REGISTER | LOGIN
Breaking News
News & Analysis

Fujitsu Is Licensee of Nantero's Carbon-Nanotube RAM

9/1/2016 11:27 AM EDT
16 comments
NO RATINGS
More Related Links
View Comments: Newest First | Oldest First | Threaded View
Page 1 / 2   >   >>
Ron Neale
User Rank
Author
Re: random CNT filaments
Ron Neale   9/16/2016 7:48:26 AM
NO RATINGS
Resistion;- If you mean for the mixed NiO-Carbon fabric NVRAM (nano tubes if you like) where the carbon cleans up the NiO so the Ni can act to localize electrons providing a resistance switching structure, like a CeRAM. and the NiO glues the stucture together. Then received wisdom would indicate it would be fabricated in the conducting state.
As far as the new Nantero model** is concerned with the resistance states determined by the width of the tunnelling gap(s) and without a requirement for forming. The device would be in either one of the two or more of its possible resistance states. Essentially open circuit
As discussed in the comments here below, what holds an all tunnelling non contact structure together is still a mystery. In one of my figures in a comment here below and in the interest of simplicity I illustrated this as a undulating gap. You have to imagine a 3D version of that held together either by adsorbed gas or NiO. Otherwise without forming it seems the structure would fall apart. The need for a separate FORMING step will eliminate any of the NV memory contenders not just carbon fabric types.

**"Investigation of Carbon Nanotube Memory Cell Array Program Characteristics," by Sheyang Ning, et al International Conference on Solid State Devices and Materials (SSDM), September 2015.

resistion
User Rank
Author
Re: random CNT filaments
resistion   9/16/2016 12:34:40 AM
NO RATINGS
This is all a very interesting discussion, thanks Ron. One more thought, what would the initial state be, all CNT gaps open or some connected or all connected?

Ron Neale
User Rank
Author
Re: random CNT filaments & CeRAMs
Ron Neale   9/13/2016 6:42:43 AM
NO RATINGS
Nonvolatile. I am merely trying to connect the dots and establish the reason Fujitsu signed a Nantero carbon fabric memory licence, rather than the other options ReRAM/RRAM, including CeRAM or even MRAM. Assuming it was not just a me too gesture, i.e we must have a non-volatile memory technology like the rest of the gang, not unknown in the electronics business world.
Perhaps they get the best of both worlds nano-carbon and CeRAM. By using NiO as the glue for the Nantero carbon nano-fabric. The carbon of the nano tubes (fabric) then provides both the sigma and pi bonding to bring nickel to the ideal state for correlated electron resistance switching. In the same way as Ni(CO)4 doping does. Negating the need for a triple layer CeRAM structure to kill surface states. Also co-deposition might be easier to work with than "Liquid Death" the other name for carbonyl Ni(CO)4; although I guess by now some pre-cursors have been found to make it safe.

nonvolatile
User Rank
Author
Re: random CNT filaments & CeRAMs
nonvolatile   9/12/2016 2:39:04 PM
NO RATINGS
It seems to me that you know something about this new age of  Nantero and a possible mix of CNT and NiO co-sputtered with C, to be a matrix insulator. It is well-known that NiO/Ni can be crucibles to form carbon nanotubes, and everything mixed together could come close to a CeRAM. But, I do not see the advantage, as the phases are all backwards and the annealing/deposition temperatures are rather high for deep nano-scales. And, you still have a percolation system.....

 

nonvolatile
User Rank
Author
Re: random CNT filaments & CeRAMs
nonvolatile   9/12/2016 2:33:25 PM
NO RATINGS
By the way, the paper that you mentioned says this:

The sample marked in Tab. 3 as NiO#j3 is the NiO film doped with carbon, by co-sputtering at 500 °C. We discovered that incorporating a little amount of carbon into NiO results in lower film resistivity.

The data is presented in fig. 6. 

 

It is a tricky business to add carbon. It can go the reverse in terms of doping, as apparently happened here. I am sorry, but I can say nothing about materials processing. It is part of a Big IP package.

I was aware of this paper and we , of course , have tried many processing techniques.

nonvolatile
User Rank
Author
Re: random CNT filaments & CeRAMs
nonvolatile   9/12/2016 2:21:10 PM
NO RATINGS
Yes. It would be possible but not recommended. Scaling could become a problem. The composite of CNT+NiO would still be a random arrangement and very variable from cell to cell.

Ron Neale
User Rank
Author
Re: random CNT filaments & CeRAMs
Ron Neale   9/12/2016 2:00:47 PM
NO RATINGS
Nonvolatile:- In a paper published in Poland,** it was clearly demonstrated that the resistance of NiO could be varied by co-deposition with carbon. Do you think it might be possible by increasing the carbon and reducing the NiO beyond the levels of that particular work, so that it acts only to bond the conducting carbon tubes flakes, or ribbons to form a CeRAM structure. Where the electrons can be localised on the Ni as is claimed to be the case for a CeRAM formed by Carbonyl doped NiO. This Polish paper was in my archives because when I was doing due dilligence and exploring possible alternative explanations for CeRAM it appeared it might be possible to use electric field to move oxygen in NiO films and change its bulk resistivity.

