I was writing Glass powder-bed 3-D printing and came up with the idea of “powder-bed” 3-D printing that is really more of a layered paste bed, and I realized that that idea itself is maybe a lot more broadly applicable than to the lye-fluxed glass-powder bed processes with full-bed baking prior to depowdering I was exploring there.
In particular, I was exploring the idea of powder-bed 3-D printing with two binders: a “sacrificial binder” that maintains the integrity of the powder bed during the printing process but is later removed, and a selectively deposited permanent “flux” binder which defines the geometry ultimately produced by the process, and which perhaps is not active until after all the layers are patterned. My objective there was to allow the flux to be blown in dry powder form onto the powder bed without disturbing it, but now I see many other potential advantages of this design approach.
One variant for printing in concrete is to trowel on plastic layers, made of a thixotropic aqueous paste of sand; water; perhaps reinforcers like chopped basalt fibers or straw; a water-soluble temporary binder like sugar, carboxymethylcellulose, clay, or gelatinized starch; and perhaps a plasticizer, which might be the same as the temporary binder. Then, on each layer, selectively deposit a permanent cement such as waterglass, slaked lime, plaster of Paris, magnesium oxychloride, calcium aluminate, or ordinary portland cement. This requires the layers to be thin enough, relative to the mobility of the cement molecules, that the cement can diffuse down to the previous layer before it sets, or they require you to deposit the cement vigorously enough to penetrate the whole layer. In the case of a slaked-lime binder, you’d have to perforate an array of air holes at the end of the process to allow CO2 to diffuse to all the layers. Such objects wouldn’t need to be baked; they’d just need the uncemented material to be washed away with a water hose once the cement was set.
Another variant is to pour layers of liquid into a vat or pit, one by one, allowing hydrodynamics to level each layer rather than using a mechanical recoater; after pouring each layer, you solidify it with the sacrificial binder (for example, by allowing excess water to depart into porous surroundings, or allowing a solvent to evaporate, or by allowing enough time for thixotropic network formation to gel it, or by spraying on a pH-changing agent to activate a pH-sensitive gelling mechanism) and then selectively deposit the permanent binder, which might be a construction cement like those mentioned above, or might be a sintering aid like those discussed in the note mentioned above and in Dercuano, or might be something else.
Another variant is to deposit the layers using spin-coating, permitting extremely fine and precise layer thickness control. In this case, you could still pattern the layer by selectively depositing a different material on it, such as a catalyst, but another possibility is to pattern it by using light or particle beams to cause some kind of change in it, like a multi-layer version of the standard photolithography process, or stereolithography on a solid substrate rather than a vat of liquid. For example, you could draw a pattern on the layer with a moving laser, move a bar of light sources across it analogous to how a xerox machine or flatbed scanner scans an image or how an LED printer prints one, press it up against an LCD that selectively permits light to pass through, project a pattern onto it optically with a lens from a projector, or press it up against a thin mica window through which an electron beam is passing. Once all the layers have been patterned, maybe you need to bake it, and then you can remove the undesired parts of the block, for example by immersing it in a solvent.
If you were using, say, UV light to pattern the layers in this way, most of the usual stereolithography resin concerns apply: you need a component that reacts to the UV, a possibly different component that does something like polymerize or depolymerize in response, and something that blocks the UV from reaching the next level down, which has already been patterned --- a UV-opaque pigment, but not so much of it that the effect only reaches partway through the layer. The mix might include other components as well, for example to affect the mechanical or electrical properties of the final product, and these might vary from layer to layer.
In the case of patterning with electron beams, which might offer the possibility of deep submicron 3-D printing, adjusting the electron energy may be a more reasonable way to set the penetration depth to the layer height than mixing in varying amounts of opaque pigment.
If you specifically spin-coat potassium silicate rather than photopolymers, your “sacrificial binder” can be the dried potassium silicate itself, while you can inkjet-print the pattern on each layer with an aqueous solution of a polyvalent-cation salt such as calcium acetate, which will cross-link the potassium silicate into insoluble alkali-lime glass. Once all the layers are printed, if it’s been adequately protected from CO2, you can redissolve the dried but uncrosslinked waterglass with hot water, leaving a 3-D-printed object in alkali-lime glass. This may be adequately precise and transparent for fabricating optical lenses; 100-nm resolution in Z should be easily attainable, and resolution in X and Y will depend on the precision of positioning and inkjet deposition, easily reaching 30 microns. Other things to selectively deposit might include aqueous lead acetate, which will not only make the glass insoluble but significantly alter its refractive index and dispersion; water-soluble dyes to incorporate into the final piece; and various kinds of paint that are intended merely to adhere to the final workpiece rather than to react with it, such as conductive copper-filled paint.
