Some comments in a book on ayurvedic alchemy led me to think about possible formulas for castable refractory.
The Rasa-jala-nidhi describes part of the preparation of a nabhi-yantra or jala-yantra as follows (p. 272, 307 of 406, Bhudeb Mookerji, M.A., Vol. 1, published in 01926):
...plaster the joint by means of the following paste, which serves as a good waterproof:—oxidised iron, finely powdered,1 molasses, and lime, rubbed together with a highly condensed decoction of the bark of babbula [Acacia nilotica, now Vachellia nilotica, the source of gum arabic]. This plaster is called “water mud”, and water cannot pass through it. Similarly, chalk, salt, and oxidised iron, rubbed together with buffalo’s milk, gives rise to a plaster, called, “fire mud”. This plaster is a strong fire proof, The joint of the basin and the crucible is to be plastered with this fire mud, ...
It doesn’t really say what the proportions of the chalk, salt, oxidized iron, and buffalo’s milk are; WP says buffalo milk is a bit higher in everything (except water, I suppose) than cow’s milk. I suspect this mixture will form hematite cemented with hydroxyapatite and contaminated with salt and carbon when heated — that the milk serves as a source of phosphates, which combine with the calcium from the chalk to form hydroxyapatite, as well as protein and sugar which act as low-temperature binders; then, at higher temperatures, the protein cross-links and forms a thermoset, holding the fire mud in shape until all the organics have been converted to carbon with a little nitrogen. I suspect the salt is just present as a catalyst and the iron oxide as an inert aggregate that is stable at high temperatures.
In other parts of the book (e.g., Vol. 1, p. 41 (76/406), or on p. 10) he does give precise proportions of ingredients by weight.
I do not regard the author of this “translation” as entirely reliable; much of his book is concerned with procedures for transforming base metals into gold, silver, and copper, a process he says he has had demonstrated to him by his preceptor and is confident he could carry out himself given the appropriate equipment; moreover he claims that the human race is 1.6 billion years old. The book he “translates” from Sanskrit says it is written by one Bhudeba of the Mukhopadhyaya family, descendant of the sage Bharadwaja who, nine hundred thousand years ago, brought to India the science of medicine; his parents are Harilala Deba and Nistarini Debi; and on the next page he explains that Mukhopadhyaya is “generally written in English for the sake of brevity” as “Mookerji”.
That is, the book is written in Sanskrit by Bhudeba Mookerji and translated into English by Bhudeb Mookerji. He says the book is “based on many of those which are still extant”, but is written in “year 5026 of the perverted Kalijuga”. The Kali Yuga started on February 17/18 in 03102 BCE, so year 5026 is most of 01925 CE and a little of 01926. So unfortunately it is impossible to know whether this “fire mud” recipe dates from 01926 or a thousand years earlier, or all the way back to Nagarjuna 1700 years earlier, to whom rasāyana (रसायन) and rasaśāstra are traditionally attributed. (On p. 20 he claims that a particular prescription, one tola of purified sulfur mixed with butter, “was prescribed more than 5000 years ago. Present-day human beings can stand only one fourth to half of the dose prescribed in those days.” units(1) and Wikipedia agree that a tola is an 80th of a seer, (though Dr. Mookerji says it's a 64th of a seer) and a seer is 14400 grains by the British Colonial Standard, so a tola is some 11.7 g if the standards haven’t shifted too much in 5000 years; elemental sulfur’s oral LD₅₀ is unknown but is over 5000 mg/kg, so if Dr. Mookerji was getting toxic symptoms from feeding a “beautiful lady” 11.7 g of sulfur (<300 mg/kg) every day for three weeks, maybe his sulfur was contaminated, probably with arsenic.)
On p. 301 (345/406) Dr. Mookerji gives the chain of weights java (barley seed)x6 = gunja; ×3 = balla; ×2 = masha; ×2 = dharana; ×2 = niska; ×2 = kola; ×2 = tola. This would make the tola 1152 barley seeds. Normally a barley seed is about 40 mg, which would make the tola some 46 g rather than 12 g, but perhaps Indian barley was smaller a thousand years ago. He also says a java weighs 6 mustard seeds (sarshapas), and mustard seeds are typically about 3–6 mg, which would put a java at 18–36 mg, unless modern mustard seeds are similarly enlarged.
Still, though, I think Dr. Mookerji’s explanation of how to make a phallus out of mercury or “rasalingam” (Vol. 1, p. 10, 45/406) by dissolving one third of its weight of gold leaf into it, then rubbing it with vegetable sour juice, putting it inside a lemon, and boiling it in gruel, is much more convincing than Sadhguru Jaggi Vasudev’s explanation that consecration or energization with a divine reverberation is what solidifies the 99.2%-pure mercury or 99.8%-pure mercury in the mercury lingas in the temple he consecrated called Dhyanalinga, in the subterranean water tank called Theerthakund.
He explains that the term “pārada” comes from “para”, “end”, and “da”, “give”, both of which words amusingly enough have the same meaning in Spanish, and in the case of “da” is in fact the same word. Spanish “parar” “stop” comes from Latin “parāre” “prepare” or “make ready”, which comes from Proto-Indo-European “*pere-” “produce, procure”, which I think very likely isn't the same as Sanskrit “end”.
Getting back to the refractory recipe, Sigma says tribasic calcium phosphate melts at 1100°, which doesn’t count as “a strong fire proof” in my book, but it’s already hot enough for a pretty wide range of uses; and the carbon would tend to support it up to much higher temperatures. WP says tribasic calcium phosphate Ca₃(PO₄)₂ melts at 1670° so maybe Sigma was talking about hydroxyapatite Ca₅(PO₄)₃OH; WP says, “Most commercial samples of ‘tricalcium phosphate’ are in fact hydroxyapatite.”
