Ceramic Additive Manufacturing Cross-Process Parameter Refinement

<p dir="ltr">With high operating temperatures, hardness, and oxidation resistance, ceramics offer superior performance to metals and polymers in extreme environments. Low-density ceramics, often oxides, carbides, nitrides, and borides, are ideal for aerospace applications, while more dense ceramics and ceramic-metallic composites offer exceptional abrasion resistances for machining and tooling. These applications benefit from unique geometries with complex features, but the same properties that make ceramics desirable create substantial challenges for fabricating complicated parts. While additive manufacturing (AM) has been widely adopted for metals and polymers due to the geometric flexibility, the difficulty of fusing ceramic layers has hindered its broader adoption in ceramics. Adapting powder bed machines traditionally used for metal AM to print ceramics leverages preexisting infrastructure and knowledge to fabricate with new feedstocks. Successfully implementing such an approach, however, requires addressing the limitations of both the printing methodology and the material. Ceramics and extreme materials demand narrow operating windows, meaning either the parameters require refinement to accommodate the material or making alterations to the feedstock to suit the process. Achieving this balance requires an understanding of process-induced anomalies, build failures, and the broader impact of processing conditions on material behavior.</p><p dir="ltr">Binder jet additive manufacturing (BJT) offers a material-agnostic approach to fabricating monolithic ceramics by forming green parts, but identifying and quantifying anomalous spreading events is essential for selecting appropriate parameters to ensure build quality. The development of automated spreading anomaly detection addressed this through image collection, processing, and semantic segmentation. Five-layer deep U-Net neural network models were trained on single build datasets with data augmentation to accelerate implementation. Segmenting alumina and 316 stainless steel highlighted challenges with analyzing optically lighter ceramic powders, where surface features were less distinct. This resulted in a lower test accuracy of 92.3% for the alumina compared to 95.4% for stainless steel, reflecting the increased difficulty in detecting subtle spreading defects. Despite this, the segmentation successfully captured trends in part geometries and printing behaviors. Alterations to recoater roller speeds successfully removed pushing anomalies from the build and captured by the segmentation, demonstrating the effective use of process parameter correction with anomaly identification. </p><p dir="ltr">Anomaly identification informs the selection of successful BJT spreading parameters, but the size of necessary feedstock particle sizes (15 μm to 100 μm) requires modifications to achieve higher final part densities. BJT feedstock requires coarse powders for effective spreading, but fine particles are preferable for sintering, complicating post-processing high-temperature ceramics like SiC. To address this, spray-dried granules composed of submicron SiC were evaluated as an alternative feedstock, offering improved flowability while retaining the benefits of smaller particles. Micro-CT imaging revealed inherent internal porosity within the granules, necessitating higher binder saturation levels to achieve sufficient green strength. Process monitoring guided the selection of parameters influencing powder deposition. Additionally, minimal interaction between the binder and a resin in the powder allowed for unimpeded depowdering. The packing of the submicron particles, however, compounded with the packing of the granules produced porous post-processed parts. While additional post-processing steps were necessary to further increase density, exploring the printing processing parameters enabled the successful fabrication of SiC parts, demonstrating the feasibility of this novel feedstock in BJT. </p><p dir="ltr">In AM processes that encounter challenges that hinder ceramic part fabrication, such as residual stresses in laser powder bed fusion (L-PBF), modifications to the feedstock can improve manufacturability. Cemented carbides, formed by suspending carbide particles in a metal binder, offer a balance of strength and fracture toughness between ceramics and metals, improving processability. Macroscopic failure modes, including plate delamination, part deflection, and recoater blade collisions, must be addressed to achieve successful builds. To assess the impact of process parameter selection on residual stress development, thermomechanical modeling of WC-Ni parts provided trends to guide build designs. Reducing laser energy density generally lowered stresses by decreasing thermal gradients, with changes to laser power exhibiting the greatest effect. Additionally, residual heat from previous layers further reduced stresses with additional preheating, highlighting reducing interlayer time as an additional strategy. These findings enabled the successful fabrication of complex geometries, including drill bits with internal cooling channels, demonstrating the viability of parameter-driven improvements. While the resulting builds achieved comparable hardness values to previous builds, challenges such as plate delamination and edge elevation from scan strategies introduced limitations related to part size and recoater interactions. Ultimately, thermomechanical modeling proved to be a crucial tool in refining process parameters, enabling the fabrication of previously unobtainable cemented carbide parts. </p><p dir="ltr">Through automated anomaly identification, feedstock adaptations, and thermomechanical modeling, this work demonstrates strategies for improving ceramic AM outcomes. The development of process monitoring via semantic segmentation in BJT provided a foundation for understanding spreading behavior and beneficial adjustments to processing parameters. The successful fabrication of SiC spray-dried powder highlighted the importance of adapting to new feedstocks to balance printability with part post-processability. Additionally, the thermomechanical modeling demonstrated how existing computational tools can be adapted to identify behavioral trends with the L-PBF of extreme materials. Together, these studies illustrate the necessity of refining processing parameters to accommodate ceramics in powder bed AM, offering strategies for approaching the fabrication of novel parts and materials for extreme environments.</p>