IN SITU CHARACTERIZATION OF INTERFACE-MICROSTRUCTURE DYNAMICS IN 3D- DIRECTIONAL SOLIDIFICATION OF MODEL TRANSPARENT ALLOYS Rohit Trivedi 1 , Nathalie Bergeon 2 , Bernard Billia 2 , Blas Echebarria 3 , Alain Karma 3 , Shan Liu 1 , Nathalie Mangelinck 2 , Cédric Weiss 2 1) Department of Materials Science and Engineering, Iowa State University, Ames, IA 50011-3020, USA 2) L2MP, Université d’Aix-Marseille III, Faculté Saint-Jérôme, Case 142, 13397 Marseille Cedex 20, France 3) Department of Physics, Northeastern University, Boston, Massachusetts 02115, USA Abstract : Microstructure plays a central role in determining properties of materials so that the fundamental understanding of the physics of microstructure selection is critical in the design of materials. Under terrestrial conditions fluid flow effects are dominant in bulk samples which preclude precise characterization of fundamental physics of microstructure selection. Experiments in thin samples, carried out to obtain diffusive growth, give microstructures that are neither 2D nor 3D. Rigorous theoretical models, using the phase-field method, have shown that the fundamental physics of pattern selection in 2D and 3D is significantly different. A benchmark experimental study is required in bulk samples under low gravity conditions. Also, the selection of microstructure occurs during the dynamical growth process so that in situ observations of spatio-temporal evolution of the interface shapes are required. Microgravity experiments on ISS are thus planned in a model transparent system by using a new Directional Solidification Insert (DSI), designed for use in the DECLIC facility of CNES and to be adapted to also fit ESA experiments. The critical aspects of hardware design, the key fundamental issues identified through 1g-experiments, the proposed experimental study on ISS, and the results of rigorous theoretical modeling are presented. INTRODUCTION The topic of pattern formation is important in several fields of sciences, and the process of pattern selection is quite complex since it occurs in a highly nonlinear growth regime [1]. During the directional solidification of alloys the interface between the solid and the liquid exhibits complex patterns that are analogous to patterns that form in other scientific fields such as combustion, fluid dynamics, geology and biology [2]. Self-similarity of many of these patterns has been shown for two-dimensional patterns. Three-dimensional patterns in many fields are either not stable, or are difficult to study experimentally in a quantitative manner. In contrast, solidification patterns in three dimensions can be accurately studied through directional solidification experiments, and the understanding of the fundamental physics of different solidification patterns will provide a general theoretical framework for a broader problem of pattern formation in nature. The study of solidification pattern formation is also very important in the design and processing of new materials [3]. The interface patterns formed by solidification largely govern mechanical, optical, magnetic and superconductivity properties of materials, so that materials and processing conditions can be designed to obtain specific patterns which give optimum properties and better reliability of the finished product. For example, dendritic patterns are synonymous with most casting and welding microstructures, and the dynamics of pattern evolution plays a critical role in the selection of metastable phases and novel microstructures in rapid solidification processing technologies. From the technological viewpoint, the importance of cellular and dendritic microstructures is quite clear since these structures are exhibited in all important commercial solidification techniques, such as continuous casting and laser welding. The properties of materials, and the reliability of materials, processed by these techniques, are governed by the microstructural characteristics of cells and dendrites. Directional solidification is a powerful technique to study pattern formation since the input parameters can be accurately controlled and the response of the interface can be quantitatively examined through the imaging of the spatio-temporal evolution of the interface patterns [4]. One of the key problems in