Determination of Selective Laser Melting Process Parameters for a Novel Material System: With Applications in Thermoelectrics

Abstract

The application of additive manufacturing (AM) to multiple technological industries has become extremely prevalent in recent years. With this move, there has been a need to qualify new materials for first time use within a conventional AM system. One such common system is that of laser powder bed fusion (LPBF). It is also known as Selective Laser Melting (SLM). As it is already extensively used with metals, LPBF is an excellent tool for exploratory research with powdered materials such as semiconductors and semimetals that are traditionally used in electronic devices in bulk crystalline form. An excellent candidate in this subset of materials is thermoelectrics. The work in this thesis focuses on the use of AM to produce thermoelectric devices with the bismuth telluride semimetal alloy, and to use the knowledge of laser-material interactions to develop a framework that can forecast optimal process parameters for this material system, and ultimately, any system that is conducive to LPBF. An EOSINT M280 Direct Metal Laser Sintering machine was used to carry out experiments on the powdered bismuth telluride alloys. These were unfit to be readily used within the machine due to their non-spherical morphology and wide particle distribution. A sieving of the powder, followed by optical and thermal characterization and a consultation of literature, provided the necessary information to create the necessary manufacturing jig and semi-analytical model to explore appropriate laser powers and speeds to use. A custom powder delivery system was developed in order to overcome flowability issues, and to obtain uniform layer height within a build. The delivery system used metallic solder stencil masks of a fixed 76μm thickness, that were able to be stacked with relative ease. The substrates upon which the AM took place were prefabricated printed circuit boards that were designed to have the electrode layout of a conventional thermoelectric module. The developed model used a semi-analytical energy balance to determine laser power/speed pairs that satisfy the thermal requirements of the melt process with some associated losses. It uses an approach to divide the incident beam into solid- and liquid- interacting components, to attempt to deliver power at a rate that would allow for melting to the bottom of the powder bed with minimal evaporative loss. Both p-type and n-type bismuth telluride elements were successfully manufactured on PCB substrates and showed interesting triangular microstructures upon their rapid solidification after laser melting. They also maintained their stoichiometry to the point that their alloy types did not change. The p-type bismuth telluride was tested using the Harman Technique and was determined to have a room temperature ZT of 1.1. For an understanding of the optimal process parameters, the semi-analytical model was compared to two widely different material systems in literature, the well-characterized Ti6Al4V metallic alloy and other attempts at AM of bismuth telluride. It was able to provide a good approximation to the empirically observed process windows for each material that used the metric of maximal relative density.