Abstract
Space filling micro-architectured lattice materials based upon unit cell arrangements of trusses that can be translated in two or three dimensions to create low-density cellular lattice structures have attracted interest for structurally efficient sandwich panels, impact protection and thermal management applications. When made from low density metallic alloys, the lattice materials and structures have attracted interest for lightweight, load bearing applications, and have traditionally been made via investment casting, micromachining and brazing-based approaches. The recent emergence of additive manufacturing processes such as Electron Beam Melting (EBM), has enabled the one step fabrication of metallic lattices from metallic powder and opened a potential processing route for their widespread fabrication. However, while significant work has addressed the use of EBM methods to make lattice (and other) structures with truss dimensions that are large compared to the powder size used for fabrication, little work has investigated the application of the EBM method to low density lattice structures made of slender trusses with dimensions that approach that of the largest powder particles. This dissertation explores EBM powder bed manufactured Ti-6Al-4V octet truss lattices with strut diameters that approach the lower fabrication limit and investigates the dependence of their density, elastic modulus, compressive strength and mode I fracture toughness upon the fabrication process-controlled truss topology and microstructure.
Octet truss lattices were designed with strut thicknesses that were only several multiples of the Ti-6Al-4V powder diameter used during EBM fabrication, thus testing the limits of this process’ capabilities. State of the art EBM processing methods were used to make a wide variety of lattice structures. Micro X-ray tomography was used to characterize the strut topology and characterize the strut roughness and internal voids that were formed during electron beam melting and solidification of the powder. This revealed the lattice strut surfaces were covered with partially melted powder particles, resulting in a significant amount of lattice mass that inefficiently supported applied loads. The use of a hot powder bed held at a temperature of 600 °C was found to result in a lamellar α/β phase microstructure and resulted in an as-built solid material elastic modulus, yield strength and ductility that we substantially less than conventionally processed Ti-6Al-4V which exhibited an equiaxed α/β phase structure. Structures had a porosity up to 0.31% by volume.
In 2015, Dong et al., reported the compressive stiffness and strength of “snap-fit’ modified topology octet truss lattices, which were fabricated by waterjet cutting 2D wrought Ti-6Al-4V parts which were “snapped” together along their large nodes to form a 3D lattice which was then consolidated through vacuum brazing. A set of EBM modified topology lattices of similar geometry (with similarly large nodes) were manufactured and compression tested in this study in the as-built condition showing similar strength and stiffness to their snap-fit counterparts. The modulus and strength of the EBM lattices was well predicted by micromechanical models modified to address their extra nodal mass. This model predicted that the peak compressive strength of the lattice was governed by inelastic buckling for all but the lowest density (most slender trusses) which collapsed by elastic buckling. The solid material tangent modulus at the strut critical buckling stress therefore controlled the peak compressive strength. The tangent moduli of the EBM lattices was greater than their wrought alloy counterparts alloying them to achieve similar strength.
Two sets of similar octet truss lattices were compression tested in the as-built and hot isostatic pressed (HIP) condition, with the HIP treatment utilized to eliminate sample porosity and increase alloy ductility. Despite the HIP treatment resulting in coarsening of the lamellar α/β phase microstructure and a further reduction in solid material strength, the HIP treated lattice stiffness and strength were similar to the untreated lattice set, in part due to the reduction in porosity to near undetectable limits. Because the solid material properties of the as-built and HIP EBM were deduced from just one tensile test which required using an estimate for their load bearing area, the conclusions of this study would be enhanced if more precise measurements are found.
The mode I fracture toughness of the octet structures was compared amongst geometrically equivalent sets of as-built, HIP treated and also HIP treated lattices that underwent an acid etching treatment that was developed to remove inefficient mass from the strut surfaces. The HIP treatment, which enhanced solid material ductility and reduced porosity, also reduced constituent material strength and resulted in a slight reduction in lattice toughness. The subsequent chemical treatment further enhanced the tensile ductility and resulted in removing a significant amount of the inefficient partially melted powder particles on the strut surfaces, resulting in structures with the highest toughness. The EBM lattice relative (scaled to cell length) toughness was greater than snap-fit wrought modified node structures, measured in 2017 by O’Masta et al., which suffered from a significant allocation of mass to enlarged nodes and also inefficient braze mass, that was estimated to have contributed approximately 10-30% to the mass (and therefore density) of the structures without contributing to the plastic strain energy dissipation during fracture. Further, the braze has been shown to reduce the ductility of the alloy, potentially weakening the structure. The EBM lattice sets did not have the large nodal mass so their effective strut lengths, defined as the edge-of-node to edge-of-node length, were longer. Analytical models show that octet lattice fracture toughness scales with the square root of the strut length, providing evidence that the ideal octet geometry has more resistance to fracture than the modified node variant, all else being equal.
