Characterization of TiC Reinforced Nickel Matrix Composite using Deep learning assisted X-ray Microscopy

Metal Matrix Composites (MMCs) have gained prominence in materials science and engineering due to their exceptional properties and diverse applications. MMCs represent a class of materials with remarkable mechanical, thermal, and electrical properties. These composites consist of a metallic matrix reinforced with ceramic or other non-metallic phases. In this study, we explore the comprehensive characterization of Ni-Ti-C-based MMCs fabricated by laser engineered net shaping (LENS) process using X-ray microscopy (XRM). Ni-Ti-C-based composites were synthesized by selectively melting nickel and titanium carbide powders layer by layer. The resulting three-phase Ni-TiC-C composites demonstrate outstanding tribological properties, including an exceptionally low coefficient of friction while maintaining relatively high hardness. XRM, with its high-resolution and non-destructive capabilities, provides valuable insights into MMCs. By analyzing the microstructure at the sub-micron level, XRM enables the understanding of the spatial distribution and alignment of reinforcing phases within the metal matrix. Notably, certain phases within the titanium carbide (TiC) domains push the limits of X-ray microscopy detection, relying on X-ray-to-visible-light conversion via scintillators coupled with optical magnification objectives. To overcome previous limitations, we employ a unique combination of specially designed high-resolution objectives and deep-learning-based 3D reconstruction. This approach yields sub-micron resolution results for dense and large MMC samples, which were previously impractical due to signal-to-noise constraints and the operating limits of X-ray scintillators. The findings using XRM significantly enhance the understanding of the mechanical and thermal behaviors of MMCs. These insights contribute to the understanding of MMC behavior and provide valuable insights into the relationship between microstructure and properties.