Thinking a bit about thermite, it occurred to me that, for sand casting or investment casting of metal objects on the scale of centimeters, it might be best to generate the metal object right on top of the mold, thus avoiding the necessity to open a hot furnace, carry a red-hot or white-hot crucible, and pour the crucible. Copper in particularly is notoriously tricky to cast in this way.
The first version of this process I saw was back in the 1990s with microwave-oven casting: some guy whose name I forget stuccoed his clay lost-wax mold with magnetite and graphite as susceptors, taped over his microwave-oven fan, and microwaved the assembly until it was hot enough to melt metal. A more recent incarnation of a similar idea used a microwave-oven-sized tiny porous-insulating-refractory kiln with charcoal inside of it to calcine magnesia to make a magnesium-oxychloride knife from mostly seawater; the refractory is transparent to microwaves, and avoids the necessity to cover up the ventilation fan, but the charcoal picks up the microwaves. The stucco guy explained that magnetite works better than graphite, but only up to its Curie temperature, at which point graphite starts to work better because of its higher resistance; presumably less pure forms of carbon such as charcoal have higher resistivities and therefore work down to lower temperatures as well.
I’ve also had success igniting arcs between pieces of charcoal and steel wool inside a microwave oven, and such arcs would also work to heat up the interior of an insulated ceramic space like that. (I also left a glass of water elsewhere in the microwave to limit the risk of magnetron overheating in case my susceptors proved too reflective.) Peepholes in an optically-opaque insulating refractory kiln, whether for use in a microwave oven or not, might permit the pyrometric inspection of the blackbody interior when the microwave and thus the arc is turned off. (I was doing the arcs out in the open on top of a bed of granulated salt, because I had no sand; the molten salt globs were easier to remove from the granulated salt than the molten glass globs were from the glass floor of my microwave oven.)
(Silicon carbide is another susceptor with a wider temperature range than either magnetite or graphite.)
If you can ignite thermite (whether by arcs, Joule heating, or by any other means), in a sand funnel scooped out of the top of a metal-casting mold, then you can presumably fill the mold with the liquid metal produced, whether that is iron from an aluminum/hematite thermite or copper from an aluminum/cuprite thermite. Moreover, if the reaction chamber is sufficiently refractory for the reaction not to melt through its bottom, it can be used to heat an insulated reaction chamber rapidly to 2500° to 2800°, a temperature that can be calibrated by the thermite’s stoichiometry rather than regulated by feedback, and which may be useful for other reactions that are difficult to perform at more convenient temperatures, such as the graphitization of amorphous carbon foam, carbothermic reduction of metals with refractory oxides, and so on.
(This process is routinely used with graphite crucibles or sand molds for welding copper conductors and railway tracks.)
Here in Argentina, on Mercado Libre, hematite (“pigmento oxido de hierro rojo”) costs US$4 to US$6 per kg, while magnetite is in around the same range. I can’t find cuprite; see the note on copper salts for more. Cupric oxide (tenorite) is much easier to prepare in bulk, but the resulting thermite acts with deadly rapidity; you might be able to reduce this menace by diluting the thermite mix with less-active hematite, excess aluminum, copper filings, iron filings, silica sand, or even — at the risk of producing hot caustic gas — blue vitriol. A small amount of excess aluminum and perhaps hematite or iron should produce an aluminum bronze rather than copper; CuAl10Fe3 is 8.5%-11% aluminum and 2%-4% iron, the remainder being copper. Aluminum bronzes are lighter, stronger, more corrosion-resistant, and lower-melting than copper.
Any excess aluminum would necessarily require that the reaction vessel be carefully purged to eliminate gaseous oxygen to reduce the risk of an aluminum fire. At ordinary temperatures carbon dioxide is not sufficiently inert to escape this danger, although paradoxically above about 2300° at atmospheric pressure the equilibrium goes the other way, and carbothermic reduction of aluminum becomes possible.
In any case, mixing enough of the desired end product into the thermite ought to tame it adequately, although thermite is plagued with hazards stemming from surprising interactions accelerating its action.
