Rabitz's office at Princeton University is loaded with the props and toys that so many chemistry teachers rely on to explain the sub-microscopic world -- rotating magnets, stacking blocks and the little Tinker-Toy balls and sticks that represent atoms bonding together.
It's all so neat. Pull off a carbon atom here, add an oxygen there, and voil`, something new!
Reality is much messier. Chemists have developed sophisticated techniques for moving atoms in molecules, but the process can be arduous, especially for bigger molecules. A single, simple tool for doing such work has, until now, been little more than a dream.
Rabitz has pioneered a technique that gives scientists at least the first prototype of such a tool. It is essentially a cut-and-paste function for editing molecules, allowing chemists to remove and rearrange individual atoms or groups of atoms from larger clusters.
In addition to fulfilling a longstanding goal among chemists, the approach is proving to have widespread uses in other fields that deal with very small-scale interactions. Other researchers have used variations on the idea to produce notable results in areas from electronic chip development to fiber-optic communications.
The technique involves tuning a laser to emit just the right colors, or frequencies, of light so the mixture resonates with chemical bonds and breaks them apart. The basic idea has been around for 40 years, but had not been successfully carried out because it seemed to require an overwhelming amount of calculation to tune the laser correctly.
In 1992, Rabitz and his former graduate student, Richard Judson, proposed a radical solution -- part brute force and part cunning strategy. Then in a recent article in Science magazine, Rabitz and two other collaborators, Robert Levis and Getahun Menkir of Wayne State University, reported that they not only used the technique to selectively snip apart molecules, but also to stitch them back together in a new configuration.
Closing the loop
Rabitz's technique harnesses the power of the molecule itself to solve its own complex equations and uses a computer to "read" the answers. Starting with a random mix of light waves, the computer guides a trial-and-error process of zapping molecule after molecule with different combinations of light until a clever piece of pattern-detecting software recognizes that the molecule has begun to move toward the desired result. It makes adjustments, zaps again and observes whether the new result is better or worse. Each mini "experiment" takes about 1/1,000 of a second.
"If you can do a million experiments in an hour, you have a tremendous lever," said Rabitz, who is the Charles Phelps Smyth Professor of Chemistry at Princeton. "Maybe the first 100,000 don't tell you much, but you can just keep on going."
In one example from the research reported in Science, the scientists removed a cluster containing a carbon and three hydrogen atoms from an acetone molecule, which has two such clusters. More surprising, they managed to remove the middle section of a three-part molecule and join the two end pieces directly together, forming a different chemical.
What had made this sort of thing so difficult was the complexity of interactions between atoms. For any two atoms bonded together, chemists could calculate fairly well what frequencies of light would resonate with the pair and shake it apart. But when many are bonded together, applying a force to one bond affects all the others and the calculations become impossible using current methods and computers.
Rabitz's technique is called a "closed-loop" method because decisions about how to adjust the laser beam are based solely on the results of the previous experiments, not on any external guidance, such as a theoretical calculation of how a bond is likely to react. In the end, the computer and the molecule "discover" the answer even though the computer was given no prior information about the physical principles that govern chemical bonding.
Century of control
The challenges that arise at this level of chemistry turn out to have quite a bit in common with problems scientists face in many other areas. At very small scales, interactions are governed by "quantum" phenomena -- the way in which electrons jump from one energy level to another, absorbing and discharging packets or "quanta" of energy. This process is notoriously inscrutable. In 1929, German physicist Werner Heisenberg postulated the "uncertainty principle," which implies that the mere act of measuring an electron's position or energy will change it, rendering the measurement of dubious value for making decisions about how to further manipulate it. Quantum phenomena, it seemed, were fundamentally uncontrollable.
Rabitz's closed-loop approach gives scientists control over quantum events.
"It really gives us insight into how atoms can be manipulated by light," said Margaret Murnane, a physicist at the University of Colorado. While many advances of 20th century physics could be seen as coming from the observation of atoms, advances of the 21st century may come from the control of their activity, Murnane said.
In computer chips, when semiconductors, which act as tiny on-off switches, get crunched too close together they start to interfere with each other in unpredictable ways. Similarly, scientists have begun to try to make computers in which the quantum positioning of electrons acts as the on-off switch. Such a "quantum computer" could be incredibly small and powerful, but would require supreme control over an unruly phenomenon.
Rabitz believes the closed-loop approach could give scientists the tool they need to make such projects work; a computer could quickly search for and discover the unique combination of light waves to produce the desired result.
Murnane and fellow Colorado physics professor Henry Kapteyn used closed-loop control to greatly increase X-ray emissions from atoms. They tailored a laser beam so that atoms in its path emitted all their X-rays in unison, instead of in a blur of interfering waves.
Kapteyn said the experiment is yielding significant insights into the physics of X-ray generation. "You can learn a lot by taking in so much data," he said. "You are crunching through tens of thousands of experiments that you have done in one afternoon."
Others have used a closed-loop system to negate the distortion that occurs when a strong light pulse overwhelms the capacity of a fiber optic cable, like music played too loudly over a stereo. The researchers, at Los Alamos National Laboratory, engineered an optical signal that appeared distorted at first, but emerged from the cable as a clean signal, like putting a tape of noise into the stereo, cranking the volume and hearing perfect music from the speakers.
As for uses of the method in chemistry, Rabitz said it is too early to tell whether laser manipulation will be cost effective enough for high-volume production of chemicals, but it certainly will be useful for manipulating substances in the lab. Because of its general applicability, he said, "it opens the prospect of going into the chemistry stockroom and taking every bottle off the shelf and dropping its contents into the apparatus and seeing what we can do."
Hints behind the jungle
Rabitz acknowledges that, although the technique produces dramatic results, it does not immediately give scientists insight into the workings of the quantum mechanical world.
"The original goal was to do attain control over quantum mechanical phenomena -- and if you can do it, great," he said. "It's a different question, but a very fundamental question, to ask, 'Well, what happened? What is the mechanism here?'"
Nonetheless, Rabitz believes the approach will ultimately shed light on the quantum mechanics that it so neatly bypasses. He has an idea for injecting test signals -- a sort of optical watermark -- into the laser pulses and then observing how those signals are distorted during the closed-loop learning process. Analyzing that distortion may reveal hidden patterns and principles behind what appears to be a muddle of quantum interactions.
"I am an optimist," said Rabitz. "Behind this jungle, there are hints."
Contact: Marilyn Marks (609) 258-3601
