As explained in Material observations, Elmer’s Glitter Glue changes in an interesting way when it dries: as deposited, the glitter flakes are dispersed almost isotropically, without any preferred orientation, but drying makes the glitter flakes mostly parallel to the surface.
This suggests that by this means we can fabricate composite materials with high loadings of fibrous or especially laminar reinforcing filler (e.g., talc, clays, carbon fibers, carborundum whiskers, graphene, MXenes; see Electrolytic 2-D cutting and related electrolytic digital fabrication processes for some fuller lists) without having to assemble them layer by layer. Specifically, you bulk up the polymer with a lot of solvent, mix in your functional filler, deposit it in an X–Y sheet of uniform thickness on some substrate, evaporate off the solvent resulting in a great loss of volume, and then break the sheet free from the substrate.
Adherence to the substrate prevents the sheet from shrinking in the X–Y plane as it dries, forcing all of the volume loss to be in the Z direction; this reorients the filler particles nearly parallel to the X–Y plane, allowing them to pack much more tightly than laminar reinforcing particles normally could, potentially resulting in a final product that consists almost entirely of the filler, bound together with a small amount of matrix. If the solution is sufficiently viscous or dense, or if surface charges maintain the filler particles deflocculated, the filler will not settle out or float out before being held in place during the drying process. Thus the filler will be evenly dispersed through the resulting matrix.
Laminar reinforcing fillers are extremely desirable as reinforcement because, while each particle of a fibrous filler adds strength in one dimension, laminar fillers can add the same strength in two dimensions. So, in theory, where most of the composite’s strength comes from the filler and not the matrix, laminar reinforcement should be able to make materials that are twice as strong. Normally, though, they can only be added as very low percentages, because they either run into one another at weird angles so that more filler can’t enter, or they clump up in stacks, so their large surface area doesn’t touch the surrounding matrix, so their strength doesn’t get transferred.
Sheets, whiskers, and other fibers below a critical dimension are “flaw-insensitive”: they’re small enough that most of their length or width lacks crystalline defects, so there is no flaw at which to concentrate stress. This commonly increases their strength by an order of magnitude or more. This effect does not come into play with craft glitter.
Waterglass is another interesting possible candidate here, since it forms a soluble mostly-silica xerogel when it dries, although it may be brittle enough at room temperature that it won’t achieve super strength. Mixing waterglass with clay to repair things has a long tradition in pottery, and it deflocculates the clay due to its alkali-metal ions.
(It is possible to further harden waterglass after it dries by exchanging its alkali ions with polyvalent cations, as is done in KEIM paint.)
Alternatively, rather than evaporating the solvent, you could try removing it by some other means. For example, you could press the mass against a semiporous membrane so as to reverse-osmotically force the solvent through the membrane, but not the dissolved matrix or the filler, similar to using frit compression to produce buckypaper (which typically has no binding matrix) or slipcasting pottery. I’m pretty sure this will work in a closely analogous way; the situation is very closely analogous to the solvent case.
Maybe it would even work to squeeze a molten matrix material through a porous material (such as a sintered frit, unglazed fired clay, or dirt) in this way, leaving only the oriented filler and a small remnant of binder, which would remain when the material was cooled; in some cases, as with slipcasting, the capillary action in the porous material would itself be enough to suck away the excess binder.
The things I’m not sure about is ① whether the currents of molten binder will tumble the filler particles as they pass (I think not; I think they’ll just press the filler particles up against the porous wall, and at any rate you can do the squeezing more slowly to get slower currents) and ② whether you can separate the composite from the frit afterwards (but in the worst case you can cut it off parallel to the frit surface while it’s still almost molten).
As I understand it, the drying process normally works by first gelling the viscous solution into a hydrogel, then contracting it into a xerogel under the influence of surface tension in the nanopores of the gel. Thus, carrying out the above process with metals using mercury as a solvent may or may not work, because solid amalgams are not gels, and the shrinking process may be different from xerogel collapse. It may work anyway, though; mercury can dissolve all of zinc, copper, tin, lead, silver, and gold to an appreciable extent, and at least in the case of gold it is commonplace to recover fully dense solid gold by heating, which is how mercury gilding works.
The IUPAC solubility series volume 25 (Metals in Mercury) has the following solubilities for some selected metals in mercury at a couple of temperatures:
| Metal | Room temperature | 300° |
|---|---|---|
| Magnesium | 2.52% | 26% |
| Aluminum | 0.014% | 5.6% |
| Tin | 1.05% | >84% (tin melts at 231°; miscible?) |
| Lead | 1.47% | 93% (lead melts at 327°) |
| Titanium | 0.000017% | 0.0035% |
| Chromium | ???? | too low to measure |
| Iron | ???? | <0.00004% |
| Cobalt | ???? | <0.00007% |
| Nickel | ???? | 0.007% |
| Copper | 0.0092% | 0.6% |
| Silver | 0.065% | 5.1% |
| Gold | 0.13% | 14% |
| Zinc | 6.32% | 70% |
Magnesium is the most tempting entry here, but I’m guessing that if you were going to dissolve magnesium in mercury and then evaporate off the mercury, you’d have to do it in a way well protected from oxygen. Aluminum amalgams aggressively extrude fibers of aluminum oxide over the course of hours when in contact with air.
