The group at RWTH Aachen University and Research Centre Julich GmbH has recently published a paper in Nature Communications and the first author was kind enough to send me a copy. Actually it is an open paper and can be found at: www.nature.com/ncomms/journal/v4/n4/full/ncomms2784.html. Just make sure you also download the supplemental information as well! www.nature.com/ncomms/journal/v4/n4/extref/ncomms2784-s1.pdf. The paper has received a lot of attention on the web and I wasn’t sure I could add much more but I am going to try anyway….
The manuscript is extraordinarily thorough and essentially points out that an ReRAM cell will have an open circuit voltage particularly when in the off (high resistance) state. Or more correctly, the cell will act as a small battery with a measurable capacity and will discharge itself through the external circuit. The effect is more pronounced in cation migration ReRAM cells (often referred to as ECM, electrochemical metallization, where a positive metal ion migrates in the cell to form a filament). However the effect is also apparent in the VCM, valence change, cells where anion (for example, oxygen ion) migration is the predominant filament formation mechanism. In the off (high resistance) state, open circuit voltages of ~100mV (ECM) are reported with formed VCM cells being ~100x lower. In the on (low resistance) state, the filament shorts out the battery and the battery effect is absence (ECM) or much reduced (VCM). The capacity of the nanobattery is measured to be a few nC for the (100um and greater) sized devices the authors report. These measurements are non-trivial and require some very careful experiments. The authors report they performed over 900 experiments on 8 different cell configurations! (The attached image illustrates one of the ECM cells studied.)
The authors identify three sources of the inbuilt potential. 1) Nernst potential (due to the different electrode potentials), 2) Diffusion potential (due to ionic gradients across the cell) and 3) the Gibbs-Thomson effect (due to the geometry of the partial filament once the cell is formed). As an aside, I wondered if the same effect could be observed in a PCM (phase change memory) cell. I think one could argue that the Diffusion potential should be negligible (no ions) but the other two might be present.
It is important to note that the discharge of the nanobattery does not impact the off state resistance to any great degree. However, it could potentially impact the read process of the cell in the off state as the potential across the cell could confuse the read out. In fact the authors point out that the charge associated with a readout could be comparable to the nanobattery capacity. But in practice this can be compensated for in a read out circuit in various ways provided the potential is reproducible from cell to cell. Also the authors point out that the voltage could allow for novel read out approaches for complementary resistive cells.
Christie Marrian, ReRAM-Forum Moderator