Technical: Resistive switching in oxide-based heterostructures (metal-oxide-metal) has been known since the 1960's and has experienced a recent (since 2000) explosion of commercial and scientific interest, particularly for nonvolatile resistance-change random access memories. Resistive switching can be defined as the process in which a material or heterostructure is able to reversibly change (under electrical stimulation) between at least two resistance values that are stable with time. Most of the competing mechanisms proposed to explain resistive switching are qualitative and have only indirect observational support. However, they nearly all involve the redistribution of dopants in the oxide layer through ion motion. The project will interrogate in a quantitative fashion the electrical potential and oxygen distribution in metal-oxide-metal heterostructures, to quantify the oxygen motion within the oxide and across the metal-oxide interface, and develop a predictive computational model that captures the kinetic aspects of the switching process in three dimensions. The integrated research and education program described aims to understand, design, and produce nanoscale resistive switches appropriate for data storage and logic devices. Testing of the hypotheses that uniform ion redistribution within a heterostructure can be quantitatively characterized and ultimately optimized for resistive switching will be studied. To do so the research will determine the distribution of oxygen near, measure the oxygen exchange rates across, and quantify the electrical potential near metal-oxide junctions and oxide heterojunctions, designed and fabricated at Carnegie Mellon. A variety of characterization methods, including secondary ion mass spectroscopy, electron holography, scanning Kelvin probe microscopy, scanning electrochemical microscopy, capacitance-based methods, and spatially resolved secondary ion mass spectroscopy to interrogate ion redistribution and transfer will be used. A multi-dimensional continuum level computational model will be developed that describes the electrochemical transfer and ion redistribution in the resistive switch, which will be verified by testing of nanoscale resistive switching metal-oxide-metal devices, appropriate for future applications.

Nontechnical Abstract

This project will have a broad societal impact by providing the fundamental underpinnings to device design and performance in novel ultra-high density data storage devices and reconfigurable logic technologies. The research activities will impact the education of a diverse group of Ph.D. and Undergraduate students through: laboratory and computational research, an advanced elective course, and a sophomore-level laboratory module. The new elective course, entitled 'Defect Interactions: Enabling Advanced Functional Materials and Devices,' will provide concrete examples of how defect interactions (such point defect motion near / across interfaces in memristors) act as enablers in advanced devices, including resistive switches, batteries, fuel cells, and high power electronics. A laboratory module will be implemented in the sophomore level core materials science course 'Defects in Materials.' It will demonstrate to students how point defects behave in electrochemical materials. Finally, the PIs will organize a research symposium on resistive switching at an international research conference.

Project Report

The project goal was to understand the functioning of a novel electronic device: a programmable resistor called a memristor. The name is a hybrid of two words: RESISTOR with MEMORY. Such devices can exsit in two states: an ON state which is conductive and an OFF state which is not. The states are nonvolatile (i.e. they retain their properties after the power is switched off) and programmable by application of voltage. Such devices are made by sandwiching a thin layer of oxide (titanium, tanatalum or hafnium oxide) between two layers of metal and can be used as memory banks. Until now, it was thought that the conductivity in the structure was due to accumulation of defects in the oxide layer: the electric field would induce ions to move leaving vacant sites in the material. This project showed that the origin of switching is different. In oxides, the current flow spontaneously constricts to a very small filament. This, in turn, locally heats up the material, lowers the reistance, and causes lasting changes in the distribution of atoms. These properties make it possible to fabricate very small oscillators (devices in which current repeatedly constricts and expands) that can serve as building blocks in computers that act like a human brain. We have also demonstrated an inexpensive method of fixing the location of the current filament. It can be accomplished by locally irradiating the device with electrons in a manner similar to that used in the TV tube. The students who worked on the project learned fabrication and testing methods of electronic devices and are now working in electronics companies.

National Science Foundation (NSF)
Division of Materials Research (DMR)
Application #
Program Officer
Z. Charles Ying
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Carnegie-Mellon University
United States
Zip Code