One thing Gordon Moore wasn't thinking about when he contributed an article to the 35th anniversary issue of Electronics magazine was creating a law of any kind.
But his prediction – because that's what it was – has indeed turned into a law and a law which has driven the semiconductor industry to its current levels of excellence.
What has now become known as Moore's Law started with an observation by Moore – who was still with Fairchild at the time – that the number of transistors per unit area of silicon was doubling on a regular basis and that the progression was likely to continue.
What he actually said was 'the complexity for minimum component costs has increased at a rate of roughly a factor of two per year. Certainly, over the short term, this rate can be expected to continue, if not to increase. Over the longer term, the rate of increase is a bit more uncertain, although there is no reason to believe it will not remain nearly constant for at least 10 years'.
A decade later, with the industry still following the prediction closely, Carver Mead, a professor at Caltech, dubbed the observation Moore's Law.
Although the industry hasn't adhered to Moore's Law exactly since then, it still uses the concept as its engine of progress.
One interpretation of Moore's Law is the number of transistors per unit area. Another is the cost of making those transistors; Moore himself recognised that process shrinks, as they have come to be known, were going to bring significant cost benefits to the industry. Now, if a leading company isn't making chips on the lastest process, it isn't making them as cost effectively as it could.
What is amazing about Moore's Law is the complexity – or lack of it – of integrated circuits at the time. Moore noted at the 40th anniversary of the article's publication: "The most complex IC we had then had just 30 components. And we had one with 60 components that would be introduced in the next year."
Here's an interesting contrast. The 4004 microprocessor, launched by Intel in 1971, contained 2300 transistors. Today, the 5th Generation Intel Core processor features 1.3billion. And, according to Intel, more than 6million of its latest Tri-Gate transistors can fit on a full stop.
The question now is how long Moore's Law can continue to be applied by the industry as silicon begins to encounter the laws of physics. Moore himself has reservations. Speaking a couple of years ago, he expected industry progress to continue 'for a decade or so'.
One of the major roadblocks has recently been overcome with the development by Intel and others of the FinFET approach. Because linear scaling is becoming increasingly difficult, the gates of the transistor have been turned on their side to allow more devices to be fitted into a given area. This is allowing technologists to address nodes smaller than the 14nm and 16nm processes beginning to enter production.
Being able to squeeze twice as many transistors into a given area of silicon on a regular basis becomes increasingly difficult. Feature sizes are now some ten times smaller than the wavelength of light being used to create them and this is pushing technology to the limits.
The challenge for the industry now is to move from immersion lithography at 193nm to extreme ultraviolet lithography: a move which has taken significantly longer than expected.
Talking to Intel earlier in 2015 about his Law, Moore said: "The industry has been phenomenally creative in continuing to increase the complexity of chips. It's hard to believe – at least it's hard for me to believe – that now we talk in terms of billions of transistors on a chip rather than tens, hundreds or thousands."
Author
Graham Pitcher
Source: www.newelectronics.co.uk