Indirect selective laser sintering of ceramics

Ceramics with intricate geometries are useful in a wide range of technical applications, but current manufacturing techniques limit the geometric complexity. Additive manufacturing (AM) is a viable approach to surmount these limitations and produce complex tailored structures. However, ceramic materials are not easily processed with AM technologies because of their high melting temperature and sensitivity to thermal shock. Because of this, an indirect approach is taken where a transient binder is added to effectively glues the ceramic particles together during shaping. Selective Laser Sintering (SLS) is capable of producing small complex polymer features, but how this capability translates to indirect SLS of ceramics is an open question. This dissertation aims to systematically investigate the limitations of using indirect SLS to produce intricate ceramic geometries. The investigation is first approached from a phenomenological perspective, where a variety of geometries are manufactured using indirect SLS and then compared to similar polymer parts produced with direct SLS. The geometry comparison is paired with a mechanistic study of particle-scale interactions using in-situ microscopy, and an analytical model is developed to describe these observations. Guidelines for the geometries possible with polymer SLS turn out to be a good starting place for the design and manufacture of ceramic geometries using indirect SLS. However, indirect SLS is further limited by the heterogeneity of the blended powder. In-situ microscopy shows that, in mixed polymer/ceramic powders, the binding mechanisms are different at different laser scan speeds and also different from polymer SLS or selective laser melting of metals. A permeation model is used to describe the binding mechanisms specific to indirect SLS and correctly predicted all of the trends observed in experiments. The settings (laser power and scan speed) that produced parts strong enough to be removed from the SLS machine were predicted by the model to produce permeation distances large enough for particle bonding. Conversely, settings which resulted in failed parts during the experiments were predicted by the model to not permeate far enough for particle bonding