Boffins prep tasty data-cramming 3D metal-sandwich chip Boffins have constructed a microprocessor architecture capable of packing in up to 1,000 times more data than today’s generation of processors. Crucially, the data in the chips is recorded and stored using the spin of electrons. Physicists at the University of Cambridge have built a 3D microchip that crams data into a three- …
Data in chips is recorded and stored as the spin of electrons
Not usually it isn't, that's actually the bigger news all by itself.
From the headline I was expecting a story about 3-d layered conventional chips, but then out of the blue you just casually drop in that it's spintronics as well? Did you not realise that the spintronics aspect of this is by far the more new and innovative cutting-edge technology? Conventional chips are being developed in 3-d, but that's just routine engineering r'n'd; spintronics is the real innovation here.
Also, the existing chips you're talking about are MRAM, not MDRAM, and they aren't really very spintronic. Magnetism is based on electron spin, and they store bits in magnetically charged cells, but hell, so did cassette tapes and they're not spintronic; it becomes spintronics when you're actually moving electron spins around the place as your primary 0/1 signalling mechanism. So, yes, there are magnetic cells in which the spin of bound electrons is polarised one way or the other, but they're immediately read out into simple spin-irrelevant conventional circuits to go from the cells round the rest of the chip. You write as if spintronics was an established technology and 3-d chips were something radically new, when it's actually the combination of the two that is the story here.
So is this article talking about processors or memory chips or both? It seems confused.
It also seems to say moving electrons 'vertically' is difficult and costly. Is the implication here that this new process solves those issues, or was that an observation about some remaining known issues here?
I get the feeling Gavin didn't fully understand whatever source material this article was based upon. Either that or I'm having a slow brain day (i.e. a normal day).
Slashdot just posted a somewhat clearer summary of what's going on here:
A major obstruction to the development of practical 3D microchips is moving data and logic signals from one layer of circuitry to another. This can be done with conventional circuitry, but is quite cumbersome and generates a good deal of heat inside the 3D circuit.
Physicists at the University of Cambridge have now developed a spintronic shift register that allows information to be passed between different layers of a 3D microchip.
'To create the microchip, the researchers used an experimental technique called ‘sputtering’. They effectively made a club-sandwich on a silicon chip of cobalt, platinum and ruthenium atoms (abstract). The cobalt and platinum atoms store the digital information in a similar way to how a hard disk drive stores data. The ruthenium atoms act as messengers, communicating that information between neighbouring layers of cobalt and platinum. Each of the layers is only a few atoms thick. They then used a laser technique called MOKE to probe the data content of the different layers. As they switched a magnetic field on and off they saw in the MOKE signal the data climbing layer by layer from the bottom of the chip to the top.'"
(Line breaks added by me.)
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
