CU breakthrough could shrink computer chips
April 20, 2009
Colorado engineers have found a way to shine a doughnut-shaped laser of light over a second light, thus trimming its edges and offering the hope that computer chips can be made five times smaller.
It’s a potentially huge breakthrough because the entire global electronics industry, and much of the success of the American economy, relies on Moore’s Law, which says transistors and circuits will double in power every 18 months without increasing in size or cost.
The assumption that iPods and computers will continue to grow ever more powerful is running up against the limits of physics, a problem that if not solved could devastate the entire semiconductor industry.
University of Colorado Assistant Professor Robert McLeod and a team of engineering students used tightly focused beams of blue light on liquid molecules known as monomers to record lines and dots thousands of times smaller than the width of a human hair on a substrate.
They then “chopped off the edges” of the lines using a halo of ultraviolet light, trimming the width of the structures significantly.
“We are essentially drawing a line with a marker on a nanotechnology scale and then erasing its edges,” said McLeod, who teaches in CU’s electrical, computer and energy engineering department.
The method offers potential new approaches in the search for ways to shrink transistor circuitry in a global electronic market that is always pursuing smaller, more powerful microchips, said McLeod.
A paper on the subject was published in the April 10 issue of Science Express, the online version of Science magazine. Co-authors included Timothy Scott and Christopher Bowman of the chemical and biological engineering department and graduate students Benjamin Kowalski and Amy Sullivan of the electrical, computer and energy engineering department.
Why not start simply with a skinnier blue light to inscribe the pattern on a substrate? INDenverTimes asked.
“There are some fundamental physics that keep you from doing that,” McLeod said. It’s determined by the wavelength of light. It’s hard to squeeze light into a spot much shorter than its wavelength.”
The law, named after Gordon Moore, co-founder of Intel, states that the semiconductor industry’s growth depends on the confidence that chips can be made ever smaller.
But there seems to be a limit of about 45 nanometers – “that’s incredibly tiny” – to how skinny a line can be made with one color.
“We don’t know how to make it smaller, and that’s a big worry,” he said. “The entire foundation of Silicon Valley is now shaking. The reason you buy a new iPod every year is that you know it can give you twice as much power as the old model” and still be just as small.
So, the CU team and a few others around the nation asked the question, “OK, if we’ve run into the defraction limit, what else can we do?”
“What if we had two colors available and made them do different things?” said McLeod. Once the question was asked, the fun began.
McLeod and his team used a tabletop laser to project tightly focused beams of visible blue light onto the photo-sensitive materials. A chemical reaction initiated a bonding of the monomers into a plastic-like polymer solid, he said.
If the beam was focused in one place, it inscribed a small solid dot. If the beam was moving the focus through the material, it created a thin thread, or line.
“One color is used just like the current process, to make changes in photo-sensitive materials to create a plastic,” he said. “Then you have the other color as a stop.
“You lay two different optical patterns down there, and have it happen at the same time. It’s like a whiteboard marker and an eraser.
You engineer the chemistry such that when the second color hits, it’s the eraser, it’s the ‘don’t go.’ “
Imagine the period at the end of the sentence. To make it smaller, the doughnut-shaped ultraviolet light shines on the same spot as the period is being formed, lopping off its edges.
The process seems capable of shrinking circuits by a factor of five, which is crucial, because as it turns out, polymers 1,000 times thinner than a human hair aren’t skinny enough for the semiconductor industry.
Each time scientists find out a way to make chips twice as small, it’s good news, but requires Silicon Valley to invest some $5 billion in new instruments.
If this process works, it can leapfrog over several smaller incremental changes.
CU and McLeod have filed for a patent on the new process.
“This is the first toe in the water,” McLeod cautioned. “Now we move toward the engineering and see if we can start making things with it, maybe nano-scale robots.”