Models—Experiment—Computation: A History of Ideas in Structural Chemistry

Abstract

Ideas about chemical structures have developed over hundreds of years, but the pace has greatly accelerated during the twentieth century. The mechanical interactions among building blocks of structures were taken into account in the computational models by Frank Westheimer and by Terrel Hill, and Lou Allinger’s programs made them especially popular. G. N. Lewis provided models of bonding in molecules that served as starting points for later models, among them for Ron Gillespie’s immensely popular VSEPR model. Accounting for non-bonded interactions has conveniently augmented the considerations for bond configurations. The emergence of X-ray crystallography almost 100 years ago, followed by other diffraction techniques and a plethora of spectroscopic techniques provided tremendous headway for experimental information of ever increasing precision. The next step was attaining comparable accuracy that helped the meaningful comparison and ultimately the combination of structural information from the most diverse experimental and computational sources. Linus Pauling’s valence bond theory and Friedrich Hund’s and Robert Mulliken’s molecular orbital approach had their preeminence at different times, the latter finally prevailing due to its better suitability for computation. Not only did John Pople build a whole systematics of computations; he understood that if computation was to become a tool on a par with experiment, error estimation had to be handled in a compatible way. Today, qualitative models, experiments, and computations all have their own niches in the realm of structure research, all contributing to our goal of uncovering “coherence and regularities”—in the words of Michael Polanyi and Eugene Wigner—for our understanding and utilization of the molecular world.