Date of Award

Spring 4-2017

Embargo Period


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Materials Science and Engineering


Petrus Chris Pistorius


Additive manufacturing (AM) has experienced remarkable growth in the past decade with applications in both rapid prototyping and rapid manufacturing for functional end-usable parts. As one of the most promising AM processes, selective laser melting (SLM) can be used to fabricate metal products line by line and layer upon layer within a powder bed system. Such process allows the building of parts with customized shapes, which brings higher design flexibility than traditional casting and wrought manufacturing. In this work, AlSi10Mg powder is chosen as the raw material for producing parts by SLM, since aluminum alloys are widely used in automotive and aerospace industries thanks to an excellent combination of low density and competitive mechanical properties. However, there remain multiple drawbacks which limit further applications of aluminum parts produced by SLM: lack of prediction of solidification microstructure, few studies on fatigue properties, and cost and time caused by the limited production rate. All these issues were studied in this work and summarized as follows: Rapid movement of the melt pool (at a speed around 1 m/s) in SLM of metal powder directly implies rapid solidification. In this research, the length scale of the as-built microstructure of parts built with the alloy AlSi10Mg was measured and compared with the well-known relationship between cell size and cooling rate. Cooling rates during solidification were estimated using the Rosenthal equation. It was found that the solidification structure is the expected cellular combination of silicon with α-aluminum. The dependence of the measured cell spacing on the calculated cooling rate follows the well-established relationship for aluminum alloys. The implication is that cell spacing can be manipulated by changing the heat input. Microscopy of polished sections through particles of the metal powder used to build the parts showed that the particles have a dendritic-eutectic structure; the dendrite arm spacings in metal powder particles of different diameters were measured and also agree with literature correlations, showing the expected increase in secondary dendrite arm spacing with increasing particle diameter. It is well-known that the fatigue behavior of cast aluminum alloy parts is largely determined by the internal defects, particularly pores and inclusions, such as oxides. This study shows that such imperfections are also present in AlSi10Mg parts produced by SLM, and serve as sites for failure initiation. The effect of hatch spacing and building orientation on tensile and fatigue properties was tested. Similar defects were found both on polished cross-sections and on fracture surfaces. The results imply that the oxide-driven pores dominate the fatigue resistance of the samples in this work. The larger oxide particles which are associated with the crack initiation likely form by oxidation of metal vapor during part manufacture. Residual porosity in parts produced by SLM mainly results from lack-of-fusion, entrapped gas, pores left in powder, evaporation of elements, and collapse of key-holes. Lack-of-fusion porosity is caused by the the insufficient overlap of melt pools in powder bed fusion and is particularly detrimental to fatigue performance due to the stress concentration at the sharp edges of the pores. The third part of this work deals with predicting lack-of-fusion porosity quantitatively by a geometrically-based model and designing processing parameters for build rate improvement without introducing porosity. The inputs into the simulation are hatch spacing, layer thickness, melt-pool cross-sectional area, and hatch rotation angle. Comparison with several data sets from the literature shows that the simulations correctly predict process conditions at which lack-of-fusion porosity becomes apparent, as well as the rate at which porosity increases with changes in process conditions such as beam speed, layer thickness, and hatch spacing. Relative to the default processing parameters provided by the manufacturer, the build rate can be improved by adjusting hatch spacing and layer thickness, and increasing the platform temperature. The simulations also show that the volume fraction of lack-of-fusion porosity is independent of hatch rotation angle. A unique combination of zero rotation and half hatch spacing as the beam offset between adjacent layers is proposed for build rate optimization.