Sulfur is another possible oxidant for powdered metals, particularly in a sealed pressure vessel. This poses the risk of the sulfur boiling if the reaction is not fast enough and pressure is not contained, but it has a couple of very interesting benefits. First, it is possible to weld aluminum with this method, producing only aluminum sulfide (melting point 1100°, well above aluminum’s 660°; very hard but decomposes in water to aluminum hydroxide and hydrogen sulfide, from which the sulfur can easily be recovered). Second, iron pyrite is a beautiful and interesting material in its own right, quite aside from its historical usage in starting fires and in semiconductor diodes. Finally, while metal oxides like those of iron, manganese, and copper inevitably leave behind a residue of the reduced metal as well as the molten oxide, sulfur can produce a pure molten metal sulfide if the stoichiometry is correct.
Such welds can in principle be executed by pressing a powder of the thermite, or even just the oxidant, in between blocks or grains of the solid material to be welded, then igniting the mass, thus forming the weld. Although the sulfides formed will weaken the weld, being as they are weaker than the bulk metal (except perhaps in the cases of aluminum and zinc), if the weld remains hot long enough for them to spheroidize, the loss in strength may be minimal. As I wrote in Dercuano, it may be possible to use such a process as a way to rapidly sinter a green body primarily composed of metal grains.
(I suspect it’s possible to weld magnesium by this method as well, a task which is challenging in general and impossible with ordinary thermite.)
By selectively depositing a small amount of oxidant in a bed of metal powder which is then suffused with an inert gas before ignition, as I wrote in Dercuano, it should be possible to 3-D print near-net mostly-metal objects with a rough surface and a small amount of oxide or sulfide trapped in spheres deep inside of them. Moreover, this should also be possible with selective deposition of a paste consisting mostly of metal powder with a small amount of oxidant and binder; the oxidant might be liquid sulfur or amorphous sulfur, in which case no extra binder would be needed; or the oxidant might be crystalline sulfur or oxides, hydroxides, or nitrates of metals, in which case the binder would be some other material such as an aqueous bentonite colloid or molten lead-tin solder. A third 3-D printing process would involve selectively jetting a binding agent, perhaps just water, into a powder bed deposited layer by layer, followed by the depowdering of the green body in the way that is currently usual for such powder-bed processes; then the green body would be ignited, perhaps after drying.
In all of these 3-D printing processes, you could use inert, dense, high-boiling, and endothermic fillers to reduce the tendency of the ignited body to evolve gas and blow itself apart; incorporating adequate porosity into the design would also help. Silica, silicon, lithium, phosphoric acid, olivine, lead, tungsten, and many other substances could be useful here. Sodium chloride and alumina are in use for this purpose today.
Other ways to heat a charge of metal in an insulating chamber immediately atop a mold, so as to drop it into the mold once it finishes melting and immediately make a cast, include the use of a small in-place carbon-electrode electric arc furnace (as demonstrated by, for example, The King of Random, RIP); a small induction heating coil, which can be placed outside the refractory chamber itself (I’ve written about sealed insulated induction furnaces previously in Dercuano); and optical heating through a peephole, either by a focused laser or by focused sunlight, perhaps previously “collimated” by passing it through a “pinhole” before focusing, in order to permit the usage of smaller solar ovens.
Peepholes of transparent silica aerogel or aerogels of other high-temperature ceramics, such as yttria or yttralox, may permit pyrometry and optical heating without loss of heat to convection, although at these temperatures radiative heat loss is probably more significant. They can also prevent contamination of the reaction chamber by reactive gases such as oxygen, though perhaps purging the vicinity or directly the chamber with other gases such as argon may be more effective, or the loss of scarce reaction products such as vapors of gold or mercury.
A more everyday way to melt refractory metals and reduce ceramics less refractory than tungsten is to heat them with a carbon arc in argon on a water-cooled copper hearth, which can provide the necessary grounding. However, this approach is not very efficient due to the high thermal losses into the copper, and might result in copper or carbon contamination of the melt.
