In semiconductor inspection equipment, a chuck used to hold a wafer is equipped with a cooling or heating system for temperature uniformity across the surface of the wafer. Surface temperature uniformity is important for increasing semiconductor inspection speed. Triply periodic minimal surfaces (TPMSs) are proposed to enhance temperature uniformity. TPMSs are a topic of increasing research in the field of additive manufacturing and are a type of metamaterial inspired by nature. TPMSs are periodic surfaces with no intersections. Their continuous curve offers self-support during the additive manufacturing process. This structure enables the division of a single space into two domains. As a result, the heat transfer area per unit volume is larger than that of general lattice structures, leading to a superior heat transfer performance. This paper proposes a new structure called a double-wall TPMS. The process of creating a double-walled TPMS by adjusting the thickness of the sheet TPMS was investigated, and its thermal performance was studied. Finally, a double-wall TPMS was applied to the chuck. The optimal designs for the diamond and gyroid structures exhibited a difference in surface temperature uniformity of 0.23 °C and 0.66 °C, respectively. Accordingly, the models optimized with the double-wall TPMS are proposed.

Recent advances in additive manufacturing have provided more freedom in the design of metal parts; hence, the prototyping of fluid machines featuring extremely complex geometries has been investigated extensively. The fabrication of fluid machines via additive manufacturing requires significant attention to part stability; however, studies that predict regions with a high risk of collapse are few. Therefore, a novel algorithm that can detect collapse regions precisely is proposed herein. The algorithm reflects the support span over the faceted surface via propagation and invalidates overestimated collapse regions based on the overhang angle. A heat exchanger model with an extremely complex internal space is adopted to validate the algorithm. Three samples from the model are extracted and their prototypes are fabricated via laser powder bed fusion. The results yielded by the fabricated samples and algorithm with respect to the sample domain are compared. Regions of visible collapse identified on the surface of the fabricated samples are predicted precisely by the algorithm. Thus, the supporting span reflected by the algorithm provides an extremely precise prediction of collapse.