First Some Background

A project log for Semisolid Metal Printing

Print metal just like plastic.

Michael PerroneMichael Perrone 06/25/2021 at 02:080 Comments

Semisolid Metal Printing has been attempted with limited success since about 2013. If memory serves, Lulzbot was unknowingly the very first to attempt semisolid metal printing, when they tried printing a solder with slightly off-eutectic composition. This key feature is why it worked as well as it did, though it seems they were not aware of that detail at the time. A few months later, a maker whose name escapes me (name started with an S? Post a note in the chat if you remember) attempted semisolid metal printing with antimony alloys of his own creation. From what I understand of the composition, would have been nearly ideal for semisolid metal printing, but both he and Lulzbot inevitably encountered nozzle clogging. This was due to an effect called granular jamming**, which is common in semisolid alloys but absent from clean thermoplastic. In this log entry I will attempt to include all the relevant context to understand how this extrusion system avoids granular jamming and other issues.

There are are at least five factors that must be considered when printing semisolid metals as you would thermoplastics. Firstly, the reactive wetting models describing metal are vastly different from the standard wetting models used for water or even polymers. Secondly, the range of viscosity and cohesive forces experienced by molten metallic systems differ drastically from those experienced by thermoplastics, making the Plateau-Rayliegh instability* difficult to overcome. Third, dissolution an of the nozzle/deposition head in the molten alloy must be taken into account. Fourth, metallic systems can experience granular jamming** which is almost entirely alien to thermoplastic systems, except for thermoplastic composites with solid powders added. Fifth, the material properties of semisolid alloys may have much more sensitive dependence on temperature than thermoplastics do, viscosity for example ranging over many orders of magnitude in as little as a few degrees Celsius depending on the alloy.

The Plateau-Rayliegh instability can be mitigated at the same time as granular jamming by sing a brush to guide the molten metal and printing from above the liquidus****** temperature so that the material only passes into the semisolid range as it passes over the brush. This also enables lower defect counts and higer resolution than with welding-based additive manufacturing, because "painting" the metal on is much more repeatable than the high energy, chaotically unpredictable electrical transients common in welding.

 One additional consideration when printing with thin fluids is capillary forces. Kundan Chaudary Et. Al. at the Wyss Institute at Harvard attempted printing eutectic alloys and ran into the issue that the capillary pressure of fluid wetting onto a substrate tended to pull thin molten eutectic alloys through the nozzle in a variable manner, dependent on the specific shape of the meniscus formed due to surface roughness as well as on the movement speed of the nozzle. More recently, Billy Wu Et. Al. demonstrated an electrochemical 3D printer for which the meniscus was stabilized by a sponge in the fluid reservoir, balancing out capillary forces at the nozzle meniscus with the capillary forces over the surface area of the sponge. Alternatively, it is possible to use active force balancing to control deposition of thin fluids at the nozzle tip, but the use of a sponge is simple and elegant for most applications, similar to how a felt tip pen or marker works. So for the cost of a bit of flow control, it is possible to make a remarkably simple extrusion system

Semisolid alloys are often erroneously compared to thermoplastics, but their physical properties differ in some key ways which can’t be overlooked when attempting to 3D print with them. They appear viscous and shear thinning when between their solidus (where all stable phases are solid) and liquidus (where all stable phases are liquids) temperatures: in such a state they are partially molten, and behave more like wet sand than thermoplastic. This is a stable state for off-eutectic****** and peritectic******* compositions, and in between these two temperatures, such alloys are never fully solid or fully liquid. They are generally not viscoelastic materials as thermoplastics are and the liquid portion tends to have low viscosity and strong cohesion/surface tension, where thermoplastics have high viscosity and lower surface tension. Granular jamming** is possible in these systems, and many attempts to print semisolid alloys in the same manner as thermoplastics have ended in failure after being unable to overcome nozzle clogging due to granular jamming**. In the extreme case of granular jamming**, It is possible to entirely press out the liquid portion of the semisolid alloy, with the frit and the liquid fraction forming new thermodynamic equilibria with new solid and liquid fractions once separated and allowed to sit for long enough. Slightly more success was achieved with ultrasonic agitation by Lawrence Livermore National Lab, the MPI at WPI, and others, but it necessitated the construction of a nozzle as an ultrasonic horn, and the ultrasonic energy accelerated the dissolution of the horn into the semisolid alloy, detuning the ultrasonic horn until the amplitude of the ultrasound was no longer sufficient to prevent granular jamming**. The cost of that system was also likely prohibitive for industry applications. Stirring while induction heating, instead of using ultrasound, was found not to provide enough force to ultimately impede granular jamming** when the melt was substantially semisolid. When the melt had a low volume fraction of solids, granular jamming** did not occur, but then the Plateau-Rayleigh instability* prevented printing and the stream of semisolid metal broke up into droplets and did not bridge between the nozzle and the workpiece. It is generally desirable to shear a semisolid melt before total solidification because breaking up dendrites improves mechanical strength of the resulting part. Adding a brush to the extruder solves all of these issues with semisolid metal printing, and makes semisolid metal printing viable and cost effective.

