV measurements were carried out and a typical bipolar resistive switching was demonstrated. Scanning Kelvin probe microscopy was used to observe the space charge activities during the switching. It was observed that, during hysteretic IV measurements, exposed positive charges appear at the reversely biased metal-oxide interface where the applied external bias follows the internal field direction within the interface Schottky barrier. By correlating the resistive switching with charge activities observed through scanning Kelvin probe microscopy, we propose a charge trapping/detrapping model to explain the switching phenomena. In the as-prepared device, interface oxygen vacancies are compensated by two electrodes. The device resistance is determined by the interface Schottky barrier. During the IV measurements, the reversely biased Schottky barrier is under large electric field and electrons are detrapped from oxygen vacancies. The exposed oxygen vacancy lowers the Schottky barrier and turns the device into a low resistance state. This effect should exist in most metal-oxide-metal devices, though it may not be the dominating factor. It is very important for researchers to take note of this effect when analyzing their experimental observations. Further investigation using planar metal-oxide-meal devices should lead to a better understanding of the resistive switching phenomena. The investigation of ferroelectric random access memory focuses on the fatigue of ferroelectric polarization after repetitive electrical cycling. BiFeO3 and Pt were used as the switching element and electrode, respectively. The combination of scanning Kelvin probe microscopy and piezoresponse force microscopy allows us to monitor both the domain evolution and space charge redistribution during polarization reversal. It was observed that charged domain walls do appear during electrical cycling, but they do not cause fatigue as suggested by a previous study. Instead, after repetitive electrical cycling, high energy electron injection occurs at the interface between BiFeO3 and Pt electrode. Defects deep in the forbidden band are thus ionized by the injected electrons and cause domain pinning. As electrical cycling continues, more electrons are injected and the pinned domains grow into the center of the ferroelectric layer, eventually leading to fatigue of the device. In the charge injection model, the interface Schottky barrier prevents the injected electrons from diffusing back to the electrode. It is thus expected that by using materials with small work functions as the electrode, fatigue can be reduced. We performed the same electrical cycling on planar devices using Fe and (La0.7,Sr0.3)MnO3 electrodes. The low work function Fe yields lower Schottky barrier whereas (La0.7,Sr0.3)MnO3 gives rise to flat band condition. Indeed, the piezoresponse force microscopy at 10^10 cycles shows negligible domain pinning in the Fe and (La0.7,Sr0.3)MnO3 device. Negligible charge injection is revealed by scanning Kelvin probe microscopy in both devices. At the same number of switching cycle, Pt device has already developed significant charge injection and domain pinning. Based on our study, we suggest that electron injection, not oxygen vacancies, causes fatigue in BiFeO3.