Memristor 'brouhaha' bubbles under

1/16/2012 12:20 PM EST

Resistive memory devices out of Titanium Oxide were published in the early 1970s. Samsung does not own the basic US Patent on that.

1/16/2012 1:45 PM EST

At first site this appears to be the “How many angels can dance on the head of a pin” debate. To me what is more important is to try and get into the market place some NV memory products that are competitive in price and performance and/or offer some unique property. I think one should always investigate the motive behind this type of argument. Is it a sign of the battle lines being drawn for the fight as to who owns what piece of IP? Or has somebody discovered that one of the memory types they have been promoting has mechanisms that overlap with others and they may not be as well protected, as they previously believed. If so, I happen to think Ovonyx/ECD will win that battle. Although I side with Blaise, reading the paper, there is clearly a need to separate the volatile threshold switche from NV phase change.
I do not think I am alone with the view that names/mnemonics should try and convey as much information as is possible with respect to what is happening inside the memory, as far as we know.
My own choice is PCM and not PRAM or PCRAM the latter is built-in a sales pitch for potential applications. If PCM is ever successful from where we are now, as well as scaling it will need to be able to serve as an NVRAM i.e. as a universal memory and that will be its unique selling point over Flash, it is just possible it could even win that battle without scaling.

1/16/2012 2:31 PM EST

Intel's claim was always 'creative', and seems to have much more input from marketdroids and patent lawyers, than the designers, or reality.
It is no surprise to now find a game of patent semantic gymnastics is the motivation.

1/17/2012 10:51 AM EST

On the subject of establishing differences. I think the electric field driven drift of oxygen ions in the oxide based memistors, is of interest because if the ions move in to unfilled locations in the atomic structure without disturbance and without even short range order changes then it is difficult to see how that can be considered a phase change. If the long or short range order is changed then that is possibly a phase change; although not necessarily a crystal/amorphous one. Given memory device PRODUCT SUCCESS,the determination of the difference or sameness will no doubt make some lawyers rich.
If there is a need for a forming process in the oxide memory and it involves a phase change in my view it will make the memory of little consequence. While in the past a forming process was OK for single capacitors it is of little interest in a 1G-bit plus memory array.

PS In my other post here I do know the difference between "sight" and "site". I will not argue that there is a use if site as a scene. It was a typo as with "switche"

1/17/2012 4:35 PM EST

I have been researching ReRAMs for 5 years now. When the Memristor concept came out I ran to read Chua's paper and immediately realized the some error as described in B. Mouttet's paper. In terms of a dynamic system, the Chua's paper provides no memory effect. Specially the claimed non-volatile memory effect. The models thaet have been published are a mockery of science. Metal-Insulator transitions are known for years; It started in 1937 with Wilson, and through the 1950's with Mott, Hubbard, Anderson, Gutzwiller, Brinkman and Rice, among others. Mot got a Nobel prize for that. Common to all this are the d-block transition metal oxides, like TiOx, CoOx etc. And some perovskites too. The coordination number changes due to electron tunneling into a coulombic interactive material such as TiOx have been confused with "filaments", oxidation-reduction reaction and a myriad of pseudo science. Understanding the "memory effect" is still a big research area in physics, but there is plenty evidence of a real and useful effect. I have no filaments in my devices, for that is a defect, not a mechanism for storage (see May 1st ,2011 issue of Journal of Appl. Phys. - cover story). If you have an ultra thin device such as HP 's (30 nm), the interaction of vacancies and hopping electrons coming from Ti, makes pretty pictures, but it is a highly unstable device and not a good memory. Detailed Ligand compensation is where the story is. This is a "many-body phase" device, and not to be understood with "classical" transport or single particle quantum mechanics as Silicon.
I suggest that next time, before running to the press, run to the literature. Now, if HP decided to expand the memristor (terrible name BTW) to include nonvolatile memories, although being a real stretch , it is actually not a bad classifier, however that is not where Chua's paper started. He has a transient out of phase "pinched" lisajour picture.... caraujo@aol.com/Prof. Carlos Paz de Araujo

1/17/2012 6:05 PM EST

From one of my comments above while I agree with Blaise Mouttet that the memory types are different, I tried to make the point made by Prof Carlos Paz De Araujo in the post above. I stated, "reading the paper, there is clearly a need to separate the volatile threshold switch from NV phase change." It appeared to me that the Mouttet's paper was concentrating on the threshold switch and ignoring the phase change memory.I suppose one way of distinguishing the types of transition, might be by grouping them as thermal and other transitions
I thought HP were claiming the memristor was a non-volatile memory, or have I missed something?

1/18/2012 6:37 PM EST

Watching this topic over the years, I've been puzzled by the firefight between two armed camps: "it's a new thing" and "it's not a new thing". I just don't understand all of the heat.


However, what does it matter? If the devices are useful, fill a niche, or enable something else, they are valuable. If they really aren't useful by reason of cost, or performance, or what-have-you, what does it matter how a memristor is classified?

1/19/2012 10:49 AM EST

I would like to add from private correspondence with with Blaise Mouttet, he admonishes me:

"Incidentally, it is trivial to convert example 4 of my paper from a volatile to a non-volatile dynamic system by including a crystallization rate equation thus forming a 2nd order dynamic system. This has been discussed in phase change modeling literature by David Wright and in terms of memristive systems by DiVentra and Pershin."

He is also of the opinion:

"This is not simply a debate on semantics or classification of "memristor". In my view this entire memristor fiasco is a direct attack on the scientific method. If scientists are going around claiming they have discovered an example of a memristor (or any other theoretical concept, e.g.
Higgs boson, etc.) they need to rigorously prove that their claims have scientific validity and show comparisons between experimental data and theoretical models. This has not been done for the so-called "memristor" claims of HPLabs and it is a(expletive deleted)disgrace.

1/19/2012 10:54 AM EST

Here is a peculiar Ge-Te-Sb materials development. Maybe one of you techies can translate it into popular science dialect:

http://www.nanowerk.com/news/newsid=23999.php

1/20/2012 11:17 AM EST

HP has responded to the discussion about the memristor. See http://www.eetimes.com/electronics-news/4234963/HP-responds-to-memristor-debate

5/4/2012 10:50 AM EDT

I recently did a SPICE simulation to show that a linear resistor in parallel with a non-linear capacitor can also produce the pinched hysteresis effect (see link).

http://vixra.org/abs/1205.0008

Thus pinched hysteresis cannot be considered as evidence of a "fourth fundamental circuit element" as claimed.