Due to its key role in shaping material properties during thermo mechanical treatments of iron alloys, wide attention has been devoted to ferrite-austenite transformation. Its characterization at the mesoscopic scale generally involves monitoring of phases content using X-Ray diffraction or deducing microstructure evolution from post-mortem EBSD mapping. However, complete understanding of transformation mechanisms would involve tracking microstructure evolution in situ. Such kind of direct observation remains challenging owing to the necessity to combine high resolution imaging and adaptability to high temperatures.

In this study, Digital Image Correlation (DIC) has been chosen as a way to compute strain field during allotropic transformation. Images are captured using a high-resolution camera which allows characterizing a large part of the sample with micrometer resolution. A home-made experimental device is designed to maximize image resolution and keep control over thermal solicitation imposed to the sample. Following previous work on DIC at high temperatures [J.S. Lyons et al., Exp. Mech. 36(1996); B.M.B. Grant et al., J. Strain Anal. Eng. Des. 44(2009)], care is taken to avoid loss in image quality. In particular, samples are put in a non-oxidizing atmosphere and the influence of the energy they radiate is reduced thanks to a system of illumination and optical filtration in blue wavelengths domain. Grains are coarsened to observe in-grain phenomena more easily. Grain coarsening in the austenitic domain has proved more efficient than strain annealing to obtain large grain sizes, up to millimeter size. Samples are characterized by Electron Back-Scattered Diffraction (EBSD) before and after the experiments to study microstructure and texture evolutions. They are heated by Joule effect, which allows attaining heating rates around 1000°C/s. Temperature is measured by a 2-color pyrometer. A regulation system is set to submit samples to various heating and cooling conditions and study their subsequent effect on ferrite-austenite transformation. Emphasis is also put on understanding coupling between plasticity mechanisms and transformation. To this end, samples are plastically deformed in uniaxial tension prior to heating in a range varying from 1 to 10%. Results show strong heterogeneities in the polycrystal response. Activation of different slip systems according to the level of solicitation is highlighted.

In parallel, a model written under the small strain format is developed to gain further insight in the understanding of iron behavior. A coupled formulation is built by mixing up thermal and transformation terms into the expression of a thermo mechanical power functional derived from crystal plasticity. In this mechanical model, dislocation glide is regarded as the main mechanism for stress accommodation. An attempt is made to reproduce distinctly the behavior of each phase. Behavior of the BCC phase, which exhibits the strongest temperature dependency is modelled by a statistical formulation inspired from [L. Stainier et al., J. Mech. Phys. Solids. 50(2002).] based on kink and jogs formations probabilities. For the behavior of the FCC phase which is ruled by the displacement of individual dislocations and their interactions with obstacles, a storing and annihilation law is considered. Transformed volume fraction is then introduced as an internal variable and a spherical contribution encompasses transformation strain. In accordance with variational principles [M. Ortiz, L. Stainier. Comp. Meth. In Applied Mech. And Eng. 171(1999)], evolution of variables, including slip increments and fraction of the material locally transformed, are computed through the resolution of a minimization problem. Simulations are carried out using Zorglib finite element software. Geometry and grain orientations are extracted from experimental data. A volume heating source mimics Joule heating and radiation conditions are introduced at the surface of the samples. Related coefficients are adjusted to fit observed thermal behavior. Simulations are shown to be in correspondence with experimental results.