October AMP_Digital

A D V A N C E D M A T E R I A L S & P R O C E S S E S | O C T O B E R 2 0 1 9 3 0 Even so, many ceramic powders con- tain small amounts of hard agglomer- ates that can interrupt flow. A related issue is the elasticity of the extruder. Disposable plastic extrud- er barrels are convenient to use but tend to swell under pressure. With a gas-pressure system, flow can be quick- ly turned on and off. It is more complex with a motor system, where the mo- tor must be rapidly backed up to stop flow. This works well with filament-fed 3D plastic printers, but the robocast- ing system generally runs with a con- tinuous flow and rapid table motions to traverse between separated printing positions. If the extruding motor is sim- ply stopped, several extra centimeters of bead are typically extruded. Another problem is bubbles in the slurry. Even with centrifugation and vacuum degassing, it can be difficult to achieve a bubble-free filled syringe. Two particular rheological issues in extruding slurries are filter pressing and dilatancy. We often observe that a material will extrude well for half the volume of a syringe and then stop, leav- ing dry packed powder in the barrel. Liquid can flow between larger parti- cles, concurrent with the flow of slurry through the nozzle. This excess liquid flow causes the remaining slurry in the barrel to dry out. It is simple to use the Poiseuille equation to calculate the flu- id flow rate through a nominal path be- tween the particles in the nozzle. This can be minimized by decreasing the particle size, increasing the viscosity of the carrier liquid, or reducing that of the slurry by reducing the particle loading. Dilatancy is well known in Oo- bleck, a high-loading dispersion of starch particles in water. The precise mechanism is unclear and there may be several, but the slurry solidifies un- der shear stress. In our systems, this means that a material extrudes well, but the pressure starts to build up and then flow stops completely. Releasing the pressure may cause flow to restart. Our parts are normally printed through a nozzle of 0.3-1 mm diame- ter. Finer nozzles require higher pres- sures and longer printing times and are much more vulnerable to particle agglomerates. Higher resolutions also require stiffer printers and very flat sub- strates to maintain an accurate spac- ing between the nozzle and the part. It is noticeable that 3D polymer printers often set a layer thickness smaller than the nozzle diameter, so that a ribbon is extruded. We tend to consider the bead as roughly square and with the same cross-sectional area as the nozzle. As with other systems, surface finish can always be improved by green machin- ing or polishing the sintered part. CURING METHODS The original robocasting system [6] was a high volume fraction suspension of alumina in water. A combination of evaporation (from the first layer) and wicking of water down into underlying layers causes the slurry to solidify in seconds. For slurries that show a definite yield point, various chemical curing systems (e.g., the epoxy-amine reac- tion or free radical polymerization) are used. Curing should not occur in the sy- ringe during extrusion, so a pot life of 1 hour is necessary. Curing after depo- sition can be increased by printing onto a hotplate. Free radical systems tend to be left with sticky surfaces due to oxy- gen inhibition. Many possible variants exist with regard to curing chemistry. Recently, we have been using photo- curing methods that provide some so- lidification and curing during printing. However, this only works with white powders and slurries with limited light scattering. Two important considerations in- clude bridging and interlaminar ad- hesion. For open-mesh parts, a slurry should be stiff enough to bridge a gap of several millimeters without signifi- cant sagging. In addition, partial curing during printing enables better bonding between layers. SHRINKAGE AND WARPING Most solidification processes in- volve a volume change of a few per- cent or more, which inevitably leads to non-uniform stresses during 3D print- ing. The degree of warping depends on the stiffness of the part as written, part geometry, and adhesion to the substrate. Conventional beam-bending theory can be used to predict warping. Despite the robocasting example, it is generally preferable to avoid us- ing solvents that can evaporate during printing. Heavily filled ceramic green bodies can be brittle, so cracking be- comes an issue if substrate adhesion is strong. MECHANICAL PROPERTIES In general, the mechanical prop- erties of the green body need only be good enough to allow gentle handling, which is rarely a problem. Going from the green body to the final sintered ceramic involves the same concerns as tape-cast or molded ceramics, and the final properties are very similar. An advantage could potentially be seen in that ceramics are quite vulnerable to stress concentrations; a clever de- sign of the printed system may reduce this problem. Fig. 2 — Multi-head instrument used for parallel manufacturing of a single part.