Pentane and CFCs are conventionally used for blowing styrofoam and similar low-temperature materials; propane is used for aerosol cans. For bread, yeast produces carbonic acid gas, and heating produces water vapor, while in pancakes, cookies, and crackers, baking powder produces carbonic acid gas from something like cream of tartar (potassium bitartrate) and baking soda, or for a double-acting powder, something like monocalcium phosphate and baking soda. Salt of hartshorn (ammonium carbonate and carbamate) similarly decomposes into ammonia, water, and carbonic acid gas at around 58°.
At around 100°, water evaporates. It has the unique advantage of weighing only 18 daltons, so a given mass of water will produce more moles of vapor than just about any other condensed matter, except that ammonia is only 17 daltons. By comparison, dioxygen is 32 daltons, dinitrogen 30, and carbonic acid gas 28.
But what about higher temperatures? For blowing foams from polymers like waterglass which require higher temperatures to soften, we’d maybe benefit from solids that don’t become reactive until those higher temperatures. For waterglass in particular it would be desirable that the blowing agents not release polyvalent cations, since those can raise the softening point of the waterglass to an inconveniently high temperature.
Various kinds of water-including complexes might work as fillers that can retain water to higher temperatures. Phosphoric acid retains some water up to 800°, although it’s probably not very compatible with waterglass; and its melting point drops rapidly from 40° with water content. But sal mirabilis, epsom salt, blue vitriol, alabaster, and the chlorides of magnesium and calcium incorporate quite a lot of water in their crystals, some of which evaporates at temperatures somewhat above 100°. Of these, sal mirabilis is free of polyvalent cations, and alabaster is relatively insoluble in water.
Baking soda alone evolves carbonic acid gas upon heating past 50° and fairly quickly at 100°, leaving washing soda, which I believe releases a second carbonic acid molecule at above 850°. This, however, leaves behind sodium oxide, which may not be desirable.
Chalk, of course, also releases carbonic acid at 500°-850°, leaving behind quicklime.
The formate and chloride of ammonium both decompose into gas upon heating (to 180° with traces of hydrogen cyanide in the first case), but in the case of the chloride the process is reversible. I think that the oxalate also decomposes to gases at 215°-265°, producing carbon monoxide as well. Oxalates and formates of metals also tend to produce gases when heated; those of lithium, calcium, sodium, potassium, and magnesium are notable here.
Aluminum trihydroxide (“trihydrate”, gibbsite) is commonly used as a flame-retardant plastic filler. It’s fairly alkaline, but highly water-insoluble, and does contain polyvalent cations. Upon heating past 220° it produces a mole and a half of water, leaving amorphous sapphire. Magnesium (di)hydroxide is similar, but doesn’t decompose until 330°, leaving magnesia; and slaked lime of course does the same thing, but at 400°-600°, leaving quicklime. Slaked lime is unstable in atmospheric air because it slowly absorbs carbonic acid gas to become chalk.
Sulfur boils at 444.6°, but it melts first and then polymerizes, and the vapor consists of mostly violet disulfur (64 daltons) above 720°; at lower temperatures as much as 10% may be red trisulfur. Ignition in air happens at a much lower temperature than this boiling. Sulfur has the great advantage that it is not water soluble and pretty inert at room temperature.
Autoclaved aerated concrete uses a reaction between powdered aluminum and water to generate hydrogen (2 daltons!) but of course the total mass of the ingredients is much larger than if you were to just boil the water.
Metal hydrides might be an alternative hydrogen source, though calcium hydride (hydrolith) doesn’t even melt until 816°, and even LiH (8 daltons!) doesn’t melt until 688.7°. These readily produce hydrogen gas with water, forming the hydroxides. Thermal decomposition of LiH is feasible, but the resulting lithium metal doesn’t boil until 1330°.