Breaking News
Comments
Newest First | Oldest First | Threaded View
Page 1 / 6   >   >>
eavedrop44
User Rank
Rookie
Re: ReRAM
eavedrop44   4/21/2014 4:26:33 AM
NO RATINGS
I love what youve got here, love what youre saying and the way you say it. You make it entertaining and you still manage to keep it smart. I cant wait to read more from you. This is really a great blog .
BlogHuts.com


nonvolatile
User Rank
CEO
Re: nickel vacancies
nonvolatile   10/17/2013 4:38:01 AM
NO RATINGS
The ratio is adjustable by doping, but a good number is 10,000 to 1. It is important to have a good Ron that is not to low, as this and Imaxon need to be well controlled. A lower than 100 microamp Imaxon is prefered. Doping for higher Ron, can be easly acomplished by "elemental"doping and does not disturb the the CO doping that heals the coordination sphere of Ni. This is another important aspect - after CO doping, the material is clean, and then one can dope for scateering in the on state independently of anything. It is this bulk control, which cannot be achieved very easily if filaments are just being connected/disconnected in some microscopic area, that makes the coordination control doping so elegant. Thank you for your question.

PhyandEE
User Rank
Rookie
Re: nickel vacancies
PhyandEE   10/16/2013 11:46:49 PM
NO RATINGS
What is the resistance switching ratio? Is it uni-polar or bi-polar switching?

nonvolatile
User Rank
CEO
Re: nickel vacancies
nonvolatile   10/15/2013 7:47:15 AM
NO RATINGS
Thank you for your question. We can see that when the CO level, introduced in the oxide by wet chemistry (i.e., already in solution in the Ni precursor, for Ni(CO)4 is extremely poisonous and not added to the precursor as such-please, anyone trying this DO NOT use Ni(CO)4, it kills), the resulting material is always in a metal-like state. This is very much thickness dependent. Thus, it is not clear if vcancies has anything to do with the metal phase, as a vacancy would be completely screened by the surrounding electrons in the Drude regime. Now, in the less conductive case, it is then possible to have a reversible Mott transition due to the disproportionation reaction of Ni(+2) to Ni(+1) + Ni(+3). The CO doping technology is able to make the disproportionation reaction to be the dominant reaction, which is specifically the charge species that can be reversed without the need of filaments. Because disproportionation reactions are simultaneous oxidation/reduction type of reactions, it can be easily confused with a redox reaction. Redox reactions can occur in severely damaged materials, but they are not very reproduceable. Well then, the CO can be electron donor via a pi bond, and an acceptor in the sigma bond. The other two defect reactions in NiO are Ni(+2) + Vox(+2) --> Ni(+4) + Vox(0){two trapped electrons} and Ni(+2) + 2 e(-) -->Ni(0). The first reaction with oxygen vacancies creates a peak of Ni(+4) in the XRD spectrum. And the metal Ni(0) also shows up in XRD. Both peaks disappear with CO doping, but they are dominant in non-doped materials. Thus, it seems that the pi bond is dominant. It is a possible that in the filamentary versions of NiO, All species occur and we have a situation that the insulating phase dominates at the small areas (the area dependence is very strong, for other reasons). Thus, when the soft breakdown occurs, the filament is strongly Ni(0) and surrounding it, there are many Vox's. Where the filament is disconected - this is seen mostly around the anode - we believe that the "CeRAM effect" (i.e. - a disproportionation/Mott transition) may be occuring to bridge the metal-like filament to the electrodes and create the metal like state of the ReRAM. Extending this to other transition metal oxides, has been achieved by our group.

Thus, I hope that this gives you the general idea. The problem of all ReRAM is the Roff state being unstable with temperature and dispersed in the cummulative probability of good bits. Thus, so far, the dispersion of the off states, specially after many write pulses makes the "freak bits" population too high for commercialization. The other two areas are the high on current and the Vset/Vreset dispersion -specially the Vset dispersion. All these things are extremelly reduced in the CeRAM approach.

As we make a 3 layer structure with Heavy doping/10 nm of low doping/heavy doping, we completely screen the active middle region from the electrodes and be electrode agnostic (CoSi, SiNi, Al..), we have cleaned up the material to have only the right species, lowered Schottky effect, and we are "born on", that is, it is already ohmic as built. The details of this will be coming as we make our next press release - probably late November. (When I said 2 months, it was mid september, so we are delayed a little) - what is holding us back? - everything, including many activities such as ALD and difficulties in nanoscale areas. But, it is going. Thanks for your interest.

