The filamentary mechanism for ReRAM/CBRAM operation has been widely discussed here and elsewhere and is becoming the ‘accepted wisdom’ for a well defined set of active materials and electrodes. Broadly speaking, these devices can be characterized by the on resistance being a function of the compliance current (the current limit during the ‘on’ voltage pulse) and being insensitive to the area of the device down to the size of the filament which can be 10nm or smaller. A second mechanism has been long proposed for a different set of material systems which I will attempt to describe here! The idea is that the active material switches from an insulating state to a conductive state under the action of an external stimulus which could be thermal, magnetic or a voltage pulse. This is sometimes referred to a Metal-Insulator (or Mott) Transition (MIT) and the materials classified as Correlated Electron materials.
An example from my previous life is the thermally induced transition in Vanadium Dioxide. At low temperatures the stable monoclinic crystal structure of VO2 is an insulator. Heat it up (to >80C) and a phase transition to a tetragonal crystal structure with metallic properties occurs. The resulting resistivity change is over three orders of magnitude. This phase change is reversible simply by cooling back to below 20C. Indeed, I was working on this material in a thermal imager (IR camera) that attempted to exploit this huge change in resistivity with temperature. The same resistivity change can be induced electrically and unlike ReRAM materials, the material reverts to a high resistance state once the voltage is removed. This of course makes VO2 uninteresting for memory applications directly. However, engineers from Samsung have combined a Pt/VO2/Pt switch element and a Pt/NiO/Pt memory element in series to reduce inter cell leakage. By applying a voltage higher than threshold of the switch element (0,6V in this case), the switch element reaches the on state and the cell can be accessed. This behaves similarly to the MIEC access device developed by IBM (see Blogs passim).
At the recent NCCAVS Memory Workshop, a presentation by Seshubabu Desu, Dongmin Chen, Lee Cleveland, and Jean Yang-Scharlotta, 4DS described their approach to ReRAMs (MOHJO™) based on strongly Correlated Electron materials. such as Pr1-xCaxMnO3 (PCMO). 4DS describe MOHJO™ as ‘based on Mott metal-insulator transition driven by field induced non-linear transport of oxygen vacancies across the heterojunction’. They emphasised this is not a filamentary process and that oxidation of the top electrode is observed, i.e. A reactive top electrode (examples with Ti, W or Ta are described) is necessary for memory operation. In support of their MIT mechanism they show a weak dependence on compliance current (see below from the 4DS presentation) and a scaling of the low resistance state with device area. 4DS see field driven ion migration across the hetero junction formed by the oxidised metallic electrode and the PCMO as being associated with the resistance change.
I must confess to a slight reservation in that vias fabricated at these dimensions for these ReRAM devices are not the ideal rectangles as shown in cartoons and schematics. Thus across the via, there will be regions where the electric field is stronger and field induced migration will greater so low resistance paths could be created. And of course, oxygen ion migration is an accepted part of the filamentary formation process in many materials. However, it is true that the PCMO’s are completely different materials and there is always the possibility of a lateral diffusion which could ‘even out’ any areas of high concentration.
Happy New Year to one and All!
Christie Marrian, ReRAM-Forum Moderator
ps Many thanks to ReRAM-Forum registered user, Dr Carlos A. Paz de Araujo, co-founder of Symetrix Corporation and Professor of Electrical Engineering at the University of Colorado, Colorado Springs. Carlos passed on an interesting review article on Correlated Electron Systems by Sieu D. Ha, You Zhou, Rafael Jaramillo, and Shriram Ramanathan from Harvard University, Cambridge, MA 02138, U.S.A. published in Future Trends in Microelectronics: Frontiers and Innovations, edited by Serge Lviryi, Jimmy Xu and Alex Zaslavsky