As we all know, computer chips have been getting smaller, faster, and cheaper year-by-year for decades now. Today, some $10 pocket calculators pack more power than the most sophisticated, multi-million-dollar mainframe computer of the mid-1950s.
As pointed out in the Los Angeles Times, if the same rate of performance improvement versus cost held for airlines, a flight from New York to Paris would cost one cent and last less than a second.
This has been made possible by ever denser and more complex integrated circuits.
- In 1971, Intel’s first microprocessor chip had 2,300 transistors.
- By 1978, the 8086 processor had 29,000 transistors.
- Less than a decade later, the 486 housed 1.2 million transistors.
- Last year, the Pentium 4 reached 410 million.
- The Itanium 2, due this year, will run big servers with its astounding 1.7 billion transistors ? all on something the size of a postage stamp.
At the same time, manufacturing has been improving at an astonishing rate, with micro-lithography taking only one second to lay down all those transistors.
As a result of these converging trends, the cost of a transistor fell by a factor of more than 10 million between the 1950s and today, according to a report in the Daily Telegraph.
In 2004, more transistors were produced than grains of rice ? and they cost less. There were 10 quintillion transistors shipped in 2003 ? that’s a “10” with 18 zeroes after it, or 100 times the number of ants in the world.
These astonishing numbers ? and the improvements in computer chips ? were achieved by building transistors closer and closer together on smaller and smaller bits of silicon. And when it comes to chips, smaller is faster and even cheaper ? up to a point.
But when they run that fast, fully half the electricity is wasted as heat ? they run hot enough to cook an egg. Moreover, a factory to build these chips costs $2 billion and covers four football fields ? the most expensive factory ever built on a per-acre basis. Even in a $213 billion industry, that cost is getting too high to pay.
With almost two billion transistors on a half-inch-square chip, the technology is reaching its limits. For example, back in 2003, a defect 20 nanometers wide could be ignored in a Pentium-3, which had circuit lines 180 nanometers wide. But a 20-nanometer defect is a show-stopper for today’s chips with circuit lines a mere 90 nanometers wide.
The fact is, Moore’s Law, one of the fundamental trends driving our civilization, is approaching a dead end as chip components start bumping up against the ultimate size of the atoms themselves. Within a decade, super-chips will run hot enough to melt metal, according to Business Week. And the manufacturing plants to make them will cost $10 billion. Bernie Meyerson, an IBM vice president, told The Seattle Times in an interview that alternatives to silicon will have to be found at the latest by 2020.
What’s to be done? The answer is to nanotechnologize computer chips. And, in the next 20 years, it will probably be the biggest payoff we’ll see from the broader nanotech trends we’ve previously discussed in Trends.
In pursuit of this payoff, the big players in the trillion-dollar electronics industry are scrambling to be the first to crack the nano-code, the holy grail of computer technology.
Among them is IBM. It has spent the last four years assembling a third-generation prototype of a carbon nanotube transistor. It’s up and running and can carry 1,000 times the current of equally sized copper wires. More importantly, it produces a fraction of the heat of a conventional nano-scale chip and promises to be far easier to make.
The nanotube is a honeycomb of atoms shaped into a cylinder not unlike a coil of chicken wire. By changing the geometry of the hexagonal bonds between its atoms, researchers can tune it either to conduct or block electricity. This on-off ability can represent a binary one or a zero, the basic building blocks of computer information.
Transistors made from nanotubes offer many advantages over traditional chips. Manufacturing them requires no toxic chemicals. They’re 100 times stronger than steel. They’re resistant to radiation. And researchers at IBM have already built nanotube transistors with electrical properties far superior to those of silicon.
Meanwhile, Intel is feverishly working with Stanford University researchers to produce nanotubes in a secret lab in Oregon. And Hewlett-Packard, which manufactures four times more silicon than Intel, is developing a novel technology called “crossbar latches” to replace transistors. These are grids of platinum wires with molecules at all the junctures that can conduct electrical signals. HP can now produce crossbars 30 nanometers wide and spaced 100 nanometers apart, which eliminates many of the problems involved in conventional micro-transistor technology.
