Suppose you do a powder-bed 3-D printing process where the binder you deposit is a flux that lowers the melting and sintering temperatures of the powder filler, then bake the block at a temperature high enough to sinter or melt the binder-modified part of the structure. This might make 3D printing in new, unusual, or very inexpensive materials feasible.
The baseline I’m coming from here is that a spool of fucking PLA on Amazon goes for fucking US$23/kg. And, yeah, PLA is light and dimensionally stable and doesn’t require your hotend to get super hot (180° is enough), but it’s also pretty weak (50 MPa), and it sure takes a long time to come out of that teeny little hole.
There are many possible binder-filler systems that could be used, but the one that seems most promising to me at the moment would use quartz sand as the filler and a mixture of sodium and calcium carbonates as the binder, forming soda-lime glass. Soda-lime glass has a glass transition temperature of about 570°, and so sintering should be possible a bit below that, while unfluxed quartz melts at 1670°. With 1100° of headroom, it should be very easy to keep the kiln temperature in the range required to sinter or melt only the fluxed part, not the unused filler. Indeed, it might be possible to pit-fire the piece, which typically reaches heats between 1000° and 1200°.
The usual mixture for soda-lime glass is about 73% silica, 13% sodium oxide, 10% calcium oxide, and 4% other impurities. Sodium and calcium carbonates, which are themselves fairly safe to handle, decompose to form these caustic oxides at 851° and 840°, respectively, though I suspect the presence of quartz will lower the required temperature. Sodium bicarbonate (aka sodium hydrogen carbonate), which is even safer to handle, decomposes to sodium carbonate starting at 50°, releasing water and CO₂. All of these ingredients are safe enough that they are used in cooking.
If calcium is omitted, the resulting “waterglass” material is water-soluble at high temperatures; this may not be a concern, depending on the application. The potential advantage of this method would be that sodium carbonate and bicarbonate are water-soluble, up to about 5% or 10% in warm water. However, that isn’t enough to properly flux the quartz, so it will have to be applied in solid form.
You could have a robot with a valved funnel carefully dribbling the appropriate amount of mixed carbonates onto each layer of silica sand, or you could have an entire row of such nozzles.
Soda-lime glass doesn’t start to flow freely until about 1000°, so as long as the temperature doesn’t get that high, the fluxed part of the print won’t soak into the rest of the unused filler too badly.
Lead glass might be an alternative: rather than calcium, you use lead(II) oxide (or, with air, galena or just plain lead), and rather than sodium oxide, you use potassium oxide, or in practice, potassium carbonate. Lead glass, however, is less viscous than soda-lime glass, which is undesirable in this case, and doesn’t have a lower softening point (still around 600°).
The only mineral more common (and thus, I hope, cheaper) than quartz is feldspar, or rather, the feldspars, which melt at a wide variety of temperatures, from 600° to 1200°. I’m not sure what, if anything, to flux them with, other than quartz itself. I guess calcium-rich plagioclase can be fluxed with sodium-rich plagioclase.
Sodium and potassium feldspars are traditionally used as fluxes in pottery to vitrify silica and boron trioxide, so in a sense this is just a slightly different angle on the soda-lime glass from the previous section.
Alumina (aluminum oxide, also known as ruby or sapphire) is the hardest mineral commonly used in ceramics, because the harder diamond and carborundum (silicon carbide) are tricky to deal with, cubic boron nitride is expensive (I think?), and tungsten carbide is about as hard as alumina, but more expensive.
Alumina doesn’t melt until the truly unreasonably hot 2072°, which was a great difficulty in the development of the Hall-Héroult process that converted aluminum from a precious metal into a cheaper substitute for steel. The trick that made it feasible was fluxing the alumina in cryolite, Na₃AlF₆, so that the mix melts at only 1000°.
