Plaster of Paris is easily formed at room temperature, nontoxic, and somewhat refractory, withstanding temperatures up to 1200°; it decomposes by releasing vitriol at 1460°. But it doesn’t insulate against heat extremely well, and it weakens somewhat when heated enough to start dehydrating again, becoming easily crumbled with the fingers. It reputedly strengthens again above 800°, but if that’s true, it must be a process that takes more than the few minutes I tried. It has a bad reputation for use in forging iron, because while it will survive up to 1200°, it slowly degrades at iron-forging temperatures. NIST wrote about the various stages of alabaster calcination in the 1940s.
Once re-calcined, the plaster remains solid (except that portion heated to white heat, which evolves a vitriolic air), but is enormously more fragile than before; rubbing it between fingers produces a fine floury powder. It also contracts slightly, producing cracks when heating is uneven. These are more serious problems for larger objects, which more easily collapse under their own weight, than for smaller ones.
Fired clay and portland-cement concrete can be foamed — aerated with bubbles — to improve their insulation capabilities, make them more resistant to crack propagation, and reduce their weight, at the expense of strength. I haven’t heard of anyone doing the same thing with plaster of Paris or lime cement.
There’s a special-effects material described as “aerated plaster” called Gypsnow: it expands rapidly when wet; remains soft; absorbs impacts. “Place 3.7 liters or 125 oz of water into a 5 gallon plastic bucket. Add 10 lbs of Gypsnow and mix with an electric drill, pour the mixture into a lined plastic form and after about an hour you can remove the item.” It claims that this is “aerated” but to me it sounds like maybe it has styrofoam or some hygroscopic polymer in it.
Mixing expanded perlite or, better, expanded vermiculite into the plaster would surely work. But I think they are less refractory than the plaster itself.
WP says vermiculite bonded with vaguely specified adhesives including sodium silicate is good to 1150°. It says perlite is only good to 850°. Hydrated sodium silicate itself will foam up like vermiculite or perlite when you heat it, but the remaining solid material is still sodium silicate, and retains the very low softening point of alkali silicates. Perhaps this is less of a problem in the vermiculite composite, where perhaps the sodium can diffuse away into the vermiculite rather than leaving points of contact very vulnerable to melting.
Abandoned US patent application US20160339606A1 describes trying to reinforce plaster of Paris with graphite and “cenospheres” (making the plaster a syntactic foam; for example, of vermiculite) so it will retain more strength for use in molding high-temperature (310+°) thermoplastics. (The patent application pointed out that once you’ve used it in molding you can wash it out with water, since the plaster becomes water-“soluble” again at the casting temperatures.)
The standard way to foam fired clay is to mix it with sawdust, coffee grounds, used yerba mate, or a similar granular material that will burn off in the kiln. This might work with alabaster plaster too. I have made a lump of plaster mixed with yerba; it is quite hard, less dense than plain plaster, and pale green, and it seems to survive fire a bit better than plain plaster; but I haven’t yet had the chance to fire such an article long enough to burn out all the organics.
Naphthalene is reportedly used in this way to make porous grinding wheels, boiling out at quite reasonable temperatures instead of needing to be oxidized away like yerba mate. A wide variety of substances would work as alternatives; they need to survive the plaster’s setting process in solid form — so water ice, for example, will not work; be capable of breaking apart into granules and remaining as separate granules, ruling out, for example, chewing gum and candle wax; not be soluble in or reactive with water; not react with the plaster itself, which I think might rule out, say, iodine; be easily removable after the plaster is set, for example by heating (as naphthalene) or dissolution with another solvent (as brimstone or rubber in disulfuret of carbon); and, ideally, be very cheap.
Other compounds with these properties include para-dichlorobenzene, camphor, brimstone, rubber, and chlorargyrite, which last requires ammoniated water to dissolve. A more extensive note on some such possibilities, though mostly water-soluble, is in Inorganic Burnout. Many common plastics including LDPE are also suitable; though more expensive than the plaster itself, they are reusable. Brimstone has the unique advantage of being very cheap as a petroleum waste product, and its common form melts into a thin liquid at 115°. However, it requires precautions against fire.
Fired-porous-clay kitty litter is another possible porous aggregate, similar to vermiculite but denser.
Ttk Ciar suggests mixing a low-boiling-point material into the mix, such as isopropanol, so that it will form bubbles when heating. Maybe a baking powder would work well, as suggested by “Ken” in 2013, who also suggested trying dishwashing detergent. I baked such a loaf of plaster with baking powder; the top of the foam seemed a bit spongy during baking, but the expansion seems to be less than a factor of 2. I turned off the oven after about 20 or 30 minutes of baking; a couple of hours later it seems to be setting reasonably well, resembling a lava rock, so I demolded it and put it in a plastic bag to finish setting overnight.
The cylinder is about 30 mm high and about 80 mm in diameter.
Upon being flamed with a butane torch, the foamed plaster turns first black before turning a brighter white than originally, signaling the presence of some off-white organic compound, maybe bitartrate of soda. It seems to be an open-cell foam; I can blow through it, which is probably why it didn’t expand further. Unfortunately it also has an aroma something like formaldehyde, so I question whether putting my lips on it was a good idea.
It was somewhat harder the next day, and another day later a powdery white efflorescence had formed on the surface, containing tiny soft white needlelike crystals. I suspect that the evaporation at the surface of the block carried with it either the unreacted baking powder or its reaction products.
I’d used an empty tuna can as the mold for this plaster muffin in the oven, and I think I’d scratched its internal plastic protective layer, perhaps while mixing the plaster with a chopstick, because there are a couple of thin lines of rust-colored discoloration on the bottom of the muffin. There was also a spot of discoloration on top, which I don’t know the source of.
As an interesting note on crack propagation, I was able to snap the muffin in half with my hands after scoring the top and bottom surfaces with fingernails, and then later to drive a metal wire through the whole height of the cylinder. No visible cracks propagated out from the wire.
Another possible way to foam it: hydrate it with water with a lot of dissolved gas, perhaps Priestley’s carbonated water, under pressure if necessary. The gas will bubble out and form a foam, but not before the pressure is released.