Probably the polyvalent cations are bound up in a salt that has negligible solubility at room temperature, so the ink doesn't harden in the reservoir, but which can react with the silicate at 250°, maybe a hydroxide or carbonate. Calcium sulfate is probably too soluble.

Presumably the silica filler is amorphous and serves to strengthen the glass and reduce the TCE, though it probably raises the cost. Maybe it also makes the paste thixotropic without needing to include organics like carboxymethylcellulose which would be hard to remove later. And I guess maybe it could react directly with the waterglass to solidify it, in effect raising the waterglass's modulus out of the water-soluble region, without requiring any polyvalent cations. Silica fume is the most likely form of silica here.

If anyone digs up more details, I'd love to see them.

Hmm, I see gsf_emergency_2 found this patent from 02020: https://patents.google.com/patent/US11499234B2/en which is later than Dercuano but earlier than Derctuo and Dernocua. But its priority date is from a provisional patent application from 02019, so even Dercuano doesn't count as prior art, even if it anticipates some of the claims. Also, though, I don't see anything that anticipates this process in notes like https://dercuano.github.io/notes/flux-deposition.html and https://dercuano.github.io/notes/powder-bed-3d-printing.html, which suggests that it wasn't as obvious as it sounds.

The patent says, "curing the material to evolve gaseous water," which makes it sound like thick sections would foam up. (You have to heat it over a length of time quadratic in section thickness to permit water to diffuse through the nascent solid.) And supposedly that is the purpose of the heating, rather than provoking the other reactions I speculated about above.

There are some interesting notes in the patent about promising functional fillers, such as carbon fiber for strength or silver flakes or conductivity. It suggests using these inert fillers at loadings of 45% to 90%.

It confirms that the silica is fumed silica, and suggests as alternatives fumed alumina (sapphire) or fumed titania (rutile or anatase), which are not products I had heard of, though titania nanoparticles are common as a white pigment in paint and in food. Waterglass can definitely crosslink with aluminate sources such as metakaolin, as used in "geopolymer cement". Titanate is news to me, if true.

It also confirms my inference that the fumed silica makes the ink thixotropic, at 15 wt%, though it uses the phrasing, "introduces a yield stress into the material" and "shear-thinning behavior".

I'm definitely going to use that phrase the next time I'm making mayonnaise.

"Why are you stirring that bowl of egg yolks and oil with a fork?"

"I'm introducing a yield stress into the material!"

The patent also confirms my inference that the silicate solution is waterglass; it specifically says sodium silicate (or ammonium silicate, which is probably not a practical possibility: http://www.sciencemadness.org/talk/viewthread.php?tid=10297)

The patent doesn't suggest using ionotropic gelation in the way I suggested at all. It doesn't mention the mineral oil either.

I wonder what the other inorganics ("functional additive, embodied in some cases by Silver flakes") are, exactly. In their flagship product.

And the size of the nozzle, why not the easier to remember 400 micrometers?

https://en.wikipedia.org/wiki/Water_glass

but if you want to plqy with some, it is readily avail8ble as a "cement" used for attaching seals to wood stove doors, and countless other low teck applications the comon name of "water glass" refers to it's solubility, so.....