New memory tech seems to be all that and then some and capable of replacing all dram and nand flash memory chips in future computers. Can some tech guys here opine on this? 'Holy grail for memory tech': New candidate for universal memory emerges in race to replace RAM and NAND — and this one doesn't use a toxic compound (msn.com) ULTRARAM, developed by Quinas Technology, a spinoff from Lancaster University in the UK, combines the advantages of both, offering fast, non-volatile memory with high endurance and ultra-low switching energies. The technology, which recently won an award at the Flash Memory Summit, boasts superior longevity than flash storage, matches the read/write speeds of system memory, and requires less power. ULTRARAM leverages a quantum-mechanical process called resonant tunneling, enabling it to deliver non-volatility with fast, energy-efficient write and erase capabilities, leading to high endurance. This combination of properties was previously considered unattainable, and why some have referred to it as the “holy grail for memory tech”. ................................................ Unlike flash memory, which uses a highly resistive oxide barrier to retain charge, ULTRARAM uses atomically thin layers of InAs/AlSb to create a "triple-barrier resonant-tunneling" (TBRT) charge-confining structure. This allows ULTRARAM to switch between a highly resistive state and a highly conductive state, giving it its unique properties. ULTRARAM's energy efficiency is certainly impressive. It has a reported switching energy per unit area that is 100 times lower than DRAM, 1,000 times lower than flash, and over 10,000 times lower than other emerging memories. Its ultra-low energy credentials are further enhanced by its non-destructive read and non-volatility, which removes the need for refresh. ................................... And one more benefit? The inventors say it can be mass-produced using existing manufacturing processes in the semiconductor and silicon industries.
Quinas ULTRARAM™ is a remarkable, patented, memory technology developed by Lancaster University. It exploits a quantum-mechanical process called resonant tunnelling, allowing ULTRARAM™ to deliver non-volatility with fast and energy-efficient write and erase, resulting in high endurance. This combination of properties was thought to be unachievable until now. ULTRARAM™ has the non-volatility of flash, with a performance that is expected to exceed that of DRAM. ULTRARAM™ is a flash-like floating-gate memory. However, unlike flash, which uses a highly-resistive oxide barrier to retain charge in the floating gate, ULTRARAM™ uses atomically-thin layers of InAs/AlSb to create a triple-barrier resonant-tunnelling (TBRT) charge-confining structure. The TBRT switches between a highly-resistive (locked) state (with no bias, ‘store’), to a highly-conductive (unlocked) state on application of just 2.5 V across the gate stack (program/erase). It is this mechanism that gives ULTRARAM™ its remarkable properties. Unlike DRAM and flash, which are silicon based, ULTRARAM™ uses III-V compound semiconductors. Specifically, the so-called 6.1-angstrom family of semiconductors (GaSb, InAs and AlSb). This allows engineering of the memories’ electrical properties to exploit the underlying physics to full effect, whilst also being capable of volume manufacture using established processes in the compound semiconductor and silicon industries. The extremely low electron-effective-mass in InAs also opens the possibility for a new high-speed embedded III-V logic to address arrays.
My dummy version Different material in memory chip significantly reduces resistance = reduced heat= reduced electricity. Has proven reliability in testing for cycles and accuracy. Is patented. Tech can be applied to existing chip making plants without major changes to production lines. Seems likely, imho, to dominate the world market going forward. Biggest advancement in flash memory in decades. I think. But im a dirt guy, so thats probably all wrong Need tech support in here
1 - massively expensive to produce relative to Si based technology. 2 - largest compound semiconductor fabs in the world are in Taiwan 3 - Ga, In, As and Sb are toxic. While commonly used in optoelectronic chip fabrication, you cannot just run out and fabricate with those just anywhere (for example, environmental laws prevent any new compound semiconductor fabs in many states). 4. - There are only 6-7 compound semiconductor fabs in the United States, half of which exclusively run GaN and SiC for the high power/high frequency industry. 5. - You cannot, contrary to what was implied in this story, easily and quickly convert a Si fab for compound semiconductor fabrication. The smallest Si fabs use tools that handle 8" wafers, while the absolute largest Compound Semiconductor wafers are 6", while most are 4" wafers. 6. - There is no reliability data on these devices, it will take more than a decade to study these and determine if they actually will be commercially viable. I was told I was going to lose my job 20 years ago as carbon non-tubes were going to replace circuit boards. Quantum computers were going to replace Si based processors - and haven't. It is unlikely that this is going to replace conventional RAM any time soon.
A bit off topic but not completely unrelated: An announcement was made (today) that Northern NY State area would be landing the award for a 1.5 Billion chip plant. The actual community was not named.
Wolfspeed (formerly Cree's Semiconductor Division) It will be on their campus in Malta, NY It is being paid for by CHIPS ACT money IF....big if, it gets funded at all. The fab will be dedicated to SiC and/or GaN for high power applications. If....Wolfspeed does not collapse under the weight of their own debt.
I am hitting all of the Nerd stereotypes here, but I used to build them for people when I was in college.
InAs/AlSb quantum tunneling has been a research topic for 30+ years in the Universities. As a EE, I could go into a huge discussion on the pros and cons, but right now there is no commercial application or use case for it at the moment. Flash tech is stable and cheap. Ram is relatively fast and cheap and the next two generations of DDR RAM is already spec'd. The research is mainly focused on next generation quantum computing, which right now and the foreseeable future will remain in the cloud and mainframe space. In other words, not something new to the 'nerds'
OT: AI development can be throttled by slow playing hardware advances and strategic control of certain technologies and manufacturing. Memory is one of those tollbooths of technology that has to exist until humans have a plan / strategy for general AI.
Another CHIPS investment into Saudi Arabian owned chip foundry Global Foundries for expansion in Malta, NY and Burlington, VT. US to Award $1.5 Billion to GlobalFoundries in Chips Act Grant
not sure but it has taken a beating the last couple of days. took some gains from nvidia today and rolled them into Palo Alto networks which got crushed today when they announced plans to provide some services for free to capture market share. been watching PANW waiting on an entry and it provided itself today, not sure how long it is going to take to recover but nvidia had become too large of a portion of my portfolio.
