In the 1988 edition of Nature's 'News and Views', J Maddox wrote that 'one of
the continuing scandals in physical sciences is that it remains in general
impossible to predict the structure of even the simplest crystallographic solids
from knowledge of their chemical composition' (Maddox 1988 Nature335 7).
There is, however, the possibility of making some progress in this direction by
combining two fundamental areas of physics: quantum mechanics and statistical
physics. The starting point is an electronic structure theory density functional
theory (DFT) (Hohenberg and Kohn 1964 Phys. Rev.136 864B, Kohn and Sham
1965 Phys. Rev.140 1133A) which independently establishes the range
(first neighbours, second neighbours, etc), type (pairs, three body, four
body, etc) and chemical character (charge transfer, atomic site effects,
etc) of the interaction energies. All these can be determined from cluster
expansions (CE) (Sanchez et al 1984 Physica A 128 334) which give access to
both huge parameter spaces (e.g. for ground-state searches) and systems
containing more than a million atoms (e.g. for microstructure studies). It
will be shown that, together with Monte Carlo simulations, CE open the
possibility of quantitatively studying alloy properties which possess a
delicate temperature dependence, such as short-range-order effects, mixing
enthalpies or dynamic processes like the ageing of microstructures. This
method is extended to alloy surfaces in order to investigate geometric
relaxations as well as surface segregation, i.e. the enrichment of one component
in the near-surface region. To establish a complementary, experimental
view of the geometrical structure and chemical composition of surfaces,
experimental low energy electron diffraction spectra are analysed by the use of a
multiple-scattering theory (Pendry 1974 Low Energy Electron Diffraction
(London: Academic), Van Hove and Tong 1979 Surface Crystallography by LEED
(Berlin: Springer)) providing a test of our DFT predicted surface properties.