** Electrical and optical properties of NiO films deposited by magnetron sputtering, by MAREK GUZIEWICZ1*, et al Optica Applicata, Vol. XLI, No. 2, 2011

nonvolatile
User Rank
Author
Re: random CNT filaments & CeRAMs
nonvolatile   9/12/2016 12:55:37 PM
NO RATINGS
Well, in the CNT case, there is no Mott-Hubbard. It is as you said, an internal thing inside the CNT. The DOS there is a result of the x-shaped Energy diagram, and it leads to a gap that can be current/voltage induced. So, it is an analogous mechanism but far different physically. The "gas" that you referred to is the "free electron" band in the Mott-Hubbard. Here, in the CNT, it is a different situation in which the arrangement of hexagonal cells and the tube diameter is responsible for the gap. Since diameters vary all over, there is no clear cut way of guaranteeing that some will be metal and some semiconductors.   is a   But, it is known that a single molecule in a FET can have up to 6 orders of conductivity modulation, depending on the gate voltage.This is an added complication as the voltage distribution in the percolation network is not uniform and will change from device to device. But, on the average, perhaps this will be under control. Fundamentally, the Band structure of a single wall CNT is like I said above, X-shaped, or like to Vs. Only very few electrons can pass through to the conduction band, and the gap is actually regulated by the diameter of each tube. So, there is nothing like Mott-Hubbard here.

Ron Neale
User Rank
Author
Re: random CNT filaments & CeRAMs
Ron Neale   9/12/2016 11:33:02 AM
NO RATINGS
Thanks Non-volatile. It would appear you have offered an acceptable way of accounting for the matrix material between the carbon of the nano-fabric, tubes, it does not have to be there. The memory switching is a form of gap modulation in the carbon itself. If I assume you are referring to a Mott-Hubbard type gap, it sounds very much like a CeRAM to me, if not then it's some form of contact modulation rather than the claimed tunneling. I did wonder if it was possible to build a non touching carbon nano-fabric structure bonded together (say by included gas) on which it would be possible to localise electrons and obtain a correlated CeRAM like operation. If I understand your comment you are saying all the fibres can be in contact and the modulation of the gap within the fibres creates the two resistance states in all, or enough, of the fibres to provide a bulk-like phenomena.

nonvolatile
User Rank
Author
Re: random CNT filaments
nonvolatile   9/12/2016 9:33:26 AM
NO RATINGS
Ron, It seems that the correct understanding has finally entered Nantero. What I mean is that in a simple way, every carbon nanotube has an intrinsic gap (if not in a metal phase). From the Dirac dispersion, the density of states is:

g(E) =  SUM(gi(E))  ,           gi(E) = 4 /(pi)πv  (1 − (Ei g/2E)^2 )^−1/2  

where v = fermi velocity.

By proper annealing, the right phase of the carbon nanotube would  be semiconducting. Since the conductivity will have a step like "switch" as g(E) changes (i.e. roughly sigma = e^2 . g(E) and E=Efermi, when the density of states changes with a gap being there or not, we have two stated of resistance. 

Now, that is just the characteristic of a single (one wall) carbon Nanotubes. A random cluster of CNT would contact each unit and in the gapless state, it would be a random network of resistors.This is still a percolation network which in the metal phase, there would be no problem modeling it that way. The contact potential from CNT to CNT would not be a factor. But, in the semiconducting phase - provided it is well prepared, the contact potential would come into play and we have to assume that 100% of all CNTs are in a semiconductor phase, as a sneaky metal path would kill the gapped state.

There are of course MANY issues in this. For example, is there a correlation cluster density variation from bit to bit? if so, what is the noise margin so that one could design a proper sense amp. Also, what is the temperature dependence of this phase transition. Being in a semiconductor phase, the temperature dependence would be high and more than in an insulator. What is the acceptable leakage as a bunch of contacting semiconductors have very imperfect space charge at each "lousy" junction, and on, and on.

CNT mesh memory, still seems so "weird" . Randomness is not a good semiconductor processing quality. In any case, this is a bulk phenomena in a single CNT, but a randomly distributed composite, a bit like carbon fiber composites. The belief system that this will yield I do not have, as the small scale and the contact potential distribution seem too uncontrollable. However, sometimes, annealing does wonders in creating a long term solution even in a metastable state.

In the case of CeRAM, the density of states has analogous switching, but not with Majorana fermions. Also, the full bulk is homogeneous and does not depend in Random Resistor Networks or the analogous "Filaments" of Nanoionic Devices of ReRAMs/RRAMs.

 

Page 1 / 2   >   >>
Like Us on Facebook
EE Times on Twitter
EE Times Twitter Feed