This paper aims to demonstrate a novel technique enabling the accurate visualization and fast mapping of deformation-induced α′-martensite produced during cyclic straining of a metastable austenitic stainless steel, refined by reversion annealing to different grain sizes. The technique is based on energy and angular separation of the signal electrons in a scanning electron microscope (SEM). Collection of the inelastic backscattered electrons emitted under high take-off angles from a sample surface results in the acquisition of micrographs with high sensitivity to structural defects, such as dislocations, grain boundaries, and other imperfections. The areas with a high density of lattice imperfections reduce the penetration depth of the primary electrons, and simultaneously affect the signal electrons leaving the specimen. This results in an increase in the inelastic backscattered electrons yielded from the vicinity of α′-martensite, and a bright halo surrounds this phase. The α′-martensite phase can thus be separated from the austenitic matrix in SEM micrographs. In this work, we propose a deep learning method for a precise α′-martensite mapping within a large area. Various deep learning-based methods have been tested, and the best result measured by both Dice loss and IoU scores has been achieved using the U-Net architecture extended by the ResNet encoder.

Despite its predominantly empirical nature, metallographic etching remains an important sample preparation method that enables microstructural characterization. The poor repeatability is due to the ignorance of etching mechanisms and the human factor, which is mainly associated with the conventional approach. This approach is most commonly practiced by manual immersion of the sample in the etchant. The solution should be gently stirred so that reaction products do not settle on the surface and uneven etching does not occur. Often, stirring is provided by the sample itself or by a magnetic bar. In terms of spatial orientation, etching with the polished surface facing upwards allows the surface to be observed during the process. Some authors recommend etching with the polished surface downwards to minimize the deposition of reaction products. In some cases, a vertical orientation of the sample is recommended to reduce pitting. In this paper, the effect of the etchant stirring method on the surface quality of a metallographic structural steel specimen will be presented. In addition to different etchant stirring methods, different spatial orientations of the sample will also be investigated. To ensure better repeatability and accurate etching time, the samples will be etched using our automated apparatus. The resulting surface of the samples will be analysed using light optical microscopy (LOM) and confocal laser scanning microscopy (CLSM). The development of more reproducible etching methods is essential to obtain accurate microstructure information without artifacts and for advanced quantitative image analysis using deep learning.

The success of machine learning (ML) models in object or pattern recognition naturally leads to ML being employed in the classification of the microstructure of steel surfaces. Light optical microscopy (LOM) is the traditional imaging process in this field. However, the increasing use of ML to extract or relate more aspects of the aforementioned materials and the limitations of LOM motivated us to provide an improvement to the established image acquisition process. In essence, we perform style transfer from LOM to scanning electron microscopy (SEM) combined with “intelligent” upscaling. This is achieved by employing an ML model trained on a multimodal dataset to generate an SEM-like image from the corresponding LOM image. This transformation, in our opinion, which is corroborated by a detailed analysis of the source, target and prediction, successfully pushes the limits of LOM in the case of steel surfaces. The expected consequence is the improvement of the precise characterization of advanced multiphase steels’ structure based on these transformed LOM images.

Chemical etching is an integral part of metallographic sample preparation. Maintaining precise etch times can be difficult and therefore repeatability is limited. The aim of this work is to improve the repeatability of sample preparation using robotization. Prior to etching, metallographic samples of S355J2 (1.0577) structural steel were finely mechanically polished. For verification, 15 specimens were prepared using an in-house designed automated etching machine with a built-in 5-axis robotic arm and 15 specimens prepared manually by an expert metallographer. The samples were etched with Kourbatoff no. 4 reagent for 8 seconds in a beaker placed in an ultrasonic cleaner at 80 kHz. The samples were then cleaned in 7 beakers of cleaning fluid also placed in the ultrasonic cleaner. The robotic etching and cleaning process was optimized and the quality of the resulting surface is at least as good as that of the samples prepared by an expert metallographer. The surfaces were compared using a light optical microscope (LOM) and a confocal laser scanning microscope (CLSM). The repeatability of the preparation process is a key aspect for obtaining a large dataset of steel microphotographs for training a deep neural network that will be used in future research.