Français
English
A solid can be in a state of zero stress at the macroscopic scale, and have non-zero stresses at its local scale. Metallic materials can be affected by these states, particularly if phase transformations are involved in the manufacturing process. The interactions between heat, metallurgy and mechanics are indeed the source of the development of these stresses, as is well known in the fields of welding, casting or additive manufacturing. As these internal stresses are generally present in an undesired and poorly controlled way, they are referred to as residual stresses. They are added to those due to external mechanical loading and can therefore contribute to premature damage to a material: reduced fatigue life, acceleration of stress-dependent physical mechanisms (diffusion of chemical species, phase transformation, oxidation).
Various techniques have therefore been developed to control these residual stress states. On the macroscopic scale of a part, they consist in locally removing an element of material, for example by drilling, and measuring the deformations due to stress relaxation. This measurement can be carried out using Digital Image Correlation (DIC), providing a deformation field and opening up the possibility of characterizing a multiaxial field. In this multiaxial context, the process of moving from measured deformations to residual stresses is based on Finite Element Analysis (FEA) reproducing the material removal operation.
This method can be transposed to the micrometer scale of the part: by integrating a Focus Ion Beam (FIB) probe into a Scanning Electron Microscope (SEM), it is possible to perform a material removal operation on a scale of a few micrometers, while capturing the image of the material as it is being machined. Having previously marked the material with ~10 nm markers, the Digital Image Correlation (DIC) method can be used to track the displacements of the markers and deduce a deformation field. Given the characteristic dimensions (plot, trench, markers), the stresses involved (~100 MPa) and the high stiffnesses, the accuracy required for a usable displacement measurement is of the order of 1 nm.
Ensuring the conditions for sufficiently accurate field measurements is the first challenge of this thesis. The first objective is therefore to develop the SEM-FIB-CIN analysis protocol based on samples with simple, known residual stress conditions: uniaxial stress of approximately known value (XRD method), grain size large enough to guarantee homogeneous mechanical fields. The second objective is to extend the method to a context of heterogeneous (intrinsically multiaxial) microstructure and mechanical fields. To be able to take into account the presence in a close vicinity of a grain other than the one treated, or even of another material phase, it is then necessary to take this vicinity into account in the finite element analyses. This context is that of steels produced by additive manufacturing, or that of an austenitic-ferritic steel. Both are highly heterogeneous, with characteristic grain-to-grain or phase-to-phase lengths of the order of a few μm; both can feature stress states of the order of 300-400 MPa, and for both, detailed knowledge of these states represents a major industrial challenge, given the cutting-edge applications targeted.
Contact : Fabrice Barbe, fabrice.barbe@insa-rouen.fr
