Chemical reactions break old chemical bonds and form new ones. That's how wax is converted into carbon dioxide and water during the combustion of candles. It is also how small molecules of carbon and hydrogen, called ethylene, are glued together to form the long polymer chains that are the basis of plastics. In both instances a jolt of energy dislodges atoms from one molecule, often initiating a domino-like rearrangement of atoms in the remaining molecules.

Chemists used to liken chemical reactions to a hike over a mountain. The height of the mountain represents the amount of energy required to kickstart the reaction -- called the reaction's activation energy. The implication was that once a reaction was underway, it was a straight shot to the trail's end.


In reality chemical reactions are more like rollercoaster rides -- a series of swales and summits, with each dip representing an intermediate chemical structure and each incline an energy requirement that must be met for the reaction to continue.

Operating under the old analogy, chemists are limited in the degree to which they can control a reaction. Even relatively simple reactions like polymerization require precise, minute changes to produce the desired results. Most industries employ catalysts -- molecules, often metals that lower activation energies by subtly redistributing the electrons surrounding an atom's nucleus. A catalyst that lands an atom nanometers off target or ends too soon may fail to improve the reaction rate or may yield an entirely different compound. A reaction to form a polymer may have similar problems if it is slightly off. For example, long-chain polymers form hard plastics that polymers with branching structures.

Having refined the reaction as much as possible based on a limited, industry and other groups are looking to scientists like Morokuma for more options. Morokuma's maps of reaction pathways, also called potential energy curves, reveal all the possibilities. His maps of catalysts interacting at various stages in a reaction help industry narrow the choices of catalysts, which they then test experimentally.

But reaction pathways are computationally costly to generate. The energy associated with bonds breaking and reforming can be captured only through computationally expensive quantum mechanical equations that yield results within 1-2 kilocalories, which is the same reliability as experimental results. Looking at just one reaction step -- one trip over the mountain -- can take days to weeks on four processors of NCSA's SGI Origin2000. And that's only for the 20 to 40 atoms that constitute a catalyst's reaction center. Trying to account for the influence of the rest of system on the reaction lengthens the calculation tenfold -- at every step -- for each time the number of atoms doubles.

1 2 3 4