In the course of fundamental research, nanotechnology has already improved by leaps and bounds and it has been applied in many areas having nothing to do with computing. The biggest markets for nanoparticles so far are in familiar products, ranging from automobile tires to the silver compounds used in conventional photography.
It’s already, a $13 billion industry. But, as explained in the April 2005 issue of Trends it’s set to explode. In fact, all over the world, laboratories are manipulating the very building blocks of matter for use as everything from cancer therapies to electrical generators and from false teeth to tennis balls, according to a report in BusinessWeek.
Given these striking trends, we offer the five following forecasts for your consideration:
First, in the next five to 10 years, nanotube technology will gradually be integrated with existing chip technology to improve performance in incremental ways and eliminate some of the heat and power problems manufacturers are now facing. For example, nanotubes could replace the wires now used in chips to conduct signals. The industry will also begin using a new type of insulator to separate those wires, known as a high-k dielectric insulator, to improve the efficiency of silicon. And chip makers will continue to confront Moore’s Law simply by finding ways to write smaller patterns on conventional chips. One research group, for example, has spent several hundred million dollars on extreme ultraviolet lithography, which writes at extremely small dimensions.
Second, also in the near term, nanotube technology will be applied outside the computer industry, finding uses in a dazzling array of devices from airport security bots that can sniff anthrax to medical devices implanted inside the body to dispense drugs such as insulin. At the same time, computer companies will find novel ways of capitalizing on the special properties of nanotubes. For example, a startup called Nantero is about to begin manufacturing a nanotube memory stash about the size of a credit card that holds 1,000 gigabytes.
Third, one branch of nanotube technology will veer away from traditional manufacturing approaches and begin using biological agents, such as DNA, to construct devices, including chips, out of nanotubes. This technique uses the double helix structure of DNA to wrap around nanotubes like stripes on a barber pole, giving them rigidity for easier handling and guiding them into place. Enzymes are used to cut the DNA, typically from simple bacterial organisms for a perfect fit. From these developments, a separate industry of bioengineering will grow off in its own direction and find its own markets, including chip manufacture.
Fourth, government and private funding for nanotechnology will undergo explosive expansion within the next 25 years. Money is already being poured into nanotech research.6 The National Nanotechnology Initiative now oversees $1.2 billion in funding at 23 federal agencies for nanotech. Individual states are spending another $400 million. And 1,600 U.S. companies are researching nanotech. New Jersey recently spent $2 million to launch the New Jersey Nanotechnology Consortium, which was then bought up by Lucent. The consortium has 60 customers, from universities to military and intelligence agencies. This will be the single hottest growth area for future investors for some time to come.
Fifth, as nanotube transistors mature, in the 25-30 year range, even more exotic computer components will come on line. As we’ve seen with silicon, no technology lasts forever. Intel, IBM, and others are even now looking beyond the nanotube toward optical computing that manipulates photons; harnessing the spin of electrons to create quantum transistors; and using “quantum dots.”
References List : 1. Los Angeles Times, April 17, 2005, “Law of Continuing Returns,” by Terril Yue Jones. ⓒ Copyright 2005 by The Los Angeles Times. All rights reserved.2. The Daily Telegraph, April 13, 2005, “The Huge Impact of a World That Keeps Shrinking,” by Alec Broers. ⓒ Copyright 2005 by Telegraph Group Limited. All rights reserved.3. BusinessWeek, April 18, 2005, “The Coming Chip Revolution,” by Adam Aston. ⓒ Copyright 2005 by The McGraw-Hill Companies, Inc. All rights reserved.4. The Seattle Times, April 18, 2005, “Dreading the Demise of Ever-Expanding Microchip Capacity,” by Dean Takahashi. ⓒ Copyright 2005 by The Seattle Times Company. All rights reserved.5. BusinessWeek, October 11, 2004, “Universe in a Grain of Sand,” by Stephen Baker and Adam Aston. ⓒ Copyright 2004 by The McGraw-Hill Companies, Inc. All rights reserved.6. The Star-Ledger, April 3, 2005, “Big Things Start Small,” by Kevin Coughlin. ⓒ Copyright 2005 by The Star-Ledger. All rights reserved.
