Simon R. Bare, UOP LLC, 25 E. Algonquin Road, Des Plaines, Il 60017.
e-mail: srbare@uop.com

The recording of XANES data, as opposed to EXAFS data, has many advantages experimentally. XANES features are intense, concentrated in a small energy region and exhibit weak temperature sensitivity. It is possible to extract a huge amount of useful information from the XANES region: chemical information in the form of valence and charge transfer, bond lengths, number and type of nearest neighbors. Most of the XANES work published is qualitative in nature. Use is made, for example, of the relative height of a pre-edge peak, the position of the inflection point, the intensity of the white line, or the presence or absence of some feature. More recently use has been made of principal component analysis; another empirical method. A quantitative description of XANES is desperately needed.

One example of the application of XANES is in heterogeneous catalysis: studying the reduction or oxidation of a metal complex on a high surface area support to form a small cluster of metal atoms. Industrial catalysts typically contain low amounts of the metals, which make EXAFS experiments difficult and very time consuming. Many chemical and structural events happen during the activation process (e.g. reduction of the complex, alloying, cluster size growing, cluster composition changing). A better understanding of these would lead to the development of improved or new catalysts. It is still not fully understood what the cluster size, shape, support interaction etc. have on the XANES of supported metal clusters.

An experimentalist's dream would be to provide as input the XANES spectrum and the elements that are in the sample, and to receive as output the electronic and geometric structure of the material. The XANES of more than one edge, or element, in the sample could be supplied. With the aid of an on-line database and artificial intelligence the power of the computational tools could be increased dramatically. There is real need for a link between data interpretation, modeling and simulation. It ought to be possible to draw a structure, calculate the XANES of that structure, then start to manipulate the structure to see the effect on the XANES.

The focus of theoretical interpretation of experimental data must be on real problems on real systems under real conditions. The theoretical tools must be readily available. Moreover, the tools should be highly automated and be relatively foolproof for the large fraction of users who cannot be expected to be theoreticians or electronic structure experts, yet must be assured of reliable results. There must be a close interaction between theorists and experimentalists.