A rchive Date
[ 19-08-2002 ]
Category
[ Science ]
sub-Categoy
[ Physics ]
|
[
http://www.pcmag.com/article2/0,4149,429435,00.asp
Tech Frontiers
By Erik Rhey September 3, 2002
Sixty million transistors were manufactured last year for every man, woman, and child on Earth. By 2010, that figure will reach 1 billion transistors a year. And if you happen to live in a major city, you'll soon be enveloped by radio signals from mobile phones. Futurists predict that by 2006, mobile-phone signals will blanket our planet, with 1.6 billion people worldwide subscribing to mobile-phone services.
Whether we like it or not, technology engages our lives. We can decide whether to own a home computer, cell phone, or PDA, but we are bound to the fact that every telephone call and bank transaction, every turn of a car's ignition, and even most of the movies we watch are sired by technology created within the past ten years - or even the past three.
Imagine recording your favorite TV show and beaming it to your cell phone, or playing an online game from your wireless PDA.
In the following pages, we leave the present behind, rocketing five and sometimes ten years down the road. What we've found will astound you. We bring you the future in the four areas of technology that we consider the engines of advancement on many levels: chip fabrication, software programming, security, and entertainment.
In his essay on programming, author and New York Times technology reporter Steve Lohr examines how future programming schemes are taking a tip from nature and going organic. The programmer's pie in the sky is a concept called genetic programming, in which programming languages mimic the biological process. Although we're not there yet, Lohr explains that we're coming close with aspect-oriented programming, which can automatically update and change itself across complex hierarchies, and intentional software, which can write its own code and customize itself to the needs of each user.
Such software, however, is useless without the processing power to run it. And at the heart of the processor are transistors—millions of them. In his article on chip fabrication, industry analyst Jim Turley points out that although we're close, we've not yet reached the limit of how small a chip can be or how many transistors can fit onto one of these dinky powerhouses. Companies are in a full sprint to make the smaller and faster transistors that will drive future technologies. Turley elucidates the processes underway to make such transistors and the challenges chip makers must overcome.
PC Magazine senior editor Konstantinos Karagiannis peers through the looking glass at a topic on everyone's mind—security. As he points out, current techniques used by antivirus programs and firewalls won't be enough even in the near future. Viruses and worms are morphing into blended threats, which use an arsenal of methods to attack multiple points of entry. And even in the air, our data is not safe: New security contraventions are going after wireless devices such as PDAs, cell phones, and notebooks with VPN connections, rendering firewalls useless.
The good news is that developers are working overtime to produce products that bring much-needed reform to the security industry. We can't stop criminals, terrorists, and hackers from using computers, but new technology can prevent them from using computers as weapons.
Then there is the truly unifying technology—entertainment. It identifies who we are, what we want to be, and what we love and hate. From "The Jerry Springer Show" to "Nova," and from Citizen Kane to Police Academy, what entertains us also serves as the historical record of our cultural identity. As writer and VRML inventor Mark Pesce shows us, future innovations in entertainment technology will make today's products seem like 19th-century stereoscopes. Pesce dives into the issue of convergence in entertainment media, focusing on how MPEG-4 is revolutionizing digital video and TV. Imagine recording your favorite TV show and beaming it to your cell phone, or playing an online game from your wireless PDA—and that's just the beginning.
Viewed from our mountaintop in 2002, here is the future.
Future Fab
Semiconductor fabrication companies deserve most of the credit for the extraordinary pace of technological innovation. Like village blacksmiths or western ironmongers, these unsung heroes pound away at stubborn problems, forging the materials and technologies that provide the rest of us with the necessities and gadgetry of a modern age.
In one of the industry's supreme paradoxes, silicon transistors are virtually free yet also prohibitively expensive. More than 60 million transistors were manufactured last year for everyone on Earth. They're as common as grains of rice.
The economics of chip production are so utterly unlike those of other industries that standard rules don't apply.
