18-11-2020 | | By Robin Mitchell
Recently, SK Hynix and Bosch have helped memory start-up company Ferroelectric Memory Company to fund their next development stage of reducing FeFET to 7nm. What is FeFET, how does FeRAM work, and how does FMCs hope to capitalise on its intellectual property?
Ferroelectric Field-Effect Transistors, or FeFET for short, is a transistor technology which places a ferroelectric material between the gate and the source-drain conduction zone. When a voltage is applied across the gate-source region that is greater than the coercive voltage of the ferroelectric materials, then the ferroelectric material is polarised. Once polarised, the transistor retains its state even when power is removed, and it is this effect that could power the next generation of non-volatile memory.
FRAM is similar to FeFET, but instead of using and FeFET, the capacitor in DRAM is replaced with a ferroelectric material. When the memory is written to, the ferroelectric material is polarised to a specific direction depending on if a 1 or 0 is written. When reading from the bit, the bit is forced into a 0 value, and the current consumption is measured. If the bit was storing a 0 at the time of reading, then the current consumption is minimal. If the bit was storing a 1, then the resulting change in magnetic polarity causes a spike in current consumption. While this means that FRAM requires to be re-written after each reading, the information is stored even when power is disconnected, thus having non-volatile properties.
FeRAM is similar to FRAM, but the process in which memory is read is non-destructive, and the production process can be made much smaller. Instead of using a ferroelectric material instead of a capacitor, the memory unit itself is a ferroelectric field-effect transistor. When reading from such a memory unit, the voltage used is less than the coercive voltage, thus leaving the magnetic polarity intact. While FeFET memory is still in its early days, it can be reduced to much smaller sizes than FRAM.
FMC raises over $20 million in Series B funding
SK Hynix, a major developer of memory technologies, have helped a start-up company called Ferroelectric Memory Company, to secure funding to further help with its development of FeFET memory. Also involved with the Series B investment was Bosch, who is also another major component manufacturer hoping to integrate FeFET technology into the next-generation devices. The total funds raised was over $20 million, and this is significantly better than their first round of investments back in 2018, which saw the start-up company secure $5.4 million.
The money secured by FMC will be used to continue their development with FeFET technology, and to shrink FeFET devices to 7nm and beyond. By shrinking these devices to such small sizes, not only can memory densities be increased, but the memory capacity can match other memory technologies that already work at 7nm. Three key areas of application that FMC is looking to integrate FeFET technology into include embedded non-volatile memory, storage class memory, and memory compute applications. If the technology can break the 7nm barrier, then FeFET would be a memory technology superior to others such as FLASH and MRAM due to their bulky size and expensive nature. FeFET also provides designs with temperature stability, magnetic immunity, and high radiation resistance.
However, instead of manufacturing the devices themselves, FMC is hoping to instead go fabless (like ARM and Xilinx), and license out its designs to device manufacturers. The use of a tiered intellectual property approach allows FMC to sell IP on the process itself, the device, and an overall design. According to FMC, their roadmap has FeFET memory becoming available on 28nm planer CMOS by 2023, and from there the FeFET memory technology will move into IoT, automotive, and industrial applications.