Interesting pick-up, mate, have been looking into them and their product when I've had a spare moment.
Nantero seems to have been around since 2001.
Looks like it works with carbon nantotubes (CNT) passing energy through to switch it on and off at junctions between the nanotubes.
In summary, an impressive array of claims, of which are superior to some of ours, though little to show in terms of mass production in the 20 years of the company's existence.
Here are a few notes:
Looking at their company state:
"According to Nantero, the company that invented NRAM, seven fabrication plants in various parts of the world experimented with the new memory last year. And other as-yet unannounced chipmakers are already ramping up production behind the scenes."
So they were meant to be producing this with Fujitsu in 2018:
Deep inside Nantero's non-volatile carbon nanotube RAM tech • The RegisterThen 2019:
Fujitsu Targets 2019 for NRAM Mass Production | Tom's Hardware (tomshardware.com)Then 2020:
First carbon nanotube NRAM products due in 2020, says Nantero ... (eenewseurope.com)Reasoning from
Leading a Carbon Nanotube Innovator Out of the Valley of Death - EE Times Asia (eetasia.com)"Five years of uninterrupted development within a high–quality fab has given us extremely important data. Fujitsu has advertised that our technology is aimed to merge within their ferroelectric business line.This project was supposed to have been delivered in 2019. This was a problem with the licensing business previously held by Nantero, as they had financial difficulty in meeting the R&D spend necessary to hit the milestones as defined by Fujitsu. When I came in as CEO, we started funding the development again and are back on target with an extended timeline.
The U.S. government has employed the use of NRAM for several applications, especially in space. Lockheed Martin purchased the rights from Nantero to pursue government business a long time ago, and so I am not certain of all the applications, but I do know from my work with Honeycomb that NRAM has had success within these small–batch applications.
Nantero has had working chips for decades. Filling small batch requests from Lockheed or anyone else hasn’t been a problem; we can easily find working parts on a wafer and fill orders for niche applications. The real challenge for a commercial memory company is getting six-sigma yield. If you must throw away 20% of your chips (not including edge chips), you aren’t going to be competitive.
The new company focus is building what we know we can do. We know exactly the heartbeat of NRAM, what it’s capable of, what we need to do with a controller outside the memory cell. We’re going to push that into the marketplace."
"Nantero isn’t profitable. It’s
made about $130 million during its 20–year existence. We’re an R&D company in transition to a new business plan. We spend a lot of money in those valleys between when you make money and when you come out with your next chip."
Fujitsu now making DRAM killer with 1,000x performance boost | ComputerworldEnduranceNRAM:"the program/erase (P/E) cycles it can endure are practically infinite, according to Schmergel.NRAM has been tested by Nantero to withstand 1012 P/E cycles and 1015 read cycles"
Weebit ReRAM:"between 100,000 and a million write cycles"
TemperatureNRAM"Another advantage is that NRAM is resistant to extreme heat. It can withstand up to 300 degrees Celsius. Nantero claims its memory can last thousands of years at 85 degrees Celsius and has been tested at 300 degrees Celsius for 10 years. Not one bit of data was lost, the company claims."
Weebit ReRAM:"The retention of most NVM devices is measured at ~10 years @ 850 Celsius or 1050 Celsius, but they normally can’t achieve such retention at higher temperatures. Weebit ReRAM can maintain its reliability at high temperatures, retaining data for up to ten years at 1500 Celsius or even 20 years at 1250 Celsius."
CostNRAM:"if its costs can be brought to something similar to DRAM's costs, then it is positioned to replace DRAM. Cost is the big issue here though, since it takes very high unit volumes for prices to get close to those of DRAM."
"Nantero hasn’t focused on embedded applications yet, but we certainly should. When we do, we will be able to compete very well on cost with MRAM"
Weebit ReRAM
Market growth projectionNRAMGlobal NanoRAM Market Trajectory & Analytics Report 2022 Featuring Nantero, Inc. - The NRAM Technology Patent Holder - ResearchAndMarkets.com | Business Wire2021: $22.1M
2026: $71.9M
ReRAM2021:~$16.2M (30%)
2027: ~$957M (33%)
Material accessibility:
Radiation toleration
NRAM"Our NRAM is immune to these issues, and when we were launched on a mission aboard space shuttle Atlantis in 2009, we performed for the duration with zero soft errors. We’re immune to disruptions by magnetic fields, and perhaps very relevant to today, we’re completely radiation hardened."
Weebit ReRAM"Weebit ReRAM cells can withstand up to 350x more radiation than flash."
MaterialNRAM"We have our chemical facility in Woburn, Massachusetts, where we can get from raw material to fully completed CNT spin coat layers on a 300–mm wafer."
Weebit ReRAMSeemingly majority silicon, something all fabs work with (afaik)
Some commentary I read from someone who claimed to have worked on Intel Optane, which also involves that on ReRAM:
"Pretty much all (but not necessarily all forever) of these potential candidates now suffer the same problem in one form or another: they are difficult to turn on their side and build them top down on sequentially stacked layers via high aspect ratio etching. This scaling innovation is why 3D VNAND with CXL ended PCRAM. The ovonic threshold switch and phase change layers require a top down deposition, so they can't be wrapped around a vertical bitline like charge trap nitrides can. PCRAM's core technology is viable, but it just doesn't scale as easily. Optane was a cool product at slightly the wrong time, and if someone ever works the device into 3D, something like it might reappear.
For those not in the industry, the 'speed up' from 3D is that you stack up a bunch of process layers without complex steps in between, then perform a short sequence of patterning, etching, and deposition steps that builds the device through all the layers at once. Other techniques need to deposit, pattern, etch, deposit, pattern, etch, over and over again. They don't have the same economy of scale. And you need economy, because margins in the memory business are very slim. Memory is cheap."
(20min delay)