FeFETs: How this new memory stacks up against existing non-volatile memory

A link to the paper mentioned in the article (paywall&#x2F;institution access): <a href="http:&#x2F;&#x2F;ieeexplore.ieee.org&#x2F;document&#x2F;7838397&#x2F;" rel="nofollow">http:&#x2F;&#x2F;ieeexplore.ieee.org&#x2F;document&#x2F;7838397&#x2F;</a><p>TL;DR - Much lower erase time than NOR Flash (3 orders of magnitude), 1&#x2F;3 voltage required for write&#x2F;erase, all other parameters typically the same<p>The results from the paper are summarized as follows:<p>- Cell size is comparable to current NOR Flash (0.045 um^2)<p>- Endurance is also comparable (~10^5 cycles)<p>- Estimated 10 year retention @105C, compared to 150C for NOR<p>- Read time &lt;25 ns, versus &lt;10 ns<p>- Write time around 500 ns (1&#x2F;2 of NOR)<p>- Erase time around 500 ns, compared with 10^6 ns for NOR (!)<p>- Energy consumption not yet measured, but claimed less than Flash<p>- Max voltage of 4.5V, versus 12V for Flash

&quot;First and foremost, FMC is aiming at the embedded NVM space.&quot;<p>It also seems that they compare it with NOR and not with NAND, which again points to the scenarios where not much is stored and not often rewritten. Then <i>it&#x27;s not a technology for SSD&#x27;s or USB sticks, how I understand it</i>, but it&#x27;s to allow to have a single chip which has both the logic and the re-writable firmware. And they claim it&#x27;s cheaper to achieve that than with the competing technologies.

This stuff is fascinating, especially the challenges in fabrication as the theoretical limits are approached. I do have a question, if an expert could help explain. The article says,<p>&quot;With FMC’s proprietary hafnium oxide, the standard gate dielectric can be made ferroelectric—even for film thicknesses that compare to the one used in standard logic transistors. This proprietary hafnium oxide integrates extremely well with all current and future processes utilizing HKMG. Therefore, a scalable ferroelectric FET finally becomes possible.&quot;<p>What is proprietary about the hafnium oxide? Is it the process of creating it, or the molecule itself?

Sounds like an improvement over FRAM, which is already a shipping product that you can buy today - albeit more expensive than Flash and not as dense.<p>The ability to build with standard HKMG (high-k metal gate) wafer processing (unlike Flash) is very handy - means FeFET memory can be incorporated into high-speed microcontrollers.