Ron Neale takes a closer look at 4DS Inc.'s reported progress in its development of metal oxide hetero-junction operation (MOHJO) non-volatile memory.
I have a number of problems with the experimental protocol that might have affected the result reported by . The intimate contact between the oxide and the PCMO required to form the device and the resulting oxide-PCMO bond must be fractured after each event. The stress of fracturing plus any oxide remnants (and the surface-states caused by that action as well as the surface states that already exist) might be important in relation to any conclusions vis-a-vis the situation in a complete device structure. Because switching is occurring, the surface of the tungsten tip would be oxidized at each switching event. Unless it is cleaned and re-polished after each event, cumulative changes in the oxide are likely to occur. Further cause for concern is the probe tip. The switching experiment is carried out in air with possible oxidation effects from oxygen or water vapor. While acknowledged in , its possible consequences are ignored. Perhaps the most important potential problem on my list is the use of tungsten, a hard material, in contact with PCMO, a relatively soft material, where the pressure will almost certainly result in some surface deformation, structural change, and even the creation of oxygen vacancies as a result of significant differences in hardness.
Another problem this writer has with the tungsten probe experiment is for one aspect of the measurement the material is described as being as being in its “as-fabricated LRS state.” However, the LRS state of interest only exists (and the Shottky-like barrier responsible for it) when a metal is in contact with the PCMO and an oxide layer has been formed and subjected to at least one switching cycle. Unless in an operating device, the surface of the PCMO material in its LRS state is always returned to exactly the same state as it was in the as-fabricated state, all that can really be said about its oxygen vacancy distribution is that it is unknown. There is confirmation that surface changes do occur where the switching process is described as “approximately reversible” in that when switching back to the LRS from the HRS the PCMO surface does not return to the same state . This is attributed to a degradation process.
While  states that Pr0.7Ca0.3Mn O3 does not require a forming process, it appears to be contradicted by statements and observations within the paper and diagrams used more recently. This can only be interpreted as the electrical operating conditions for the first switching event when the oxide first forms on the top electrode surface are the same as subsequent operations.
Although  makes a good case based on the results of the ultra violet photo-electron spectroscopy and NEXAFS data and other analytical techniques, reservations are expressed. In the discussion they state, “As suggested by the analysis of our data, the transition may be due to the electric field-induced doping of a Mott insulator.” The authors also state in referring to the inhomogeneous distribution of oxygen vacancies in the HRS, “This may enable the system to locally achieve the level of concentration of oxygen vacancies (i.e. 5%) that are required to reach the Mott state.”
Consideration must be given to the possibility that because of the very localized nature of the Mott-like transition to obtain the insulating state the return to the metal state may involve a different mechanism, something akin to burn-out or fusing.
In  for the CeRAM the correlated electron transition was accounted for not by the movement of vacancies but by the controlled localization and de-localization of electrons based on control of the screening effect of electrons. While 4DS Inc may have what can be claimed as a reversible metal-to-insulator transition that may be suitable for memory device applications, it may actually be a Mott, or Mott-Hubbard transition, or something in between. Or, more simply, is there more than one way of achieving a Mott-like or Mott-Hubbard insulator-to-conductor transition or something in between?
Or is it a marketing promotional ploy by 4DS Inc. in an attempt to cover all bases in the light of claims by others that they have developed a correlated electron NV memory that is receiving serious attention? While preparing this report, I made every attempt to obtain information by way of clarification from 4DS Inc. on a number of the points raised in this article, as well as any new information they might have to support their Mott-like transition claims. They did not respond to inquiries. In particular, I would like to receive from 4DS a plot of log current as a function of log voltage for the two states of the device. This might help remove the possibility that thermionic emission, surface states, and trapped carriers might have more to do with the memory effect than Mott-like insulator-to-metal transitions.
 S. Asanuma, H. Akoh, H. Yamada, and A. Sawa. “Relationship between resistive switching characteristics and band diagrams of Ti/Pr(1-x) Cax MnO3 junctions,” Phys. Rev. B 80, 235113. Published 8 December 2009.
 Hong Sub Lee, et al. “A New Route to the Mott-Hubbard metal-insulator transition: strong correlation effects in Pr0.7Ca0.3Mn O3,” Scientific Reports. 23April 2013.