„OxRAM can be an alternative to eFlash in some products“

Any flavor of RRAM + any flavor of selector is stopped in some companies due to trade-off between power and stability

DESIGN&ELEKTRONIK: MRAM currently is the most promising kind of memory as a replacement for SRAM cache and even DRAM. However, MRAM likely will need new transistor structures below 5 nm. Which ones?

Jo De Boeck: For Embedded at all nodes the density of embedded MRAM is defined by front end to deliver enough current for MRAM switch. Specifications gets tighter but the success of embedded MRAM depends on the capability to use the logic as such - whatever is the solution at N5 and below - without special transistor dedicated to MTJ. That being said, even denser MRAM could be obtained with the introduction of a selector – the diode type-  to be able to have front-end transistors delivering current for more than 1 MTJ. Selection is coming from the diode.

For Standalone the transistor is too big for standalone DRAM-type application, three main breakthroughs need to happen: a 2 terminal selector that can be scaled below 40 nm pitch, MTJ patterning itself needs to go down below 40 nm pitch, and MTJ memory window needs to improve.

DESIGN&ELEKTRONIK: Besides MRAM  the list of other interesting options includes a spin-orbit torque MRAM and ferroelectric RAMs as DRAM replacements. Furthermore the industry is working on at least five candidates for storage-class memories, mostly variants of crossbar and resistive RAM structures. Which ones could survice the research phase and be suitable for mass production?

Jo De Boeck:  SOT MRAM is a viable candidate for SRAM application, it is at research level now and 5-10 years ahead. Ferroelectric 1T1C is a candidate for high end SCM - close to DRAM - it has an unknown status in industry, possibly it’s in development 2-4 years ahead. Ferroelectric 1T, in NAND or NOR structure for SCM is on research level. PCM + OTS is in production as so called Xpoint memory. Any flavor of RRAM + any flavor of selector is stopped in some companies due to trade-off between power and stability. Some other companies still try to develop it, the goal is to reach market 2-3 years from now. Some start-up back to drawing board and alternative solutions at material level (Weebit, 4DS, ...). There is no clear winner yet.

DESIGN&ELEKTRONIK: Imec is doing work on a version of OxRAM. Which are the target businesses and which are the biggest advantages regarding these applications?

Jo De Boeck: OxRAM in imec is mostly targeting embedded Flash replacement in logic nodes bigger than 28 nm. It is cheaper than eFlash, and can be easier introduced than MRAM. It can be an alternative to eFlash in some products.

DESIGN&ELEKTRONIK: Dr. De Boeck, thank you very much for your time and your interesting answers!

Dr. Jo De Boeck

received his engineering degree in 1986 and his PhD degree in 1991 from the University of Leuven.  Since 1991 he is a staff member of imec (Leuven). He has been a NATO Science Fellow at Bellcore (USA, 1991-92) and AST-fellow in the Joint Research Center for Atom Technology (Japan, 1998). In his research career, he has been leading activities on integration of novel materials at device level and new functionalities at systems level.  In 2003 he became Associate Vice President at imec for the Microsystems division and in 2005 started Holst Centre (Eindhoven) and became CEO of imec-Netherlands.  As SVP he headed imec’s Smart Systems and Energy Technology Business Unit. He is part-time professor at the KU Leuven and held a visiting professorship at the TU Delft.  In 2011 he was appointed CTO of imec corporate and is member of imec’s Executive Board.

Matchmaker+ Anbieter zum Thema

zu Matchmaker+

1. „OxRAM can be an alternative to eFlash in some products“
2. „EUV can certainly make the design rules simpler“
3. Any flavor of RRAM + any flavor of selector is stopped in some companies due to trade-off between power and stability
4. Disclosure