Future of Moore's Law, Physics, and Nano Devices

[lp100]

Hi!

The other day, I read about Moore's Law. To me it seems a miracle how one could predict in 1965 that the number of transistors would double every two years.
However, there has to be an end to this exponential growth due to limits of nano technology, fabrication of semiconductors, fundamental physical laws, and so on.

How small can transistors be nowadays? What might be the future of nano devices? Any thoughts?

I'd like to hear what you guys think!

 

[anorlunda]

Every year that I recall in the past 40 years, pundits predicted the end of Moore's Law. They were always wrong.

A partial explantion is that Moore's Law sets expectations. Industry players expect that if they don't meet Moore's Law improvement levles that they won't survive. Conversely, advances that exceed Moore's Law may be held back as a cushion to meet next year's expectations. I don't think you'll find hard evidence of these things, it is just a reasonable speculation.

Expectations have been met using more than just making things smaller. Ingenuity and innovations of all kinds have been used.

I will not attempt to predict which innovations will prove dominent in the future. But I have confidence that Moore's Law expectations will continue to be met for some time to come.

 

[russ_watters]

The wiki article on the subject has all the relevant details. Currently, the smallest are 20nm and if the trend holds, they will be atomic size in 2020. So I would say that's the limit under a basic interpretation.

Note however that by some related measures and interpretations, limits have already happened. Processor performance used to scale with Moore's law, but that ended 10 years ago, for example.

 

[lp100]

Thank you!

That would mean that devices cannot become smaller than the size of an atom, using the current techniques. Since electrons are much smaller than atoms, would it even be possible to design devices at the size of electrons?

 

[Khashishi]

There's no way to build devices at the size of electrons.

 

[nsaspook]

Intel has 14nm production yields, 10nm in pilot and 7nm in R&D. Most in the industry expect 5nm is possible using current silicon technology. Scaling down past Silicon will be a long time coming due to the costs.