Interesting. One thing that I think would help, especially amid the fact that NVM is still in its infancy despite decades of development, is if companies would standardize on the terminology. I know everyone thinks they have no true competitors and that their product stands apart from the crowd, but for a new technology to catch on, people need a way to compare apples to apples or even apples to oranges.
It finally sunk in! What is called "memristors" by HP are indeed nothing else but memory devices utilizing "resistance switching" effects. Such effects are well known since decades and are in no way related to the hypothetical concept of non-volatile memristor/memristive systems. It might turn out that the whole "memristor" story needs some critical reflection.
But I think we at EE Times have been making this point for some time.
It remains to be seen whether there is something in the detail of HP/SK Hynix memristor structure that differentiates it from other resistence-switching two-terminal devices.
When taken literally ReRAM is a broad classification of non-vol memory. While often used to refer to metal/metal-oxide layered structures it could also be used to identify phase-change memory and other forms of memory. This all tends to depend on common usage, which is still developing.
As a result ReRAM tends to be used to refer to metal-oxide or silicon-oxide layered structures
Conductive bridging RAM for forms of memory based on movement of metal ions to form bridging filaments
Phase-change memory for memory based on thermally-driven phase change in chalcogenide layer
And memristor for the HP technology, because it is HP's choice of terminology. HP disclosed a memristor based on titanium oxide layer, which suggests it is in the metal-oxide ReRAM camp.
Dylan, I agree. Usually, RRAM or ReRAM papers will present commonly used data to show memory-related performance. These should include (not comprehensively): 1) a basic I-V showing the SET (High->Low) and RESET (Low->High) transitions at typical current and voltage, 2) Switching for a respectable number of cycles (100 at least), 3) Some indication of stability under high temperature or voltage stress, 4) a plot of distribution of resistances for different states, 5) Schematic or photo of structure showing mechanism.
For a memristor demo, it usually involves sweeping the device to form closed figure-8 curves at different rates, and looking for gradual dynamic changes of resistance. By definition, this device is very easy to disturb (even a small current should do it), so can't be a reliable memory at all.
It is really astonishing that the connotation "ReRAM <-> memristor" is still spooking through the scientific scenery. A scientific justification, i.e., a valid physical model, for the existence of "memristors" is missing so far. Maybe, the whole "memristor" concept is not consistent with fundamentals of physics.
The stochastic nature of the complex resistance switching effects sometimes involves current-voltage characteristics which might remind of the pinched hysteresis loops that are thought to be the fingerprints of "memristors". Such observations, however, are not sufficient to claim that a "memristor" has been found.
Agreed . Why does there have to be a memristor pnenomena to match the dielectric equivalents? It could well be that all the benefits that we obtain from electrical conduction and electronics are only possible because the claimed proerty does not exist. Another way of saying, " Maybe, the whole "memristor" concept is not consistent with the fundamentals or physiscs".
For marketing, PR and money raising purposes call your resistive memory what you like and play all the curve matching games you want to, especially if it can be shown to work in a real, competitive and useful sense. The real measure of success is design-in wins and profitable application. I used to park my car in a public car park where some graffiti artist had written on the wall in large permanent letters "DON'T BELIEVE THE HYPE", words worth rembering in the ressitive memory game.
Another subtle difference is the ReRAM has to be limited from irreversible breakdown by a transistor, diode or other resistor in series. While the memristor is at least theoretically an independent entity.
I suppose that most of the former "memristor" enthusiasts are now aware that they have been taken in by an erroneous publication. Quite simply: The NATURE paper "The missing memristor found" (Nature 453, (2008) 80-83) is based on severe physical misconceptions. Some in the scientific community seem now to catch on to these problems (see: "On the physical properties of memristive, memcapacitive, and meminductive systems" by M. Di Ventra and Y. V. Pershin)
The editorial board of the journal NATURE should actually know how to re(tr)act in such cases. Or they are merely hoping that the whole "memristor" hype will sink bit by bit into oblivion.
NASA's Orion Flight Software Production Systems Manager Darrel G. Raines joins Planet Analog Editor Steve Taranovich and Embedded.com Editor Max Maxfield to talk about embedded flight software used in Orion Spacecraft, part of NASA's Mars mission. Live radio show and live chat. Get your questions ready.
Brought to you by