The HP paper was completely non-sense. It was the best piece of pseudo science ever done by a major corporation. Thank you for separating me from that crowd. Unfortunately, or fortunately, this is how I see the filament based and Conductive bridge devices. Uterly nonsense.
Thank you for your interest. Yes, the Data will tell, and its agreement with the model.
"Tough questions" only means they have found the right person. The two memristor Nature and Nature Nanotech papers by HP in 2008 are the typical cases. What would happen if theoretical physicists like Meuffels and Soni were listed as the reviewers? I don't know whether Chua was in the reviewer list. But surelly favorite comments have been given.
Your paper could be different from the HP's, becuase you are a serious physicists talking about the serious physics problems. But since you are talking about physics, then physicists will be the reviewers. As I have said, there are many arguments in MIT. One of the difficulties is to show the direct evidence. Good luck to your submission.
Thank you for your congratulations. The MIT in the heavy doped devices is caused by the Anderson process, which shows that the electron localization is due to the heavy defect concentration or heavy doping. Your are correct in stating that this is a fundamental process - that there are many views. What you may be missing is that I am a professional and a professor, and I have published in Nature and Science before. So this is not some fantasy. In the last 6 years I made sure I new the theoretical side of this by reading 7200 papers in the subject and carefully carved the device physics, which is the combination of the proper green's function and an open system in which electrons and holes can come in and out of the device. As far as priority, we published already the transport theory in the nanoscale Mott transition and got IV characteristics. Final refinement and generalization to other materials will follow. In the case of NiO, it is well known that it is a Charge Transfer Mott insulator. The question is, how does it switch states in an thermodynamic open system like a device? What are the controlling mechanisms? And, do you have THE DATA? All this has been carefully studied by myself and my team. This is a big departure from just Breaking down NiO to make filaments. This is fixing NiO with CO to control the oxidation numbers that induce the Charge transfer MIT. Sounds like science fiction, I know, but it is a real thing and it is repeatable. The devices work well and in predictable fashion. As far as the papers coming out of this effort, I suggest that you also look at the issued patents of Symetrix. And, as far as the understanding of the theoretical side of MIT, let us see what the reviewers will say, but I appreciate your concern about "tough questions". Thank you for your interest.
I have to say that MIT (metai-insulator-transition) is a very traditional topic, although there are many arguments, for example, the correlated electron state in a doping process. The discussion of the mechanisms in a resistive switching process involves a lot of papers, and MIT is one of them. It is interesting but also very fundamental. I heard that you are submitting your paper to a journal such as nature. They could ask you many tough questions. Anyway, congratulate to your outstanding device performance!
Correct. Both ways would be fine. But, I do believe that you must have not only tools but also be able to create experimental scenarious that you can predict the outcome. This is important in device development, especially when the materials cannot be taken for granted, like in the case of ReRAMs. In this case, you must ask many questions, starting with the mechanism that goes into the metal/insulator transition. What turns it on and off? Is it carrier dominated, density dominated, Tunneling dominated etc. In checking all these issues you can start fine tunning the device physically and your hypothesis in a step by step way. Thus, the proper tool at the proper time is key - but most important is the asking of the proper questions. Of course this is true in science overall, but when applied to a very specific set of questions under an educated question asking system based on a physical model is key. As the questions are answered and support the physics of the device model(not just math), then a "device paradigm" comes into place, that may be completely different than other people's. In the case of ReRAM, this has happened by introducing the carbonyl ion into the materials in such a way that you can make it always conducting, for example. And, then not always conducting, but compensated enough that the transition from metal to insulator and vice versa can be controlled to a 5 nm thick region in even Aluminum or CoSi2 and other electrodes.So, it is a different beast to start with. Knowing the always conducting state had no filaments imposed, we then can build up from there and understand better. When such departures from other people's experiences occurs and you can explain it, you have something. This is our case. Many device architectures can be tested and poperties predicted with high accuracy. I had many such experiences in my professional life. Thus, it may sound old fashion but I do not see how anyone would just "go to the kitchen and cook a device" without physics. And, now I add, without chemistry and Materials Science. In closing, this is very complex device -filaments or not. In the filaments area, the insulating state is the begining. In CeRAM case, the conductive state, usually hole current is where we start. Very different starting points, but it is a cloo for the device understanding to do both types of devices (RERAM and CeRAM).
I am sorry but this is too speculative. It is also not confirmed by many forms of experimental results. I suggest that you read the literature on Metal/Insulator phase transitions. Explanations with "filaments" can be easily debunked.
Leakage as an evidence of extremely fine filaments, etc. is just not what is happenning physically. It can happen in the normal NiO without doping control. But, hundreds of experiment with model verification contradicts that possibility. The switching mechanism in CeRAM is due to an electron-electron interaction at the orbital level. Verification of this is solid. The same can be acting in the filamentary devices at the points of conect/disconnect areas. The filaments are continuous Ni(0) metallic phases surrounded by lots of defects with activation energy of 0.48 eV. This makes the off state very unstable with temperature. This is not the case in CeRAM. The quantum phase transition is temperature independent and has been tested from 4 degrees Kelvin to 150 degrees C. Since the interaction energy between electrons is short range coulombic, the cleaner material without too many defects is able to have in loco switching. The filamentary material, even with filaments, still requires such an interaction to exist in order to shut off conduction. It is only a matter of materials design to make the interaction the clear mechanism - such a design can only be achieved by proper charge compensation of Ni, to force it into a reversible disproportionation reaction. Without that precise doping, the switching still occurs in filamentary devices, but the control of the Vset and Vreset is really difficult and random. Thus, you see a lot of ReRAM proposals with a layer that is somewhat stoichiometric and a layer non-stoichiometric or of a different material so that charge trapping in the interface become dominant and a new thing is now controlling the storage states, but it is still very defficult to make a defective material behave the same way everytime and everywhere in the array. This has been done for 12 years without much success for product stability. Our way is better - not only it works better, it has well explained physics to propose moedl checking experiments. Thanks for your interest.
Of course HRTEM is best for finding/confirming filaments. But the matrix around the filaments may also conduct with area proportionality. If it's negligible leakage level, certainly it can be said to be "filamentary."
Maybe should clarify, check resistance vs. area. A filamentary case is virtually area-independent. The resistance would depend on the writing current. Alternatively, the writing current would depend on area.
As long as the proportionality with area is violated, it is filamentary, i.e. localized at some points. Almost certainly at multiple locations. The single filament is an idealization, the current density would be unphysical.