Unfortunately this is not true. Compliance simply limits the amount of current and has nothing to do with whether a filament is present or not. If the bulk switches to a conducting state, the current will flow like in a metal. It will increase with the ability of the metal to pass more current. But, once the voltage/current goes to zero, there is no more current. Now, as you re-start from zero upt to the turn-off voltage, the mechanism to turn off the current means that something inside the material is able to stop flow of carriers. That something when there is a filament, would be some kind of filament break or disconnect reaction. What kind of reaction? in these transition metal oxides, the first thing that comes to mind is a change in oxidation number that would create some distortion in the local fields. and, in many cases, this is what happens to filaments near the anode. The anode can source electrons or holes, but mostly there will be a deficit of electrons a few lattice constants under the anode, just enough to break a filament. In our case, this random breaking of filaments by an electrochemical reaction is taken out of the equation by providing a highly conductive doped layr of NiO. Then, in a very thin middle layer, only about20 atoms thick, the active switching region is built. Enough analysis with XPS etc. has shown that this switched region is uniform and has different dominance of oxidation numbers of the right kind (+1,+3,+2) and none of the filament kind in the conductive state or otherwise(+0, metallic NiO). So, not only there is physical evidence from the spectroscopic data, but also from the electrical data, as hole injection from the anode controls the shut-off, Now, the step of making it conductive, with the compliance, is proportional to this hole current that appears going from zero volts to Vreset (turn off voltage). Experiments can be designed to literaly control how many electrons you want to come in at the seting voltage, and how many holes to come in to turn it off. That is, you must "erase" the amount of excess electrons that created the conductive state with incoming holes. This can be very precise and vary with doping levels to the point that the maximum "ON" current in the ohmic side of the IV curve, at a maximum, will be exactly the value of the compliance current. None of this could happen with any kind of filament through the perfectly metallic constituint Ni(0+). And, if the buffer layer, which is conductive, was made up of filaments, it would extinguish holes right away and never get to the thin middle layer of 20 atoms thick. And, if the thin layer would be of filaments, it would never be under the metal/insulator interface near the anode to have a disconnect reaction. This and many other considerations of what kind of potential energy landscape the Ni ion sees with or without doping, make it almost indisputable the filament argument in CeRAM.
Resistion: I think the validity of your point, filament versus bulk switching, would rest with you providing our readers with an explanation of the mechanism that allows the formation of an insulating filament in a material that it is claimed is highly conducting in its as-born state. If you consider a planar parallel electrode structure once your insulating filament is formed why would the material surrounding your filament that is still in its conducting state and subjected to the same applied voltage not switch to the insulating state? The claimed lack of a need for forming is a key point.
thank you Carlos, will do...if anyone else is interested pls drop me an email, we will be orgazining in Vancouver in 2015 with Santosh Kurinec a session on emerging memories...BTW, our book on this topic is out: http://www.amazon.ca/Nanoscale-Semiconductor-Memories-Technology-Applications/dp/1466560606...conference link is www.cmosetr.com, Kris
Congrats Carlos, you did it again...would you be interested in presenting this at emerging technologies symposium in Vancouver in 2015? preliminary program at www.cmosetr.com, cheers, Kris (firstname.lastname@example.org)
A Book For All Reasons Bernard Cole1 Comment Robert Oshana's recent book "Software Engineering for Embedded Systems (Newnes/Elsevier)," written and edited with Mark Kraeling, is a 'book for all reasons.' At almost 1,200 pages, it ...