ADVANCED MATERIALS & PROCESSES | SEPTEMBER 2024 1 7 achieve full density and mechanical attributes. Additionally, to fully utilize the potential of ceramic AM in industry, it is crucial to fine-tune the process parameters for various materials over the span of several applications. The integration of machine learning (ML) and artificial intelligence (AI) into ceramic AM procedures is expected to play a critical role in overcoming these challenges. For example, the application of ML techniques (Fig. 1) such as statistical learning theory (SLT), support vector machines (SVMs), Bayesian modeling, neural networks, and reinforcement learning (RL) is revolutionizing manufacturing processes. These technologies are at the forefront of predictive maintenance, quality control, classification, regression analysis, and process optimization, heralding a new era of manufacturing efficiency and innovation[1]. Presently, ML and AI are enhancing efficiency, quality, and performance across several domains. contemporary product development and manufacturing strategies. EMERGENCE OF CERAMIC AM TECHNIQUES Among all the AM techniques, ceramic additive manufacturing has demonstrated the potential to overcome unique challenges posed by ceramics in traditional manufacturing, such as high melting points, brittleness, and hardness. Ceramics are renowned for their resilience under extreme conditions and find extensive application across industries including biomedical, aerospace, automobile, refractory, chemical reactors, and electrical components. However, AM harnesses these properties to fabricate intricate objects layer by layer, showcasing ceramics’ versatility in various applications. This not only underscores ceramics’ critical role in advanced manufacturing but also navigates their complexities to deliver groundbreaking solutions. Currently, a few techniques recognized by ASTM International are used for 3D printing ceramics. These include: • Binder jet 3D printing involves applying a liquid bonding agent onto a bed of ceramic powder layer by layer, cured by UV light, then followed by debinding and sintering to obtain a final component. It is suitable for larger parts, but can result in high porosity. • Stereolithography or vat photo- polymerization uses a UV laser source to cure photosensitive resin with ceramic powder suspension, layer by layer. It is ideal for producing complex geometries with fine details. • Inkjet printing is similar to binder jetting, however the major difference lies in the liquid binder with ceramic suspensions referred as ink and this ink is cured into layers using UV rays to form a component. This process offers a high precision component and is suitable for complex part building while taking longer times. • Fused filament fabrication (FFF) uses a filament loaded with ceramic material to build a structure layer by layer using heat as fusing method. This is the most affordable and fastest building technique among all the AM methods, however, FFF offers less precision compared to other options. ROLE OF MACHINE LEARNING AND ARTIFICIAL INTELLIGENCE Although ceramic AM stands at the cutting edge of manufacturing innovation, providing a sustainable and efficient means to fabricate ceramic components with unparalleled precision, its evolution is ongoing. Given that ceramic 3D printing encompasses several phases—design, printing, post- processing, and quality validation— the final output’s integrity often encounters obstacles due to the physical, mechanical, and thermal characteristics intrinsic to the materials used. These challenges necessitate exacting control throughout the design, debinding, and sintering processes to (a) (c) (b) Fig. 2 — Potential scenarios for AI application in additive manufacturing: (a) A 3D surface plot of a SiO2 3D-printed coupon. AI can analyze this data for defect identification and process optimization. (b) A PCA plot of load data versus time of 3D-printed ceramic coupons, demonstrating PCA’s use in data analysis, pattern recognition, and material classification based on performance over time. (c) Images of a lattice structure: a CAD design (left) and a printed coupon using vat photopolymerization (right). AI can compare the CAD design with the printed object to ensure accuracy, detect defects, optimize geometric design and process parameters, and control quality.
RkJQdWJsaXNoZXIy MjMzNTA5MA==