Content area
Abstract
In laser powder bed fusion (LPBF) additive manufacturing metallic parts are fabricated layer-by-layer. Each layer is formed by a series of tracks where the laser interacts with the metal powder feedstock to form a melt pool. The laser and scanning parameters govern the dynamics of these tracks. Due to rapid melting and solidification, the material track, and therefore the formed layer, will retain a morphology indicative of the melt pool dynamics. By linking knowledge of melt pool dynamics, the input parameters, and resultant surface morphology, the surface can become a tool for determining process health. Furthermore, the same dynamics that form certain surface features have also been linked to other defects such as porosity. This work aims to correlate surface condition to internal defects and microstructure for the purposes of using the surface as a means of gauging broader part health.
This work fabricated samples across three materials to determine which conclusions were material dependent and which were material agnostic. Three commonly used and well-studied alloys were selected for the majority of this work: IN 718, CoCrMo, and 17-4 PH stainless steel. Each material was fabricated using the same various combinations of laser power, scanning speed, and hatch spacing. The design of the experiment was completed based on fabricating parts with different volumetric energy density (VED) values. The VED value is calculated based on a commonly used combinational formula of major LPBF process parameters. It was determined to be a flawed metric, and a later section of this work successfully explores modifications to the VED formula to address several issues.
Only the top surface of the samples was characterized as that is the surface data available during fabrication and it also provides the post-mortem melt pool surface for analysis. The surface was characterized using confocal laser scanning microscopy. This high-resolution technique provided a three-dimensional (3-D) height map and a 3-D point cloud of data from which to calculate surface roughness parameters. Fast Fourier transforms (FFT) of the surface were also generated from the point cloud of data. Porosity measurements and microstructural characterization were performed using metallographic methods.
The results reveal a meaningful relationship between surface condition and internal part quality. The calculated roughness metrics were correlated with relative density and maximum defect size. Additionally, observations of surface morphology indicated different trends between specimens with distinct surface formation mechanisms. It was found that the FFT effectively differentiated various surface conditions and melting modes. It was also determined that the microstructure of specimens in the high relative density process window for scan speed and laser power have similar microstructures while possessing evolving surfaces. This indicates that the surface could be a useful metric in capturing process drift while still producing equivalently adequate parts.
This work successfully establishes the surface as a clear indicator of part health and process dynamics. The future work section suggests several avenues for applied and fundamental science projects to further advance this field.