 

resistion
User Rank
CEO
nickel vacancies
resistion   10/10/2013 3:36:06 AM
NO RATINGS
I just recalled some papers mentioning nickel vacancies in NiO, maybe CO doping eliminates O vacancies but adds Ni vacancies?

nonvolatile
User Rank
CEO
Re: ReRAM
nonvolatile   10/9/2013 11:58:01 PM
NO RATINGS
It is not possible to look at this matter with the eyes of an indepemdemt electron description. That is, without the ability to show that the change in oxidation number is the result of electron localization, which causes the repulsive coulombic force responsible for the charge gap, one can go around circles. The disproportionation reaction is responsible for the charge gap. It is completely not dependent on position of the Ni ion, so that, after the defects such as oxygen vacancies are compensated by trapped electrons, it is totally possible to have a local MIT which would complete the filament or break it. This can be easily be missed measured as a redox reaction, if you are just checking the oxidation number -like EELS. So, it is not an easy thing to be 100% sure. Other approaches in the electrical measuremnt must be taken. One example is to measure the capacitance of the insulating state as a function of frequency. You can immediately see that the low frequency CV plot is hysterectic due to interfacial polarization and this causes the Vset and Vreset  divergence. This also show that the many oxydation states on the surface of a non compensated NiO is to blame. On the other hand, the compensated NiO is a smooth CV plot without hysteresis. 

I am not at liberty tight now to disclose a few other things. This will come out in due time. But, as we all know, the conductivity is always proportional to DOS. If you use the independent electron approximation (typical of metals and semiconductor), you can never show the f=non-rigid DOS afforded by TMOs. Only if you let DOS=0 and DOS= finite, that you can show the teo states - insulating and metal. Thus, it does not matter if you have filaments or not to have these two states. Somewhere in the filament model, where the filament is broken, an oxidation number-disproportionation -reaction will contribute to the closing and opening of the filament. Thus, as Ron says, this is a conundrum, until you get to reliability. Then, when you see your off state unstable with temperature and your Vset/Vreset variations, you would ask, How cab I control this? well, with the right device physics. Which then requires the stable material to satrt with. Off coarse we do not erradicated all defects, but enough to have a better device and one that we do not have to breakdown the oxide and have these filaments inside the material as the only means of switching. Thus, let us say that we have "quantum scale" filaments....still, how do the close and open? with a in-loco disproportionation. So, why not make the bulk all like that to start with. 

Honestly, we can dope the device to the tipping point that we have a soft electroforming like behavior - sort of born-off. One could interpret that such a device has Nanofilaments. Thus, after a slightly higher voltage than the norma set voltage, the device works just fine. But, if you go a litlle more in doping, it is no longer the case and you have a born-on device. So, the ability to go from always conductive down to just insulating enough to need a little push, is for me a clear demonstration that we have a point. Doping can eliminate the need for filaments. Thus, I am  not so worried about this. It may even be possible that the filaments are doped grain boundaries. Still, doping with CO passivates the device and yields control.

PhyandEE
User Rank
Rookie
Re: ReRAM
PhyandEE   10/9/2013 10:54:09 PM
NO RATINGS
Nonvolatile said my HRTEM comment on the filament observation is rather speculative, Resistion said Nonvolatile is priori in his NiO devices, and Ron Neal said it is like a Conundrum.

It is interesting. I have collected a bunch of papers on this topic. In the reports, filament size ranges from microns to nano, and to atomic scale. HRTEM and CAFM are the two facilities frequently used. There are also some theoretical modeling works.

No matter what, as Nonvolatile has said, DATA talks, I think we should see the data first. The outstanding device performance as described is a good news to the resistive switching study. I have to mention that although it is difficult to observe the filaments through HRTEM, it does not mean that filament should or should not work. For this reason, I think the work by nonvolatile has good chance to be accepted. It is not the kind of works by HP.

Ron Neale
User Rank
Blogger
Re: ReRAMa A Conundrum
Ron Neale   10/9/2013 6:27:49 PM
NO RATINGS

Resistron a Conundrum in the limit can filaments switching be bulk?

If the size of the area of a filament type memory device is reduced to the point where the filament is the same diameter as the device electrode area, can that now be considered as a bulk switching effect?

If the switching depends on the collapse of the pre-switching current into a nucleating point then it might be expected that the contact area of such a device would always need to be larger than the filament. These would be devices where thermal effects play a major role in the switching event.

However, in a device where conducting precursors determine the location of the filament (i.e. devices that normally require forming) the location of the filament would be positioned stochastically.

