October 24, 2002 11:00 AM PDT
Tomorrow's tech: The domino effect
- Related Stories
A new era of molecular circuit chipsOctober 23, 2002
Nanowire or nanotube? Intel looks aheadSeptember 12, 2002
IBM chip fights power leaksSeptember 9, 2002
Stopping light could lead to quantum advance in computingJanuary 19, 2001
Scientists at IBM's Almaden Research Center in San Jose, Calif., have built and operated working computer circuits at a nanoscale using an innovative approach in which individual molecules stream across an atomic surface like toppling dominoes.
The new "molecule cascade" technique represents yet another experiment exploring the far reaches of science to find ways to harness the quirky behavior of atoms, molecules and quantum spins as an alternative to silicon--the underlying element powering all commercial computing. A sense of urgency is permeating the semiconductor industry, which foresees reaching the computing limits of silicon in about 10 to 15 years.
"Our job is exploratory research. We are trying to come up with possible scenarios of how you could do computations if--and that's a big if--any technology is going to replace silicon," said Andreas Heinrich, a physicist at IBM and one of the authors of the findings to be published Thursday in the online edition of Science Magazine.
In this case, the scientists formed circuits by creating a precise pattern of carbon monoxide on a copper surface. Nudging a single molecule initiated a cascade of molecules, much like tipping a single domino triggers a laid-out pattern of dominoes to fall in sequence.
They applied this effect to computation by viewing each cascade as a single bit of information. For example, a toppled domino was considered a "1" while a standing domino was thought of as a "0". The researchers applied the same principle to entire arrays of cascaded and non-cascaded molecules where the fallen pattern represented a "1" and those left standing represented "0."
By designing intersections of two cascades, the team created the features required for complex circuits. Some intersections acted as crossovers while others were designed as fan-outs, where one cascade triggers two or more paths.
This method allowed the scientists to create digital-logic elements that are 260,000 times smaller than the ones used in the most advanced semiconductor chips available today. Putting that in perspective, the most complex circuit they built--a 12-by-17 nanometer three-input sorter--is so miniscule that 190 billion of them could fit on top of a standard pencil-top eraser. A nanometer is a billionth of a meter, which is the length of about five to 10 atoms in a line.
"The use of a molecular cascade to do a reasonably complex logic operation represents very clever science," said James Heath, a professor at the University of California at Los Angeles. "This work is pretty stand-alone in its uniqueness--taking a simple system and getting complex behavior out a few simple rules of molecular interaction."
While the experiment has generated great awe and excitement in the scientific community, there are several hurdles to overcome before the technique shows even any promise. The researchers said the experiments could only be carried out using an ultra-high-vacuum, low-temperature scanning tunneling microscope, and it took several hours to set up the complicated cascade patterns in an extremely clean and controlled environment.
"The possibility of an implementation of this new paradigm in real desktop computers is difficult to predict," said Wolf-Dieter Schneider, a professor at the University of Lausanne in Switzerland. "You'll need molecules to do this cascading at room temperature, to do the cascading more quickly and...some 'read-out mechanism.'"
The biggest drawback is that there is no reset mechanism, so these molecule cascades can only perform a calculation once.
"We can do any arbitrary computation once, but it is very important to be able to do the computation repeatedly," said Heinrich. "That is something we take for granted in current silicon technology."
The IBM team--Heinrich, Don Eigler, Jay Gupta and Christopher Lutz--now plans to focus on conducting similar cascade-based computation using other fundamental interactions, like electron spin. Cautioning that the planned research still has to bear out in experiments, the team said it believes that it can reset the cascades, allowing repeated calculations like any current computer circuitry.
"Imagine using magnetic impurities like iron, cobalt or nickel to build atomically precise structures to do cascade computation," said Heinrich. "That would have a reset mechanism using an external magnetic field."
Modern computation is based on silicon technology, which over the past 40 years has shown exponential improvement in speed and integration. Still, the most widely studied element is expected to show its limitations in the coming decades.
Scientists have already looked at other elements on the periodic table related to silicon, but to little avail.
"The problem with any new material is that there are so many material integration issues," said Bob Gasser, the director of Intel's Components Research Lab. The lab is studying ways to extend the life of silicon as well uncover alternatives.
Over the past decade, scientists have turned to cutting-edge science hoping to discover viable methods that one day may succeed silicon. The disparate techniques studied include using basic units of DNA to replicated the actions of a processor to studying the ability to stop and start the motion of light for quantum computing.
While great advances have been made, scientists said they are far from knowing what practical applications will emerge.
"I do know that when you learn how to manufacture and manipulate things at the level of a molecule or a few molecules--as the IBM group has done here--you learn a lot of other things as well," said Heath. "Applications out of this and related work are probably going to come from where you don't expect them."