PhD defence by Christina König

Principal supervisor

Professor Jörg Ralf Jinschek, DTU

Co-supervisor

Professor Peter Mayr, TUM
Senior researcher Alice Bastos da Silva Fanta, DTU

Examiners

Professor, Dr. tech Dorte Juul Jensen, Chair, DTU Construct
Professor Dr. rer. Nat Marc-Georg Willinger, TUM
Professor Manas Upadhyay, Ecole Polytechnique

Chairperson at defence

Professor Thomas Willum Hansen, DTU

Abstract 

Additive manufacturing (AM) of metals enables the layer-by-layer production of complex components but also imposes highly non-equilibrium conditions. Rapid heating and cooling cycles during deposition generate microstructural inhomogeneities, including irregular grain structures, pores, cracks, residual stresses, and variations in phase composition. These features strongly influence the mechanical performance and reliability of AM parts, yet they are challenging to characterize because fine-scale defects are embedded within large volumes and remain inaccessible to conventional microscopy methods. 

This thesis develops a high-throughput scanning electron microscopy (SEM) framework that combines large-area imaging with sub-micrometer resolution. Automated acquisition, contrast-based defect detection, and machine-learning-driven segmentation enable statistically significant quantification of pores, grain morphology, and other microstructural heterogeneities. Applied to additively manufactured stainless steel, the approach reveals clear correlations between deposition strategy, defect distributions, and grain structure, providing a systematic means to evaluate processing–microstructure relationships. 

To replicate the thermal histories inherent to AM, a dedicated in situ SEM heating platform was established using micro-electromechanical (MEMS) devices. This setup achieves heating rates up to 1000 °C/s with controlled temperature profiles, enabling direct observation of grain growth and phase transformations under process-relevant conditions. Experiments on duplex stainless steel demonstrate how cyclic thermal exposure modifies grain boundaries and alters the ferrite–austenite phase balance. 

Together, these methods address the central challenge of connecting AM processing conditions to microstructural inhomogeneities across both length and time scales. The results advance the quantitative characterization of additively manufactured metals and contribute to a more predictive process–microstructure–property framework for the design of reliable AM components.