Mechanical properties of metallic components produced by casting processes are strongly dependent on their microstructure including the presence of defects such as pores and cracks formed during solidification. This microstructure is usually characterised on 2D sections by optical or scanning electron microscopy after careful preparation of the surface. This characterisation, however, does not give information about the manner in which the microstructure has been generated and the origin of the defects. In situ observations are now possible thanks to the development of ultrafast X-ray microtomography carried out at high resolution on both the ID15 and ID19 beamlines. The further advantage of tomography is that the observations are in 3D which is very important to study the development of dendrites and the formation of defects during solidification of alloys.

The solidification experiments were carried out at beamline ID19 with an Al-10wt%Cu alloy using cylindrical specimens of 1.4 mm in diameter and about 3 mm in height. The specimen was glued on the top of an alumina rod placed on the rotating stage. It was heated until completely molten and then slowly cooled at 3°C/min while being supported by its own oxide skin. The solidification experiment was carried out in a furnace made of two MoSi2 heaters enclosed inside a cubic-shaped chamber. The furnace has a hole at the bottom through which the specimen is inserted and there are two windows on the sides to allow the passage of X-rays. The scan time to take 450 projections over a 180° rotation of the specimen was 22 s. Radiographs were recorded using the FReLoN 14-bit dynamic CCD camera.

This experiment allowed in situ observations of the continuous growth of the dendritic structure of the alloy as shown in Figure 132, both in cross-section and in 3D. The figure clearly shows the various coarsening mechanisms which operate during solidification, namely dissolution of small dendrite arms and filling of the gap between two adjacent arms. However, since the solid fraction increases during solidification, these mechanisms are due to both solidification and coarsening induced by the reduction of the solid-liquid interface area.

Fig. 132: 2D and 3D observations of the evolution of a dendrite with solidification time; the temperature and the volume fraction of solid are indicated within brackets.

To isolate the coarsening mechanisms, similar experiments were also carried out while maintaining the alloy isothermally in the semi-solid region and thus at constant solid fraction. For these experiments, the specimen was heated in a specially designed electrical resistance furnace at 6°C/min until 570°C and held isothermally at this temperature. The scan time was 30 s and a scan was taken every 65 s to characterise the evolution of microstructure during holding. At this temperature, the solid volume fraction was 72%.

Figure 133 shows details of the microstructural evolution at the scale of the dendrites. The observed coarsening mechanisms are somewhat different from those proposed in the literature so that the published models do not properly account for the observed kinetics of dendrite evolution. Improved models have thus been proposed based on our observations.

Fig. 133: Sequence of 2D (left) and 3D (right) images extracted from the volume of the specimen held for various times at 570°C showing the different coarsening mechanisms occurring on the scale of the dendrite arms (dark grey on the 2D images): a) progressive small dendrite arm melting; b) progressive interdendritic groove advancement; c) progressive interdendritic groove advancement and joining of the tips of the dendrite arms, leading to the formation of entrapped liquid.

These experiments clearly demonstrate that in situ tomography is a very powerful technique for characterising the microstructural evolution of solid-liquid mixtures both during solidification and isothermal holding. The physical mechanisms that operate at the scale of the individual dendrites can be directly observed, thus allowing more realistic models to be proposed. However, the liquid and the solid phases must exhibit sufficiently different absorption contrast to be clearly distinguished and to allow for quantitative analysis. There are still some other limitations, particularly in terms of kinetics of the phenomena, but the continuous improvement in the equipment and techniques will make it possible to overcome these limitations in the future.


Principal publication and authors

N. Limodin (a,b), L. Salvo (a), E. Boller (c), M. Suéry (a), M. Felberbaum (d), S. Gailliégue (e) and K. Madi (e), Acta Materialia 57, 2300–2310 (2009); S. Terzi (a,f), L. Salvo (a), M. Suéry (a), A.K. Dahle (f) and E. Boller (c), Acta Materialia 58, 20-30 (2010).
(a) SIMaP, Université de Grenoble, Saint-Martin d’Hères (France)
(b) MATEIS, INSA Lyon, Villeurbanne (France)
(c) ESRF
(d) STI-LMX, EPF Lausanne (Switzerland)
(e) Centre des Matériaux, ParisTech, Evry (France)
(f) ARC CoE for Design in Light Metals, The University of Queensland, Brisbane, (Australia)