So far we have considered the case of a perfect single crystal. With the help of the same principle it should be possible to visualize individual grains in the bulk of polycrystalline material. In conventional absorption tomography one usually neglects deviations from the ideal transmission behaviour due to occasional diffraction effects when individual grains happen to fulfill Bragg’s law for certain angular positions. These grains appear less intense on the detector, the ’miss- ing’intensity being transferred to the diffracted beam. Due to the fact that the reconstruction is based on a large number of projections, these occasional diffraction effects do not lead to noticeable image degradation.
Another situation arises, if one intentionally aligns the polycrystal in such way, that the diffraction vector of a strong reflection in one of the crystal grains is parallel to the rotation axis. In this case the grain will diffract for all angular settings during the tomographic scan and the volume occupied by this grain will receive a higher attenuation coefficient, as the grain appears more absorbing in the projections.
This special configuration can consequently be used to determine the position and shape of an individual grain in the bulk of polycrystalline material in a non-destructive way. Note that this information is inaccessible in conventional absorption mode, where the individual grains cannot be distinguished. This technique might be of great interest for the time-resolved, three- dimensional analysis of grain-growth in the bulk of polycrystalline material.
3.2.7 Conclusion
We propose a new three-dimensional crystal characterization technique, based on the combina- tion of X-ray diffraction topography and computed microtomography. The approach is appli- cable to high quality single-crystals, provided the dominant contrast mechanism is the direct image one. Integrated, monochromatic beam diffraction topographs may then be regarded as projections of the local Bragg reflectivity. Rotating the sample around the diffraction vector
3.2. TOPO-TOMOGRAPHY 61
will not significantly vary the diffraction conditions and allows to acquire a complete set of projections (Radon transform) which serves as input for the tomographic reconstruction. The reconstruction yields an approximation of the three-dimensional distribution of the local Bragg reflectivity inside the crystal and allows to analyse the orientation and position of individual dislocations inside the crystal.
A possible extension to the case of polycrystalline material is suggested. It is expected, that the same principle allows to determine the shape and position of individual grains in the bulk of a polycrystal.
Part II
Applications
Chapter 4
Grain boundary wetting
This chapter is devoted to the application of Synchrotron Radiation micro-imaging to a specific phenomenon in metallurgy: the wetting of grain boundaries by a liquid metal. After a short introduction and bibliographic survey of previous work we will present our experimental results obtained in the system aluminium - liquid gallium. In-situ measurements of the penetration process in bulk specimens allowed for the first time to demonstrate the direct link between grain boundary wetting and grain separation in this system. Combining different X-ray charac- terization techniques, we could also establish a method to analyse the wetting process in bulk polycrystalline material. We will conclude this chapter by the tentative interpretation of the results in terms of a fracture mechanism.
4.1
Introduction
When a polycrystalline solid is exposed to a liquid phase, rapid penetration (10−3−100µm/s) of the liquid phase along the grain boundaries takes place in a variety of metallic and ceramic systems [CG92]. This process leads to the formation of liquid films with a typical width in the order of 0.1 - 10 µm. The presence of such macroscopic, intergranular liquid films can be used to assist certain industrial manufacturing processes like high strain rate superplastic forming and sintering or chip fragmentation in rapid machining. On the other hand, the contact with a liquid metal may also entail severe degradation of the mechanical strength of the material. The latter phenomenon, known as liquid metal embrittlement (LME) [NO79, FJ97, JPB99, Gli00] is of relevance when structural metals are exposed to liquid metal environments under the simultaneous action of external or internal stresses. This is for example the case in common industrial processes like galvanising, soldering and welding (see e.g. [FCJ94]). The subject has gained renewed interest in recent years due to the active international efforts to develop ’safe’ (undercritical), accelerator driven nuclear reactors where liquid metals serve as targets and cooling fluids.
Despite numerous efforts, the detailed mechanism(s) which lead to the formation of intergran- ular liquid films are still far from being well understood. A variety of models have been proposed in recent years [Fra94, BKA95, Rab98, GN99], however none of them can fully account for the different aspects of the penetration process. This might indicate, that the working mechanism is not the same in different systems. The situation is further complicated by the fact, that the problem is not well defined from an experimental point of view. The results depend not only on parameters like temperature, composition of the involved materials and the detailed nature of the grain boundary but also on parameters like:
• impurity segregation from the melt at the solid liquid interface
• total amount of liquid metal
• geometry of the sample (e.g. thin foils in TEM versus bulk specimens in SEM)
• thermo-mechanical treatment of the base material (residual stress)
• surrounding atmosphere (presence of oxygen, water,. . . )
• sample preparation and observation technique (in-situ versus ex-situ, volumetric changes during solidification)
Differences in these conditions complicate and may even prevent the comparison and analysis of experimental results. The analysis of the experimental data is further complicated by the fact, that the process takes place in the bulk of the sample - whereas conventional observa- tion techniques are restricted to the sample surface (or polished sections) and thin foils. As a consequence, there is only limited and rather scattered data available, which might serve to corroborate or reject the different models which have been proposed.