Local Atomic Structure: What can we learn from our neighbors?

Introduction

It is notoriously difficult to study the local environment of atoms directly. Conventional crystallographic approaches to structure determination find the periodic arrangement, or long-range order of atoms. In well ordered crystals the local structure is just the projection of the average periodic structure into the unit cell. However, in materials which are not perfectly ordered, this approximation breaks down. For example, in the semiconductor alloy Ga_xIn_1-xAs, the In-As and Ga-As bonds are different lengths yet in the crystallographic structure the In and Ga randomly occupy the same lattice sites: all of the In-As and Ga-As are the same length which is the weighted average of the real bond lengths. Clearly in this case, if the structure is to be properly understood the local structure must be solved explicitly.

Our Interest

We have been developing the atomic Pair Distribution Function (PDF) technique to do just that. This is a powder diffraction technique which has been used for many years to study glasses and amorphous materials. We are extending its use to look at disordered crystalline materials. What allows us to do that is the development of advanced x-ray and neutron sources which deliver large fluxes of high-energy x-rays and neutrons. We routinely carry out experiments at NSLS at Brookhaven National Laboratory, CHESS at Cornell University, IPNS at Argonne National Laboratory, MLNSC at Los Alamos National Laboratory, and in the future at the Advanced Photon Source at Argonne National Laboratory as part of the Midwest Universities Collaborative Access Team . This allows us to collect data over an extremely wide range of momentum transfer Q. These data are Fourier transformed into real-space coordinates to yield the atomic pair-pair correlation function directly: a real-space map of the environment around the atoms in the solid. In this way we study the structure by getting to know the neighborhood where the atoms reside.

Important recent developments include a full-profile, least-squares structural modelling program which is analogous to the Rietveld refinement of powder diffraction data; and improved data-correction routines which allow us to recover the experimental PDF of a material with extremely high reproducibility and very good accuracy. This work is described in various papers listed in the publication list on this home-page.

Return to Simon Billinge:Home Page