New Breakthrough With Transistors For Chips

The Economist news magazine is reporting a new breakthrough in insulation material for transistors in chips to stop electrons jumping ship because of how small transistors have become,

www.economist.com...

As the article states there are anything up to 50 billion transistors on a chip, but they have become so small that electrons cause interference because they can jump from one transistor to another, thus a problematic limiting factor. The new insulating material is called thin-film amorphous boron nitride (a-BN). It can insulate each transistor so the electrons have to behave themselves and be faithful to their home transistor instead of messing up the business at hand.

Moore's Law "that the number of transistors which can be crammed on to a silicon chip doubles every two years" which he predicted in the 1970s was in danger of becoming no law at all because of these pesky electrons causing anarchy everywhere as the transistors were trying to get smaller so they could multiply and work as a team to become ever more empowering chips for human development.

This breakthrough is very important because we were reaching the limits as this article states,

www.technologyreview.com...

Only expense is a future limiation. yet for certain uses money is not an issue. I expect that all this tech will eventually filter down though as manufacture innovates and becomes cheaper.

There are some transistors that take advantage of electron tunneling and those are typically what is found on a processor.

It's a feature of quantum mechanics that electrons will tunnel through things, including many insulators. So this is a serious breakthrough for sure.

Generally with computer tech you need to look back at the patents from the 1980s or before as quite often thats when they discovered the techniques but theres a massive difference between knowing how to do something and getting it to a state for mass production.

Realistically shrinking the die only gives you a certain value as you also need to look at what you want the darn thing to do in the first place and the x86/64 arch is a prime example where theres generally a lot of cruft that could be trimmed out as who really needs hardware enabled BCD functions these days and you could implement the code yourself and while making exe's bigger the fact that your chips run faster makes it a better tradeoff (basically see ARM)

Think some of the fabs are now sampling for 3nm processes and probably smaller so theres still plenty of life left in the current tech tree.

a reply to: Maxatoria

Generally with computer tech you need to look back at the patents from the 1980s or before as quite often thats when they discovered the techniques but theres a massive difference between knowing how to do something and getting it to a state for mass production.

I used to work for a guy that said - "Ideas are cheap. Implementation is expensive"

#truth

Wow.. I guess we will have some wild open world VR mmo in the future.
Here are something to relate to :

Skylake i3 6100 = 1,400,000,000 transistors (1.4 billion)
Skylake i5 6600K = 1,750,000,000 transistors (1.7 billion)
Skylake i7 6700K = 1,750,000,000 transistors (1.7 billion)

Skynet = 40,000,000,000 transistors (40 billion)

a reply to: Riffrafter

a-BN is graphene’s cousin (hexagonal boron nitride instead of carbon) and we are still waiting for graphene tech (got a ski jacket, some bike tires, a tennis racket, a fishing pole (I think), but no “killer app” for graphene. Just saying that it is difficult to realize lab top tech in real world fashion.

Cool news none the less! I keep waiting to wake up in Star Trek world (I hope it is just a matter of time!)

originally posted by: TEOTWAWKIAIFF

Cool news none the less! I keep waiting to wake up in Star Trek world (I hope it is just a matter of time!)