A Material Difference for DFT

See allHide authors and affiliations

Science  04 May 2007:
Vol. 316, Issue 5825, pp. 662
DOI: 10.1126/science.316.5825.662c

The accuracy of density functional theory (DFT) calculations, which is limited by the approximate treatment of the exchange-correlation (XC) functional, can be tested for small systems by wave-function methods. In systems too large to easily apply such a test, especially those with strong electron correlations, it has often been hoped that the errors in total energies would cancel out when differences were evaluated. An example where this cancellation fails is the low-coverage adsorption of CO on close-packed surfaces of copper and platinum, for which DFT calculations favor the threefold hollow site over the experimentally determined “on top” site (binding to just one metal atom) by at least 0.4 eV. Hu et al. performed high-level quantum chemistry calculations on small metal clusters (copper and silver) to evaluate the XC energy error of DFT. The XC correction (the difference between the energy from DFT and that from the higher method) varies for different methods and continues to change as cluster size increases. However, because of the short-range nature of this error, the differences in the XC correction between different types of sites on the cluster converge to a constant at cluster sizes as small as ~20 atoms. By applying this correction to the DFT results, the authors obtained the correct on-top site preference for CO on copper, as well as the correct result for silver. This approach can also be applied to bulk systems and defects. — PDS

Phys. Rev. Lett. 98, 176103 (2007).

Navigate This Article