But the factories that produce these mighty mites are so expensive that consolidation, partnership, and cooperation are essential. Companies and countries that were rivals are now uneasy associates. They delve into the murky realms of next-generation fabrication with the economic and scientific hopes of the world behind them.
The economics of chip production are so utterly unlike those of other industries that standard rules don't apply. The cost of materials is almost zero, but the cost of tooling is phenomenal. A new chip-making factory costs more than a billion dollars to set up and equip, and it's obsolete inside of three years. That's a burden of $1 million per day in amortization hanging over every foundry.
The mask set, or film, for a new type of chip runs another million dollars and must be replaced with every modification. Raw materials like silicon and aluminum are practically free, because chips use so little of them. Unlike with TVs and cars, the incremental cost of producing more chips is negligible. That first chip is very expensive. After that, they're all free.
We've all been introduced to Moore's Law ad nauseam, that often-misquoted rule of thumb that says we'll cram twice as many transistors onto a square inch of silicon as we did 18 months ago. In any other industry, a 58 percent compound annual growth rate (CAGR) would be astounding. Yet the chip-making industry treats its CAGR as a birthright. Anything less is viewed as a failure by investors, pundits, competitors, and national governments.
Keeping up requires continually shrinking silicon transistors and squeezing in more per square inch. Faster Pentiums, smaller cell phones, and smarter antilock brakes all owe their existence to the ever-shrinking transistor. Already, these minuscule building blocks of technology are smaller than bacteria. Today's silicon transistors are only a few dozen atoms thick. Yet further shrinkage leads to manufacturing problems of a new nature.
Behavioral Issues
Some of the problems are well understood, which is why electron-beam and X-ray lithography are replacing the traditional techniques. Manufacturing future chips is the easy part. Predicting their behavior is another matter entirely. When transistors shrink to a near-atomic level, unintended side effects emerge. Solid-state electronics becomes squishy quantum mechanics. Chip making gets weird.
All chips connect tiny silicon transistors with tiny aluminum or copper wires. Now that transistors are becoming smaller and faster than these wires, speed is limited by the time it takes electrons to move through wire, which is about half the speed of light. Faster transistors are irrelevant; shorter wires are the key. The best way to make wires shorter is to make chips smaller, so we're back where we started. Transistors are now obstacles around which wires must be routed.
Today's smallest transistors are only 0.1 microns across (about 0.000004 inches), compared with the 0.8-micron transistors on the first Pentium. As transistors become smaller and wires shorter, new behaviors set in. Transistors and wires are already starting to act in ways that classic electronics theory didn't predict. Individual electrons go renegade and tunnel into places they don't belong.
In today's chips that channel a trillion electrons at once, these anomalies are benign. But when tomorrow's chips massage a few hundred electrons here and there, a few loose ones can upset everything. Probability and quantum mechanics will play bigger roles. Future chips will be mostly air, with delicate bridges of silicon and copper spanning chasms of empty space. Sapphire, diamond, and SON (silicon on nothing) buttress the structures as fabricators strive to preserve classical electronics for a few more years. Beyond that, optical or quantum computing may hold sway.
Making Masks
The million-dollar mask set is one area of progress. Each mask is the template for millions of identical chips. If the mask has any flaws at all, they'll be replicated in every single chip, producing lots of shiny but useless little squares, destined for paperweights.
The features on the mask are smaller than the wavelengths of visible light; they're literally invisible, even under a microscope—not to mention hopelessly complex. No human can tell whether a mask is perfect, but the economic pressure to assure it is flawless is huge. Future fabricators will rely on automated knowledge systems to verify a mask before it's used.
Chip fabs are rarely razed and built anew. Instead, they're continually updated and have empty rooms for future improvements. Chip makers have no idea what will go into those rooms, but they're pretty sure that someday, something important will be invented that they can slide right in.
Jim Turley is an industry analyst and technology writer. |
|