NPACI Computation Supports Rewriting the Reference Book on Boron Chemistry

Published 04/30/1998

For information, contact:
Ann Redelfs, SDSC, 619-534-5032, redelfs@sdsc.edu

SAN DIEGO -- Boron plays a key role in compounds used in the production of semiconductors, photovoltaics, high-temperature protective coatings, and other industrial and consumer applications, yet a recent computation by chemists with the National Partnership for Advanced Computational Infrastructure (NPACI) suggests that the definitive reference on a fundamental property of boron is wrong.

The work, published in the April 30 Journal of Physical Chemistry, was conducted on the CRAY T90 at the San Diego Supercomputer Center (SDSC), the leading-edge site for the Partnership. Funded by the National Science Foundation through the Partnership and a chemistry grant, the research has produced the first NPACI publication.

"It came as a surprise to me that for an element as important and simple as boron that there was any uncertainty, but there is," said Peter Taylor, computational chemist at SDSC and leader of the Partnership's Molecular Science thrust area who, with Jan Martin of Israel's Weizmann Institute of Science, conducted the boron computation. "If you do industrial chemistry for a living, this is a key piece of information."

In developing an industrial process, chemists must calculate whether a given chemical reaction releases heat or requires heat to be added. To make this calculation, a chemist must know the "heats of formation" for the compounds and elements in the reaction, and this information can be found in standard reference tables. The fundamental quantities in these tables are the heats of formation for the elements.

The accepted heat of formation for boron in the JANAF Thermochemical Tables has an error of more than two percent. A larger heat of formation value, with an error of less than 0.1 percent, was suggested by a 1977 experiment but rejected by the compilers of the JANAF tables.

The Taylor and Martin computation, with an error of less than 0.5 percent, almost exactly matches the rejected 20-year-old value. "Even the generous error bar in the tables is not enough to hold the higher value, meaning the currently accepted value is just wrong," Taylor said.

The new result comes from an ab initio computation -- one that works from the fundamental physics equations describing the atoms -- of a reaction for which the initial reactants' heats of formation are precisely known, in this case boron trifluoride and fluorine. Taylor and Martin first calculated a binding energy for boron trifluoride accurate to within 0.1 percent.

With the computed binding energy, a heat of formation for boron can be easily calculated with an error of less than 0.5 percent -- considerably more reliable than the reference.

"This is an excellent paper which resolves an important and long-standing experimental controversy," said Karl K. Irikura of the Physical and Chemical Properties Division at the National Institute of Standards and Technology, the compilers of the JANAF tables. "The authors have demonstrated that theory has superseded experiment for this type of chemical problem. As these methods become adopted elsewhere, we can expect the quality of thermochemical data to increase rapidly. This benefit will eventually carry over to chemical process simulations. Such simulations are not only important in the chemical industry -- they are critical to the atmospheric predictions that form the basis for global environmental policy."

The computations used the T90 at SDSC and, in particular, consumed a significant amount of disk space -- 60 gigabytes' worth. "Getting access to 60 gigabytes of disk storage is not easy to do on a workstation-class machine," said Taylor, who is also the Partnership's chief applications scientist. "This is an example of the other sorts of questions that require access to the type of high-performance computing facilities provided by the Partnership."

With the revised value for boron's heat of formation, industrial chemists will be able to devise more efficient manufacturing processes and have a better understanding of the boron compounds used in protective coatings, rocket propellants, iron and special-purpose alloys, and industrial and consumer cleaners.

SDSC is a research unit of the University of California, San Diego, and is sponsored by the National Science Foundation through the National Partnership for Advanced Computational Infrastructure and by other federal agencies, the State and University of California and private organizations. For additional information on the Partnership and SDSC, see http://www.sdsc.edu/ and http://www.npaci.edu/.