SEM-WDS x STEM-EDS: Improving how we measure Mn partitioning in retained austenite in medium-Mn TRIP steels

The characterization of local phase composition in a microstructure always requires a compromise between the size of the volume analyzed and the size of the microstructure considered. Methods probing small volumes with a small probe size like TEM-EDS that allows reaching high spatial resolution are generally faulted for their statistical representativeness. Conversely, SEM measurements by WDS/EPMA become complicated when the analyzed features become micrometric, even if they allow the analysis of larger fields. In this work, we compared these two methods for the same nanoscale microstructures, for medium Mn duplex steels, annealed at different temperatures, and thus presenting different enrichment by partition of Mn, as predicted by thermodynamics. This comparison work allowed us to develop a method of correction of SEM-WDS data to make them comparable with STEM-EDS data, thanks to a blurring filter based on the statistical distribution of compositions and convolutional simulated images. Thanks to its improved statistical validity, the corrected value obtained by SEM-WDS explains even better than STEM-EDS the tendency observed by HEXRD that thermal martensite is only present at the highest intercritical annealing temperatures in the microstructures. The development of this innovative method has the advantage of being user-independent, more accessible, and statistically more relevant than TEM, and its use can also be extended to other binary and multi-phase alloys for compensating the effect of a poor grain size/ probe size.