Aluminum, zinc, and tin are also soluble to a useful extent; you could dissolve a significant amount of zamak 3 (96% zinc, 4% aluminum) in hot mercury.
Rather than using unfashionable and costly mercury, it might be better to try to dissolve other metals in affordable and nontoxic magnesium or zinc, and then use an elevated temperature to vaporize the magnesium (boiling point: 1091°) or zinc (boiling point: 907°) from the alloy.
Water’s vapor pressure at 25° is about 3.2 kPa, 24 mmHg, at which rate it evaporates fast enough to be useful. Zinc melts at 419.5°, and its vapor pressure is well approximated by log₁₀Pₘₘ = -7198/T+9.664 (McKinley & Vance 01954), where T is in K, so it reaches that pressure at 869 K = 596°. As metal-fabrication processes go, that’s a pretty moderate temperature, which is why zinc fumes pose such a risk of metal fume fever. You might want to evaporate off the zinc in vacuum or under argon or nitrogen. (At 600° you have to use ammonia to get zinc nitride, so just nitrogen is adequately inert for this.)
Unfortunately, there’s no IUPAC solubility series volume on the solubility of various metals in molten zinc, but there are lots of phase diagrams for zinc alloys. In particular, molten zinc can dissolve about 10% Cu at 596°, and eutectics and near-eutectics used in soldering include Sn₉₁Zn₉ (KappAloy9) 199°, Zn₉₅Al₅ 382°, and Cd₈₂.₅Zn₁₇.₅ 265°, so molten zinc is evidently capable of dissolving substantial amounts of aluminum even at much lower temperatures. Below 596° no other structurally useful metals melt, but metals that melt at lower temperatures than copper include gold, silver, and of course lead, tin, and magnesium. So we might reasonably guess that, like copper, substantial amounts of those metals can dissolve in molten zinc at 596°; a paper suggests it can handle 20 mol% of silver. And in particular you ought to be able to dissolve bronze in zinc at that temperature, then evaporate off the zinc, or most of it.
More excitingly, a calculated phase diagram suggests that zinc should be able to dissolve about 3 mol% nickel at 600° and about 25 mol% at 873°, and another suggests 2 mol% iron at 600°.
Typically the critical dimension for flaw-insensitivity is a few tens of nanometers, which is an entirely practical thickness at which to electroplate. It occurs to me that if you want a lot of high-aspect-ratio sheets, you could make them out of a metal in the following way. You start by plating a nanolaminate consisting of alternating layers of your desired metal and some other metal or material that is easily etched later, using an etchant that will spare your desired metal; you might deposit, for example, 20 nm of each metal. Then you pulverize the nanolaminate (perhaps easiest if you initially plated it onto a metal where it had terrible adhesion, or onto a layer of graphite), for example by ball milling, into particles of, say, 1 μm. Then you etch these particles with the etchant and separate the resulting metal sheets, which are 1 μm × 1 μm × 20 nm in the example I’ve given.
If the adhesion between the layers of the nanolaminate were sufficiently poor, maybe you wouldn’t even need the etching step.
These high-aspect-ratio flaw-insensitive metal particles are suitable for use as a functional filler to make an ultrastrong composite material, whether the binder is an organic polymer, a geopolymer, waterglass, another metal, or something else.
Some pairs of metals cannot be plated from the same bath; in that case you have to move the forming nanolaminate back and forth between two baths, rinsing it in between. In other cases, you can make a bath which will plate only one metal at one voltage and a mixture of two metals at a different voltage. In other cases (chromium and titanium being notable here) you can grow an anodic oxide layer by reversing the voltage; this may be sufficiently thick to etch later but sufficiently thin to permit plating metal on top of it.
An alternative to moving back and forth between baths is to consume all the platable metal in one bath, leaving only, say, alkali metals; then you can inject the new metal directly into the bath. Indeed, you may be able to “inject” the new metal simply by turning off the inert cathode and switching to a cathode that will dissolve, or increasing the voltage on the cathode. By using a thin electrolyte (say, 1 mm) and cathodes even more closely intercalated (say, 0.1 mm, perhaps foils of two metals stacked alternatingly with dielectric sheets between them, like a multilayer capacitor) you may be able to switch back and forth more rapidly between metals than with a rinse tank.
Another possible alternative separator is to deposit not an anodic oxide film but the insoluble hydroxide of a metal in solution, such as magnesium, which will deposit on the cathode, just as metals do (see Fast electrolytic mineral accretion (seacrete) for digital fabrication?). Magnesium hydroxide in particular is easy to remove with many acids.