Soudage à l'arc de l'acier à l'aluminium : Prédiction de l'épaisseur de la couche de composé intermétallique

Joining of aluminum to steel is an interesting method to work towards the reduction of car weight. Among joining techniques, gas tungsten arc welding, also known as TIG (tungsten inert gas) welding, in particular appears as a desirable technique. However Fe and Al react together to form a brittle intermetallic compound layer (IMC) which thickness directly impacts the strength of the joint, hindering their viability at the industrial scale. In this work, we propose to develop a robust model to predict the formation of the IMC for various welding conditions. Before tackling simulation itself, we design a novel method to thoroughly analyze the morphology of the IMC at both the microscopic and macroscopic scales, quantifying its thickness, its interfaces with the base metals and its discontinuities through several descriptors. The evolution of morphology is studied with respect to arc current and welding speed separately, and with respect to welding energy which combines both parameters. Arc current turns out to have a significantly greater influence on the IMC compared to welding speed, which explains the manifest inconsistency of welding energy as a parameter for quantification. Then, we develop a finite element thermal model to evaluate the temperature histories in the weld, calibrated with surface temperature measurements on the real setup. This model is chained to a volume diffusion model and the simulated IMC thicknesses are compared with cross-sections of welded samples. The model turns out to be robust over a wide range of arc currents and welding speeds and gives relevant estimations when changing the backing plate material from steel to pure copper. Further investigations reveal that the conductivity of the Cu backing plate has a negligible influence compared to the increased thermal conductance with the Fe working plate. Finally, the computed transient formation regime of the IMC highlights three stages of growth, delimited by the extrema of temperature variations. Finally, to challenge the choice of computing IMC growth with a pure diffusion model, the formation mechanism of the IMC is observed in more details by studying extit{in situ} the reaction of liquid Al with solid Fe by X-ray tomography and X-ray diffraction with synchrotron radiation. This experiment will reveal different relative amounts of the possible IMC phases depending on both the temperature and the morphology of the samples (bulks or powders). It will allow to quantify the evolution of IMC thickness and morphology with time. The latter point confirms a strong dissolution of the layer by liquid Al while it simultaneously grows into solid Fe, but only at long time scales, reaffirming the relevance of a pure diffusion model in quick transient process such as gas tungsten arc welding.