A new competitor is on the scene...ECRAM... at least many many years away though....
here's someone elses commentary...
""
Posted by Ron Neale
Add to My Watchlist
What is My Watchlist?
 (20min delay)
|
|||||
|
Last
$1.65 |
Change
0.155(10.4%) |
Mkt cap ! $343.3M | |||
| Open | High | Low | Value | Volume |
| $1.51 | $1.66 | $1.51 | $1.822M | 1.137M |
Buyers (Bids)
| No. | Vol. | Price($) |
|---|---|---|
| 1 | 49197 | $1.65 |
Sellers (Offers)
| Price($) | Vol. | No. |
|---|---|---|
| $1.66 | 4227 | 1 |
View Market Depth
| Last trade - 16.10pm 24/06/2025 (20 minute delay) ? |
| WBT (ASX) Chart |
A recently published paper has provided important new information the operation of electrochemical NV memory.
“Unveiling ECRAM switching mechanisms using variable temperature Hall measurements for accelerated AI computation”
The newsworthy essentials at EENewsEurope
https://lnkd.in/ee98Egax
In brief the ECRAM is configured FET-like as shown in the topmost box in my sketch below. The extended part of the WO3 channel is related to some important Hall effect measurements
The competitive marketing position for ECRAM over ReRAM is the claim that it is based on bulk switching, low power, three terminal and its operation is topocratic. Meaning there are no chemical, spacial or device structure changes during a complete switching cycle.
ECRAM is therefore positioned as superior to ReRAM where filaments and stochastic effects prevent topocratic perfection.
For the ECRAM experimental evidence is presented to establish that the change in electrical conductivity of the channel is related to oxygen vacancies moving in and out of the channel from the HfO, a redox action in the channel. Plus the benefits of an electronic double layer(EDL)
The role of the upper electrode as a reservoir is not clear, it is just possible it is used as part of an initial “Forming” process to control the density of vacancies in the in the as deposited amorphous HfO in order to establish device-to-device consistency.
There is some asymmetry in the ECRAM transfer curve for negative going pulses. It would appear to compensate for that asymmetry the amplitude of pulses used are reduced, from +1.5V for potentiation to -0.8V depression.
The superior competitive position claims for ECRAMs would seem to ignore that with scaling the behaviour of a ReRAM is topocratic-like, because there are only a limited number of places for ions and vacancies to find themselves. Which means with scaling a single ReRAM nano filament will be a topocratic-like low power device. In my sketch below I, top row, I have illustrated how such a device might operate.
There are even options for oxide ReRAMs to create 3 terminal devices, similar in structure to the ECRAM. The first of which is illustrated in the second row of figures in my sketch.
The third row in my sketch is a conventional ReRAM where the value of SET current has a linear relationship with conductance (g = Iset/k). For that reason ReRAMs might prove a better option than ECRAM for in-memory processing AI accelerators and some neuromorphic applications.
None of the points I have raised for discussion should detract from the brilliant piece of analytical work that the cited ECRAM paper represents.
I have tried to contact cited authors for some clarification of details. so far without success.