Another potentially interesting powder-bed 3-D printable end product, which I didn’t appreciate at the time, is the possibility of printing in yttria-stabilized cubic zirconia. This could be done either by sintering or fusing a bed or green body of powder of the ceramic with any of the thermite compositions described above, or, more outlandishly, by selectively oxidizing a bed or green body of metallic zirconium powder (just US$12–33/kg from China in 2015-19, according to the USGS!) with the appropriate percentage of yttrium present (about 10% on an yttria basis, ideally already as the trivalent oxide, US$3–8/kg, but conceivably just as the metal, US$34–48/kg, or as some other salt such as the acetate, formate, nitrate, or sulfate, all of which are water-soluble; or as the hydroxide). The oxidation process would inevitably leave bits of the reduced oxygen-donating metal behind, probably trapped inside the zirconia mass, probably weakening it somewhat and potentially cracking it. (The candidate donor metals — iron, copper, chromium, and so on — are strong, hard, and tough, but cubic zirconia is much harder, so it cannot transfer mechanical loads to them unless it cracks.)
No such concern about donor metal remnants applies for oxidation to zirconium carbide, which can perhaps be done with just zirconium and carbon; however, the temperature is probably high enough to melt any remaining zirconium, so depositing a bit of zirconium into a bed of graphite or diamond dust might be better than vice versa. Zirconium carbide is even more refractory than zirconia (to the point that I doubt the above-mentioned thermites can sinter it), but its standard enthalpy of formation is smaller, only -207 kJ/mol to zirconia’s -1080 kJ/mol and alumina’s -1675.7 kJ/mol; so it’s conceivable that it might need a boost to fully fuse upon ignition. Analogous considerations apply to zirconium boride, for which this process is already in use, under the name “self-propagating high-temperature synthesis”.
One clever trick from zirconium boride SHS, probably applicable to this entire class of processes, is to include metallic magnesium in the mix to capture unwanted oxygen from the feedstocks, preventing it from outgassing; the magnesia can then be removed with mild acid leaching after cooling. Glucina might work better for such oxygen immobilization in the sense of occupying less volume, and it is harder than magnesia, though slightly less refractory; but removing it later requires more aggressive chemicals, and of course it is considered very toxic.
Cubic zirconia is the stable structure for zirconia above 2370°, so in the case of producing zirconia it is probably necessary for the temperature to exceed this level. Candidate alternative stabilizing dopants include calcium, titanium, and magnesium; calcium and magnesium oxides in particular might be particularly easy to handle, and they provide superior mechanical properties to yttria! However, historically sintering them has been too difficult, a problem this thermite-printing process might solve; they also have worse high-temperature stability than traditional yttria-stabilized zirconia.
Including aluminum in the zirconium mix might offer some additional advantages. I think it won’t give you hotter results — while I think aluminum has a higher energy density of 83.8 MJ/ℓ than zirconium; alumina’s molar mass is 101.960 g/mol, while zirconia’s is 123.218 g/mol, giving respectively standard enthalpies of formation of -16.4 MJ/kg and -8.8 MJ/kg, advantage aluminum; however, while alumina’s specific heat is the low 0.88 J/kg/K, zirconia’s specific heat is a super-low 0.27 J/g/K, so I think zirconium as a thermite fuel will actually get hotter than aluminum, though I think it’s entirely likely I’m misunderstanding how to apply the thermodynamics. Also, though, the resulting alumina–zirconia composites offer better mechanical properties than either ceramic alone, being harder than zirconia but tougher (higher tenacity) and consequently stronger than alumina.
Several of these powder-bed processes might benefit from the powder bed being pressed in a hydraulic press, as for hot isostatic pressing but without the heat, at the time of ignition. This would tend to accelerate the reaction dangerously, but it might also diminish the tendency for the reaction to blow the workpiece apart or produce porosity, or for the heat produced to deform the workpiece.
Of course other similar metals, such as titanium, tantalum, hafnium, niobium, vanadium, molybdenum and thorium, can be used instead of zirconium to 3-D print similar ultra-high-temperature ceramics in similar ways; titanium carbide and molybdenum boride, among many other examples, have been made by SHS. As another example, there’s a 1997 paper by Sundaram et al., that got titanium diboride by SHS of magnesium, amorphous boria, and rutile.