A new computational approach to accurately and reliably divide a molecule's electronic properties into its component atom parts has been developed by scientists in the UK.

While chemists typically think of the properties of molecules as a sum of their atomic parts, molecular characteristics such as electron density are impossible to experimentally divide into their atomic components, and have proven difficult to derive computationally. Now, Timothy Lillestolen and Richard Wheatley at the University of Nottingham have developed a simple algorithm that they say gives atoms with the smooth, spherical shapes and charges that would intuitively be expected.

The method can answer the important question "How many electrons are there on each atom of a molecule?"

'If you ask a chemist whether the properties of a molecule can be understood in terms of the properties of its constituent atoms, you are likely to get an affirmative answer,' says Wheatley. 'However, the tools available for accurate chemical calculations treat the molecule holistically, and do not allow information on separate atoms to be easily extracted.'

To tackle the problem, the team developed an iterative stockholder approach to calculate the electron density of each atom. Electronegative atoms were correctly predicted to have negative charges, while hydrogen atoms bonded to electronegative atoms had a positive charge.

'The beautiful results produced by the iterated stockholder method have taken us by surprise: if it has an Achilles' heel, then we have yet to discover it!' says Wheatley. 'However, our results so far are preliminary, and we are still working on applications of the method to a wider range of chemicals,' he adds. The team are in talks with several software companies over the possibility of incorporating the work in quantum chemistry programmes.

'This is a slick and simple idea that provides a way to answer a question that is in most chemists' minds: how many electrons are there on each atom of a molecule?' says Peter Knowles, who develops computational methods at Cardiff University, UK. 'Further work should answer the question as to whether there is a practical use for this simple theory - can we use it to predict relative chemical reactivity, or to construct a simple force field for intermolecular interactions, using these atomic charges?'

James Mitchell Crow