Wetting of molten metal systems on solid surfaces is still not fully understood, but there are some rules of thumb you can follow to get perhaps 90% of the way there. Many of the concepts applicable at room temperature in water do not apply to high temperature metals, because the fluid medium is conductive, because covalent bond behavior is rare at high temperatures, because the energies involved in hydrogen bonding are lower than the ambient thermal energy, and because many more chemical reactions are spontaneous at high temperatures. Generally speaking, surfaces with metallic character, high surface energy, favorable lattice constants on exposed surfaces, favorable condensed matter behaviors and favorable chemistry will have good wettability for at least some molten metal alloys. Even so, the brush material must usually be engineered along with the molten alloy for best results. Aside from this, the other main factor to consider is the reactive wetting model. This model takes into account chemical reactions which may occur at the solid-liquid interface, altering the surface material of the substrate. It is possible to alter the surface material in ways which are either favorable or unfavorable to wetting. This causes droplet spreading or dewetting behaviors which depend on the surface properties of the liquid, the properties of the new solid surface, the properties of the old substrate, and the reaction rate. A special case of reactive dewetting is with oxide formation. Because oxygen is common in our atmosphere, some may dissolve into the molten alloy and diffuse to the brush surface, oxidizing it and causing dewetting. Therefore when operating in ambient atmosphere containing oxygen, it is desirable to have the brush be less reactive and less electronegative than the molten alloy, so that the molten alloy galvanically protects it. This introduces some oxide to the molten alloy, but these oxide layers are disrupted by the brush as the system prints and may be negligible when the materials and process temperatures are properly chosen. Other means of protecting the workpiece and brush from oxidation (and sometimes nitriding) also exist, but galvanic protection often suffices as an inexpensive and elegant solution. Solubility in the molten metal is also usually correlated with good wettability, but is not always desirable for 3D printing. Sometimes materials which dissolve in a molten metal have very tenacious liquid coatings due to the change in composition and rheology******** very close to the material surface, the quintessential example being copper metal bonding to tin solder, where brushing off the solder after wetting may actually take some effort. Dissolution is a reasonable concern because it can alter the composition of the feedstock alloy in an undesirable way or damage the brush so that printing fails, and usually liquid metals wet onto things which they also dissolve well. To mitigate this concern there are two strategies: using materials which are wettable but not soluble in the feedstock alloy, and using materials where dissolution is manageable or even beneficial to the feedstock alloy. Refractory materials tend to dissolve more slowly because they have lower diffusion coefficients and are therefore desirable as print brush materials if they are wettable. Ferrous alloys also often have sufficiently low dissolution rates and good wetting properties to be used as brush materials, and often their microalloying into the feedstock material can be beneficial to its mechanical properties. Another point of interest is that sometimes, materals with low but nonzero diffusion into the molten metal enable the formation of Kirkendall voids*********, where more of one metal phase diffuses out of a region than the other phase diffuses in. These voids can serve as nucleation points for dewetting events, but may also be manageable depending on how severe they are. Lower process temperatures reduce the diffusion rate and retard dissolution and Kirkendall void formation.

*The instability which causes streams of thin fluid to break up into droplets.

**Where solid phases in the molten fluid stick together to prevent flow.

***An effect which draws current-carrying conducting matter towards the center of the current.

****Where different vaporization rates of the alloy components induce a strain on the surface of the fluid.

*****Where active trace elements modify the diffusion rate and surface tension of the fluid.

****** Eutectics are alloys which have a lower melting point than their component ingredients. Off-eutectics generally have a solid and liquid phase above their solidus temperature and below their liquidus temperature, and will retain those same phases during solidification and cooling. The important point is that they have a semisolid range.

******* Peritectics are alloys with a semisolid range where the solid and liquid phases turn into a third phase while cooling and solidifying. The important point that they have a semisolid range.

******** Fluid properties of a material

*********Holes at the interface between two materials which form due to diffusion effects