The future of chips and Moore's law

"The first 30 years of the integrated circuit had from two to five times the impact on the US economy, as the first 30 years of the railroad. Or, to put it another way, the transformation of the nineteenth century US economy by the railroad took 60 years, to achieve half the effect, that microelectronics had over 30 years." Kenneth Flamm, `More for Less: The Economic Impact of Semiconductors' December, 1997.

This is the result of Moore's Law, the prediction, which has held oh-so true, that the number of transistors on a chip will double every 18 months. But this just has to come to an end--doesn't it? We have, after all, read believable predictions that the lithography technologies that we currently use to build chips are running into roadblocks.

The thing is, history teaches that at least in this field, we always find ways around, or through, every roadblock that comes up. And it doesn't seem as if that's going to change:

Near term

Encouraging news is coming from Extreme Ultraviolet LLC, a consortium of big chip names that are working hard to make small chips: Brought to our attention by RCFoC reader Grant Perkins, the January 11 ZDNet News describes how the consortium plans to use `extreme UV light,' which has a shorter wavelength than the light used to etch today's .13 micron chips, to create future chips with features as small as .07 microns (70 nanometers). One benefit of such tiny elements is that these chips will run at 10 GHz! 

But this isn't as far as EUV is likely to go--they expect that this technology can eventually create chips with elements as small as .03 microns (30 nanometers), and perhaps even smaller. A very readable overview of what's involved in this process is available from Lawrence Livermore Labs. While describing the EUV process, this paper gives us a fascinating glimpse into just how `perfect' the components of an EUV system must be: 

"...the total thickness of each mirror's coating must deviate less than an atom." 

This is not your father's shaving mirror 

Nor is it science fiction --"We expect to have the first full field-scanned images by April 1," says Chuck Gwyn, Director (EUV), Lawrence Livermore. 

There's another contending technology for our future chips called `Electron Beam Lithography,' or `E-beam.' The major difference is that EUV technology essentially `prints' each layer of a chip all at once (similar to the way a photographic enlarger imprints an entire picture on a piece of paper with one quick exposure.) E-beam, however, uses an electron beam to individually draw every tiny feature on a chip, line by line. It's the difference between a parallel and a serial process. So for E-beam technology to compete with EUV, it will have to draw very, very fast.

Which of these technologies will win the hearts of our next generations of chips? It seems as if EUV has the edge at the moment, but we never know. And as we're about to see, there's always the opportunity for a dark horse technology to pop up, and to change all the rules. 

Farther out

Even with the potential advances in the art of semiconductors that we've just explored, we're eventually going to get to the point where we'll be carving things on our chips that are only one atom wide. And at that point (actually before), the rules that electrons follow in the `larger' world break down. (For example, for elements below 100 nanometers, electrons just don't stay where `they're supposed to,' causing current to leak, which compromises the device). 

So--will these atomic limits spell the end of everything that Moore's Law has wrought, short of our moving into the esoteric realms of DNA computing, molecular self-assembly, or other such obscure fields?

Two recent reports give us a hint of some of the alternatives that researchers are pursuing: 

The first, brought to our attention by Bob Este from the January 10 ElectronicsWeb, describes Toshiba's first room temperature `Single-Electron Transistor,' or SET. Although still very experimental, this is a fully-functional bit of non-volatile memory that draws tiny amounts of power, which may, in Toshiba's words, eventually "...point the way forward toward personal digital assistants (PDAs) with capabilities far surpassing those of today's most advanced personal computers." 

SETs work differently from conventional semiconductors, by injecting a single electron (rather than many) into a tiny bit of silicon, and relying on quantum properties to allow it to retain (and read out) its stored value of one or zero.

The second example, brought to our attention by reader John Hudock in the January 11 New York Times, describes how a team of University of California scientists headed by professor Pierre Petroff, have created `the world's dimmest light source'-- a mushroom-shaped bit of silicon that can be enticed into emitting just one single photon, on-demand. And this, they say, may lead to extraordinary future computers that operate using just a few photons, rather than any electrons at all.

We're all going to have to wait a while to really understand this very new and counterintuitive field of quantum computing; it reminds me of the difficulty I had in first understanding how `holes' could migrate in the early days of transistors. Happily though, we all do have some time to let these quantum rules sink in--"Practical photon-based quantum computers are a long way off," according to Los Alamos scientist Raymond

Laflamme. 

But we shouldn't get complacent--this is an area we do need to become familiar with. And here's one example of why: the January 11 Wired News explains the current thinking, that a quantum computer with only .2 kilobytes of qubits (Quantum Bits), "...would be able to run roughshod over most computer security systems in operation around the world today." 

So, these brief glimpses into the unimaginably strange world of quantum computing may be the harbinger of significant changes to come. One day, for certain tasks, quantum computing might make today's 42-million transistor P4 chips appear as slow, and as antiquated, as the original 30-ton, 17,000-vacuum tube ENIAC computer seems today.

Moore's Law does indeed have a lot of life left in it, as we continue to find ways around, or through, every obstacle in our

path. Of course, that's what drives, and is driven by, the rapidly changing face of computing!

Jeffrey Harrow



Senior Consulting Engineer


(Technology and Corporate Development Group), Compaq


Note: This is an article from the `Rapidly Changing Face of Computing', a free weekly multimedia technology journal written by Jeffrey Harrow. More discussions around the innovations and trends of contemporary computing and the technologies that drive them are available at

www.compaq.com/rcfoc. The writer's opinions do not necessarily reflect the opinion of Compaq. The RCFoC is copyright 2000, Compaq.