In those devices where defects allow some form of true bulk switching effect in a region close to the surface of one electrode with the defects or metal precipitates acting as a series resistor carrying most of the current through the bulk which have been alluded to here in the comments. Is a type which might be considered filament for large area and might claim to be bulk with scaling but as the electrode area is reduced will only work if the required defect path is certain to be present.

Would a multi-filament device where the set current would determine the number of filaments that come into play (weakest first) be considered a filament device or when the total area is involved a pseudo bulk switching device?

 

 

resistion
User Rank
CEO
Re: ReRAM
resistion   10/9/2013 7:53:04 AM
NO RATINGS
With all due respect to your MIT model, I'd caution against completely dismissing the filamentary devices, at the very least because they are so prevalent. And can you guarantee no doping or thermal non-uniformity or defect-free device? Your model is already based on insulator phase appearing locally at one electrode. The burden is on anyone to prove they are free from defects, rather than assume so a priori.

nonvolatile
User Rank
CEO
Re: ReRAM
nonvolatile   10/9/2013 12:58:57 AM
NO RATINGS
The paper by Meuffels and Soni, does debunk the HP model with an electrochemical argument. Actually in two lines we can show that the HP model is not a physical model. There is no way to show a Mott transition with a classical argument though. The paper lacks on that. Now the original Memristor concept is a naive mathematical explanation of a circuit element. It lacks physical understanding of the meaning of conductance In an ultra thin TMO film. The original paper also had no intrinsic long term retention. It was a phase difference misinterpreted as memory. Although paper by Meufffel and Sony correctly addresses the gross aspects of the HP model, it fails in explaining how such a device would work if TiO was more stoichiometric as in other RRAM devices with filaments. The evidence is strong of MIT in such devices. Where HP went wrong is that the two fluid model was modeling a bad device and not what a true Memristor, If such a thing is allowed to exist, should not be based on a circuit element, but instead on a model of TMO conductivity based on quantum mechanics. Conductivity without scattering has to start at least with Lubo"s formula. Then at mesoscopic scale and non rigid density of states, the current has to be calculated completely different from a simple Drude"s model! Not Boltzmann Transport either. The TiO layers were 30 nm thick. Ballistic transport with e-e interactions and e-oxygen vacancy interaction should have been used. None of these approaches were used. I was amazed that a drift-diffusion two fluid model was used. Vacancy flow had only the effect to trap or de trap electrons and if they moved they had no way to flow as electrons into the electrodes. So what was understood as an MIT with crazy hydrodynamic explanations was nothing but charge trapping with some movement of defects aided by the tunneling current. A simple test of retention over temperature for different pulse width would show low activation energy of the off state and low memory effect with high temperature. I can even guess that the activation energy would be in the order of 0.35 eV, typical of titanates oxygen vacancies de trapping. Very naive pele did that work. Poor Hynix, I wonder if they will ever make a device, I do want to leave it here for the record that if the Titanate is balanced as we did with NiO, it does work as a CeRAM. We have beautiful data on that. In fact we made a 10 nm thick titanate active region and it works just fine. We will present both NiO and TiO data in our papers. ,

Page 1 / 6   >   >>


Flash Poll
Top Comments of the Week
Like Us on Facebook
EE Times on Twitter
EE Times Twitter Feed

Datasheets.com Parts Search

185 million searchable parts
(please enter a part number or hit search to begin)
EE Life
Frankenstein's Fix, Teardowns, Sideshows, Design Contests, Reader Content & More
Max Maxfield

Energizing the Young Engineers of Tomorrow
Max Maxfield
15 comments
It doesn't seem all that long ago when I was a bright-eyed, bushy-tailed young engineer. Now I feel like an old fool, but where are we going to find one at this time of the day (LOL)?

Jolt Judges and Andrew Binstock

Jolt Awards: The Best Books
Jolt Judges and Andrew Binstock
1 Comment
As we do every year, Dr. Dobb's recognizes the best books of the last 12 months via the Jolt Awards -- our cycle of product awards given out every two months in each of six categories. No ...

Engineering Investigations

Air Conditioner Falls From Window, Still Works
Engineering Investigations
2 comments
It's autumn in New England. The leaves are turning to red, orange, and gold, my roses are in their second bloom, and it's time to remove the air conditioner from the window. On September ...

David Blaza

The Other Tesla
David Blaza
5 comments
I find myself going to Kickstarter and Indiegogo on a regular basis these days because they have become real innovation marketplaces. As far as I'm concerned, this is where a lot of cool ...