So you could imagine fluxing an alumina powder bed with just enough cryolite to get the grains to sinter together into a glass at around 1000°. You can probably do this with an arbitrarily small amount of cryolite; it melts at 1012°, and I believe it will wet the alumina grains immediately and begin to dissolve them, recrystallizing or vitrifying upon cooling. So the question is merely how much cryolite is needed to wet the alumina grains enough to form a solid mass.
I don’t know how hard or strong the resulting cryolite-cemented alumina aggregate will be.
Lead melts at 327°; tin melts at 232°; but 63% lead and 37% tin melts at 183°. So you could flux lead filings with tin filings, and then heat the piece to anywhere between 183° (or even a bit less) and 327°. Lead costs about $2.20/kg; tin costs about $22/kg. So the mixture costs about US$10/kg, which is not outrageous, but not cheap either. (You could probably reduce the price further by reducing tin content, at the cost of a higher and less crisp melting point.)
However, lead is very dense (11.3g/cc), so a kilogram of this metal is not very much, and the tin-lead alloy is very soft; you can nick it with your fingernails. Tin itself, at 7.4g/cc, is considerably less dense.
Another problem with tin-lead solder is that it shrinks when it solidifies, resulting in a rough surface. Type metal, Gutenberg’s great invention, is a variant with a significant quantity of antimony (US$6.60/kg, melts at 630°), which prevents this shrinkage and improves hardness further. The eutectic alloy is supposedly 84% lead, 12% antimony, and 4% tin, which works out to US$3.50/kg; it melts around 241°. (The antimony-lead eutectic alloy melts at 252°.)
Type metal has a Brinell hardness of around 20, which is four times that of lead, but one sixth that of soft steel. This suggests that its tensile strength might be .36 * 9.8 * 20 = 70 MPa, a little better than PLA.
Another possible alternative is pewter, which is tin fluxed with about 1% copper (US$4.40/kg) and 5% antimony. This is considerably harder than tin because of the alloying elements, and I believe immune to tin pest, but it’s even more expensive.
Bronze (copper, which melts at 1084°, alloyed with about 12% tin; bronze melts at about 950°) and brass (copper alloyed with around 40% zinc, US$2.20/kg; zinc melts at 420°, brass a bit past 900°) are other possible alloys that could be shaped by this method. Sprinkling 12% pricey tin filings into a copper bed to lower its melting point by 134° seems ideal.
Brass is commonly used as a binder for other metals in brazing; you could, for example, use a powder bed of iron filings and deposit brass filings onto it before baking. You could easily get most of the strength and cost of the final piece from the iron filings.
The inexpensive metals — those less expensive than copper — are aluminum (US$2.20/kg), arsenic (US$2.20/kg), iron (US$0.88/kg), manganese (US$0.80/kg), silicon (US$2.40/kg), and zinc (US$2.20/kg); and we should include carbon, since iron is commonly alloyed with carbon (super cheap, depending on purity; flake graphite costs US$1.50/kg) to make steel or cast iron.
Among these, zinc in particular (even without making brass by being mixed with copper) seems like it would be a good choice as a binder for an iron or steel filler, or possibly even for aluminum. Zinc and aluminum form a variety of useful alloys, and apparently there’s a technique called “diffusion soldering” similar to this. I’m not sure what would be needed to remove the aluminum-oxide layer from the surface of aluminum powder that has been exposed to air, though, and both zinc powder and aluminum powder are a bit of a fire hazard.
Cast iron melts around 1150° to 1200°, while pure iron melts at 1538°. Steels have intermediate melting points; mild carbon steel is the most common family, and ASTM A36 is a typical mild carbon steel; it has up to 0.29% carbon and 0.28% silicon, and according to the iron-carbon phase diagram, its melting point should still be above 1500°. Cast irons typically contain 1–3% silicon and 2–4% carbon, although the eutectic point is at 4.3%. At 3.5% carbon, the melting point is reduced to 1200°.
