Saturday, February 25, 2012

New Scientist screws up reporting on memristor debate

New Scientist recently erroneously reported on an alleged "online spat" regarding memristor papers posted on ArXiv by myself and a group led by Hyongsuk Kim (link). EETimes gave slightly better coverage but still gets it wrong (link).

The subject of the so-called "spat" is whether or not the term "memristor" as originally defined by Leon Chua in 1971 is justifiably applicable to ReRAM, phase change memory, or other forms of 2-terminal memory resistors.

My position is that there is insufficient scientific proof that the memristor as mathematically defined by Leon Chua in 1971 or, as more broadly defined by a 1976 paper by Leon Chua and Sung-Mo Kang, is a correct model for ANY form of ReRAM, phase change memory, or MRAM as claimed by Stan Williams of Hewlett-Packard.

Hyongsuk Kim's position (according to my understanding) is that it is inappropriate to call a device a memristor or memristive system unless it meets certain mathematical criteria regarding invariance of the zero-crossing hysteresis effect when the characteristics of the input signal are changed.

Therefore my position and Kim et al.'s position are not that different (although I believe Kim et al. misunderstands the point of my paper). In addition, there is some evidence from the literature of TiO2 memory resistors that, given Kim et al.'s argument, HP's so-called "memristor" can not be considered a memristor or a memristive system.

This may seem like a mere trivial mathematical argument to some but if the wrong models for ReRAM are being proposed and accepted by the scientific community this will hinder technical progress and innovation in this area resulting in a very negative economic impact if not remedied.

My formal response to Kim et al.'s paper is posted at this link.