General. Glasses are materials that have been known for a long time and have already been made by the Assyrians and Egyptians. Today glasses are used in many technological applications such as glass fibres for fast transmission of data or as electrically isolating extremely thin layers in semiconductor systems (e.g. in the processor of a computer), and in most cases they cannot be replaced by other materials. The term "glass" however is not restricted to the classical silicate glasses which are based on silica (SiO2). In general, amorphous (greek: without shape) inorganic solids, without any long range structural order are denoted as glasses. Many different compounds of nitrogen, phosphorous, boron etc. are amorphous, and are hence glasses. Even metals which have been rapidly cooled from the melt and contain chemical additions can be obtained as (metallic) glasses.

While the atomic structure of crystalline solids can be studied excellent with the methods available nowadays, the analysis of the structure of amorphous solids causes enormous problems for study, and hence our knowledge on the structural constitution of such solids is still poor. This was the motivation for installing the Collaborative Research Centre (Sonderforschungsbereich) SFB 408 entitled "inorganic solids without translational symmetry - synthesis, structure and modelling" at the University of Bonn. In this SFB we contribute with two projects (TP). In one of the projects (TP A11), we work on modified oxide silicate glasses, and in the other project (TP B5) we study the structure of glasses applying methods of transmission electron microscopy (TEM). While the project TP A11 was just started at the beginning of 2001, in TP B5 we have successfully installed, developed and applied various methods using electron microscopes for structural analysis.

The pair distribution function (PDF) is a "fingerprint" of a glass. The PDF represents the local atomic density measured at a distance from an arbitrary atom. Therefore maxima in the PDF show the favoured distances between pairs of atoms (see figure). The PDF is gained by means of scattering experiments, where usually X-rays, synchrotron radiation (= high energy X-rays) and neutrons are used. Electron scattering has only been applied recently. This method combines the advantage of the wide-spread availability of X-rays with the high angular resolution achievable with synchrotron and neutron radiation. We have developed the technique of energy filtered electron scattering which has turned out to be an important instrument for the SFB 408. Now the technique is applied more or less routinely on our trans-mission electron microscope with field emission source (CM300FEG) in combination with the imaging electron energy filter. Many oxide glasses such as barium silicate, barium germanate and, in particular, the amorphous nitride ceramics in the system Si-B-N-C could be widely characterised with the PDF from electron scattering data.

Homogeneity and phase separation. The ordering and the distribution of the different chemical species (kinds of atoms) in a crystalline solid is enforced by its three dimensional periodically ordered structure. Such an ordering principle does not exist in amorphous solids and, hence, the chemical composition of the material may vary locally - intentionally or by accident. This phenomenon is called phase separation and its origin may either lie in the thermodynamics of the system or in the preparation of the material. However, the distribution of the chemical elements in a glass in turn determines the structure at a scale larger than approx.>= 1 nm, which is somewhat larger than the size of simple molecules. In glasses an ordering at this scale is denoted as medium range order (MRO). Only few imaging techniques exist which are able to detect a phase separation or inhomogeneity on the nanometre scale. We have developed electron spectroscopic imaging as a method of energy filtering TEM (EFTEM) to image the distribution of single chemical elements at high resolution (better than 1 nm). In the amorphous ceramics of the system Si-B-N-C it was possible to prove chemical homogeneity on the scale above 1 - 2 nm. One of the remaining problems we currently work on is the influence of the projection effect in electron microscopy: despite of a possibly existing phase separation, the distribution may appear homogeneous in elemental maps.
Methodic studies on the detection of small crystallites, i.e. of structural inhomogeneities of a size >= 2 nm in an amorphous matrix, were performed successfully. The method is still considered for the characterisation of amorphous solids to detect eventually existing crystalline nuclei.