So, you could take a bed of ASTM A36 filings and selectively flux them with 3.25% carbon and 1% silicon, then heat them up to almost 1200°, or maybe a bit more, but not past 1500°. The part you’ve selectively fluxed should sinter, and then you should be able to bake it more thoroughly to make a more homogeneous cast-iron part; deformation from the printed shape should be almost zero due to the silicon content forcing carbon to remain in graphite form.
It might be a better idea to use a much smaller amount of carbon and/or silicon, so that when the powder is heated, only a small part of each filing around each carbon grain is melted, rather than the entire filing; this way, the printed part will not liquefy completely, and the finished part will be high-carbon steel rather than cast iron.
Tungsten carbide, one of the most important industrial ceramics, can be made by reacting metallic tungsten with carbon at 1400° to 2000°; it melts around 2800°, while tungsten melts at 3422° and graphite sublimes at 3600°. So you could “flux” a graphite powder bed (US$1.50/kg) with powdered tungsten (US$200 per “short ton unit”, which is 7.19 kg of tungsten, thus US$28/kg) and heat it up to 2000° or a bit less.
Alternatively, you could “flux” the graphite with tungsten trioxide, which melts at 1473°, and heat it only to 900° to react it with the graphite and immediately produce the tungsten carbide.
All of this might need to happen under pressure, I’m not sure.
Here in Argentina, the common alternative to stainless steel or silver used in silverware and whatnot is an alloy, invented by Qing China, called “alpaca”, from the 19th-century brand name of a German company; worldwide this is typically 60% copper, 20% nickel (US$22/kg, melts at 1455°), and 20% zinc, working out to US$7.50/kg. It looks like silver, it’s bactericidal like silver, and it’s strong and easy to electroplate with, for example, silver.
The CRC Handbook of Mechanical Engineering gives the melting point of the ASTM B122 formulation of alpaca in its quaint folk units of measurement as 2030°F and its modulus of elasticity as 18 Mpsi, which are 1110° and 124 GPa in SI.
Alpaca probably cannot be 3D-printed in the way discussed above, because it melts at a higher temperature than copper, zinc, or alloys thereof. If you were to try selectively depositing copper and zinc into a bed of powdered nickel, you would have the problem that the product formed would be much larger than the powder it was deposited into.
However, perhaps you could deposit powdered nickel into a bed of powdered 75%-copper brass, and then heat it up to about 1000° or 1100°. If the nickel has sufficiently diffused into the brass, the brass will melt and run away, leaving a solid alpaca object coated and perhaps permeated with liquid brass, as long as the nickel doesn’t diffuse too far and become too dilute to prevent melting. This approach would eliminate the extra processing steps that commonly attend powder-bed 3-D printers: the careful brushing of the powder from the recesses of the solid part and extra processing steps to eliminate porosity from the solid part.
This process may depend sensitively on grain sizes and morphologies and on the temperature profile of the process. It’s necessary for the nickel to have diffused enough into the alpaca part to solidify it, and for its grains to have sintered together enough to hold together, before the unmodified brass becomes liquid; but if the nickel diffuses too far, you will lose surface detail, as some brass outside the desired part acquires enough nickel to keep it from melting, while some of the alpaca inside the desired part loses enough nickel to allow it to melt.
This approach should also work for printing carbon steel: by increasing the carbon content of the unwanted part of the metal to the eutectic 4.3%, it should become a low-viscosity liquid cast iron at a sharp eutectic melting point of 1148°, permeating the pores of the sintered higher-melting steel and contributing carbon to it.
Adding so much carbon is not quite as trivial as it sounds, because the iron weighs 7.8 g/cc, while carbon black weighs only about 2 g/cc, and graphite about 2.2 g/cc. So 4.3% carbon by weight is about 15% carbon by volume, which is significant. I think it should still fit into the interstices of the iron filing bed if the carbon particles are sufficiently smaller.
(And yes, this is an even bigger problem for alpaca.)