For cases like that, where the permanent binder consists of polyvalent cations, you might be able to do your patterning with an array of transition-metal electrodes in contact with the layer being patterned, rather than with actual ink jets. By putting a positive charge on one of the electrodes, you anodically dissolve some of it and pump the resulting (generally polyvalent) cations into the workpiece. This requires that the workpiece have high enough ionic conductivity for this to be acceptably fast.
Doing things exactly backwards, you could spin-coat layers of some viscous solution of some salt with polyvalent cations, like calcium acetate, and then print “inks” onto them containing water-soluble compounds that abruptly stop being water-soluble once they see polyvalent cations; waterglass is one example, but sodium alginate is another, much gummier example, while e.g., ammonium phosphate is an example that can form extremely strong and refractory minerals with many polyvalent cations, and plain old sodium carbonate can form carbonate minerals with many polyvalent cations. So you could imagine, for example, a single spin-coated layer of zinc sulfate (maybe mixed with other water-soluble solids), on which one of your inkjet nozzles can create the biocompatible mineral zinc phosphate, another can create zinc carbonate (smithsonite), a third can create fluorescent-green zinc silicate (willemite), and a fourth can create the biocompatible antibacterial hydrogel zinc alginate, all with submicron precision at least in Z. After you’ve finished printing out your array of multi-material objects in a block of mostly zinc sulfate, you can wash away the unreacted zinc sulfate. And maybe you could have other layers that provide other polyvalent cations.
If you’re printing in a pH-sensitive gel like some of the carrageenans (or, again, waterglass), you can maintain the pH at a level favorable to gelation of the whole layer during the printing process, but then pattern each layer by loading some places on it up with a bunch of buffering agent to keep it favorable to gelation. So, for example, if your gel is stable at acidic pH but dissolves in basic environments, you could dump a bunch of a citrate or acetate buffer into the parts you want to keep; or, if it’s stable in basic environments but dissolves in acidic ones, you could dump a bunch of borate buffer in there. Then, once you’re done printing all the layers, you can immerse the block in a base (or, respectively, acid), which will immediately dissolve all the unprotected parts, while taking a much longer time to attack the printed object.
Returning to the powder-bed context, we could consider the problem of iron powder metallurgy. We can build up a powder bed of iron, layer by layer, and deposit a binder in it that works as a sintering aid (in Dercuano I suggested graphite, copper, or iron phosphide), and then bake the result to sinter the iron; but iron is not very rigid at sintering temperatures, and will tend to sag. Suppose we mix some dehydrated active alumina in with our iron powder, and repeatedly compact the powder bed during printing enough to kind of squish the iron particles together with the alumina particles, forming a friable but solid block. Now, when we bake this solid block, the iron gets gummy and soft, but the alumina particles remain as firm as ever, and have enough contact with each other to prevent any macroscopic deformation. When the loaf comes out of the oven, the fluxed part of the block is no longer friable; it’s a solidly connected network of welded-together iron particles with an interpenetrating network of alumina particles. Light wire brushing removes the unsintered volume, leaving the metal part, and lye, muriatic acid, or oil of vitriol can destroy the alumina within it, leaving the iron part with great porosity. (Infiltration may solve this, or it may be desirable.)
Alumina particles may not be the ideal “sacrificial binder” here; they’re a very poor binder, so they have to occupy a lot of the volume to work at all. There are sol-gel processes for producing alumina gels, and forming some kind of gel to encapsulate the base powder particles could allow firmer holding with much lower volume; the most common sol-gel processes are aqueous, which would be a disaster for iron powder, but some take place in other solvents that won’t attack the iron. But there may be better binders available.
For example, silica gel deposited from tetraethyl orthosilicate might do a better job of holding the iron particles in place. In its crude form its polymerization takes far too long, but perhaps if the iron powder is mixed with dry glass fibers of length comparable to the particle size, even slight polymerization would form a continuous network. Amorphous silica can handle iron’s sintering temperatures for the little while that’s necessary, and then you can remove it with molten lye or hydrofluoric acid without damaging the iron.
Shrinkage is already a big problem in ordinary pressed powder metallurgy, and the sacrificial-binder approach can solve it almost completely, at the cost of increased porosity.