Hitachi joins IBM's quest for sub-32nm semiconductor technology

As of this morning you can add Hitachi to the list of cohorts IBM has gathered in its quest for sub-32nm circuitry. Hitachi's 2-year semiconductor research agreement -- a first between IBM and Hitachi -- puts them under a loose-knit alliance with AMD, Chartered, Freescale, Infineon, Samsung, Sony, Toshiba and STMicroelectronics. The notable, non-member here is of course, Intel; that little company making "45-nm processes" and "Hi-K metal gate" house-hold terms amongst geeky, type-A adopters of consumer electronics. Yes, we're looking at you.

[soviet_vexxer @ Mar 10th 2008 6:15AM]

how small can these things theoretically get? i mean at some point (many years from now) some kind of new technology will have to take over wont it? could these get down to a fraction of a nm?

[Zappacky @ Mar 10th 2008 6:30AM]

The problem is going to occur when things hit the 'few atoms thick' size. You can't do a whole lot with something that consists of only 1 atom, so that's going to be our lower bound for size.

In addition, quantum tunneling will pose a serious problem around the 'few atoms thick' area too. MOSFETS have to deal with this problem already.

[soviet_vexxer @ Mar 10th 2008 6:39AM]

i found this randomly (remembers relays and tubes)

Extrapolating the trend, optimists should expect another radical change by 2015, which happens to coincide with the date when the fastest computers will match brains in terms of raw computing power, according to frequent estimates.

The remaining series of faster and faster additional revolutions should culminate in an "Omega point" (Teilhard de Chardin, 1916) expected around 2030-2040.

[Leo @ Mar 10th 2008 1:42PM]

yeah great predictions. more likely is that by 2040 we'll have computers that are self aware and more intelligent than humans (and obviously MUCH faster thinking). MY prediction is that there'll be robot unions and robot political parties because they'll want representation as we'll treat em like the slave purposes they were built for... they wont like that... some of em will probably kill themselves when they become self aware... robot depression - that's gonna be weird.... oh yeah and human robot war duh.

Sorry to sound like a whiney bitch but stop using the word random incorrectly. (vote me down accordingly)

[francisco @ Mar 10th 2008 6:32AM]

isnt it crazy how all these companies have to form a partnership just to compete with Intel? No one company can go at it alone. Crazy

[Sam Winter @ Mar 10th 2008 7:07AM]

The next step after 32nm is 22nm, then 16nm. Where they go from there is up in the air. It has been claimed that transistors cannot be scaled below the size achievable at

16 nm due to quantum tunneling, regardless of the materials. But obviosly these companies are working on something.
It's actually really crazy. Check out these size comparisons from wiki

90nm - size of HIV virus
45nm - Intels current manufacturing process
20nm - size of a bacterial flagellum
6nm-10nm - size of a cell membrane
2nm - diameter of DNA helix
1nm - diameter of single-wall carbon nanotube
225pm (0.225nm) - diameter of cesium atom
53pm - Bohr radius (distance between hydrogen atom and it's electron)
0.001pm - size of a proton


http://en.wikipedia.org/wiki/Orders_of_magnitude_%28length%29

[Kurian @ Mar 10th 2008 8:05AM]

That picture is starting to get a bit annoyingly overused.

[John @ Mar 10th 2008 8:35AM]

What's the cell processor got to do with this? Isn't it still 45nm?

[David @ Mar 10th 2008 10:03AM]

"isnt it crazy how all these companies have to form a partnership just to compete with Intel? No one company can go at it alone. Crazy."

It's not about competing with Intel per se (except in prestige), as the alliance partners serve a different market. Intel is in a relatively unique position of owning the market for a product that is both high volume and relatively high in cost. This allows them to easily keep their fabs full and to amortize their R&D and fab construction costs over a lot of parts. For the various alliance partners, each of which serves either a lower volume or a lower cost market, it's harder to earn enough to justify the huge costs and partnerships to share costs makes more economic sense.