I think this phosphate-formation reaction probably has the stoichiometry to actually work. Human milk has about a 1.4:1 to 1.7:1 Ca:P ratio, while cow’s milk is 1.24:1, while TCP is (by moles) 1.5:1 and hydroxyapatite is 1.67:1, so if you wanted to make calcium phosphate cement out of cow’s milk, you’d have to add about 0.25–0.5 moles of calcium for each mole of phosphate in the milk. Buffalo’s milk is reported to have higher calcium, except in Argentina:
Buffalo milk is characterized by high calcium content (about 1.5-fold Ca than in cow’s milk) as was apparent from several studies, except in BM from Argentina (Patino et al. 2007) which had Ca content comparable to that of CM.
And buffalo-milk phosphate content seems to be similar to that of cow’s milk. So it’s possible that buffalo milk wouldn’t be calcium-deficient in the same way — but I really have no idea whether we’re talking about Indian buffalo from 01926, 01826, 01626, 01226, 00426, or what, so no way to even guess at the calcium content of their milk.
Even if cow’s milk doesn’t work, though, different sources of phosphate and calcium (or possibly aluminum or boron) would surely work to form calcium phosphate (or aluminum phosphate or boron phosphate) at high temperatures.
The 01950 MIT dissertation of Wm. David Kingery (“the father of modern ceramics”), “Phosphate Bonding in Refractories” seems very much worth reading here. He outlines a wide variety of recipes then in use or disclosed in patents, and in particular mentions:
Aluminum, chromium, magnesium, and zirconium oxides react chemically with phosphoric acid at 200°C to form a bonded material. The metal phosphate reaction products are refractory and stable. Rather than using the oxides, the halides of magnesium, tin, thorium, calcium, barium, aluminum, zirconium, or titanium may be used with phosphoric acid to form a bonded refractory. After mixing the constituents to a pasty constituency, the plastic mass is formed and heated to approximately 1000°C to effect the bonding reactions and form the final product. Using aluminous refractory materials, phosphoric acid reacts to form a film of aluminum phosphate around each particle, which acts as a bond. ...
Aluminum hydrate [i.e., hydroxide] may be used with refractory clay, filler and phosphoric acid to form a bond which becomes permanent when heated to 100–300°C. The addition of aluminum hydrate to refractory compositions of zircon, silicon, etc., and phosphoric acid is also advantageous. Use of this material allows final heat hardening at temperatures of about 600°F [i.e., 315°] rather than the 1200°F [i.e., 650°] which must otherwise be applied. ...
With aluminous materials, alkaline earth acid phosphates or ammonium acid phosphates may be used in place of phosphoric acid. On heating to 200–300°C bonding action is obtained...
Acid phosphates may also be formed by addition of triphosphate [i.e., tribasic phosphate] and an acid which readily reacts with it forming mono- or bi-phosphates [i.e., hydrogen or dihydrogen phosphates]. This process may be used with alkaline earth phosphates, preferably calcium which is less expensive than other materials. ...
And that’s exactly what I speculated above is going on with the buffalo milk: it contains some amount of calcium phosphates which are not yet tribasic. Also, though, this sort of points out that the acids formed by burning the protein in the milk might help to acidify the phosphates further, perhaps replacing some of its hydroxyls with hydrogens. He also mentions a very exciting possibility of refractory inorganic elastomers:
A study of the use of metaphosphates in refractory mortars by Herold and Bust indicated that rubber-like metaphosphate polymers form a clay-gel cement with clay. However, a large amount of the phosphate was required to form an adequately plastic mass.
He goes on to mention some other possibilities:
In all, the oxides and/or hydrates of thirty-four cations were tested. ...Oxides of a highly basic nature react so violently with phosphoric acid that a porous friable structure results (MgO, ZnO, CaO, La₂O₃, BaO, SrO). ...
A large number of weakly basic and amphoteric oxides did react with phosphoric acid to produce a bonded cement-like product. Included in this group are BeO, CuO, Cu₂O, CdO, Fe₂O₃, Fe₃O₄, SnO, Pb₃O₄, Al(OH)₃, Ti(OH)₄, Zn(OH)₄, ThO₂ and V₂O₅ in addition to calcined ZnO, MgO, La₂O₃ and CaO [with partially neutralized acid].
This suggests that the iron oxide of Dr. Mookerji’s recipe might not be an inert filler as I thought.
On p. 17 (23/94) of his dissertation he has a chart of “modulus of rupture” (flexural strength, the same as the tensile strength for a homogeneous material) of an alumina mix with 10% kaolin bonded with various amounts of various phosphate cement mixes; aluminum is best at some 1300 psi (9 MPa), followed by beryllium (1200 psi, 8 MPa), magnesium (almost the same), and, surprisingly, iron (1100 psi, 8 MPa). Just bonding the mix with phosphoric acid alone yielded 750 psi, same as with barium, while calcium and thorium did substantially worse (as he puts it, “Calcium, barium, and thorium additions to phosphoric acid decrease its effectiveness [as a cement for alumina and kaolin].”). Moreover, all the good performers showed strength continuing to grow linearly with the cement fraction — he only tried up to 7% or 8% cement by weight, and while the weaker bonds like Ca and Ba were leveling off at that point, the stronger bonds weren’t, suggesting this was only a small fraction of the strength they could potentially develop.
Later on (p. 33, Fig. 9) he measures a strength of 2500 psi (17 MPa) with 13% monoaluminum phosphate (Al(H₂PO₄)₃) cement. He makes the monoaluminum phosphate sound super tempting: it gradually increases in viscosity as the water content decreases, and then you can fire it to a berlinite gel, much like waterglass. But these solutions “precipitate at pH values greater than about 2.8” (p. 34), so maybe you can do the same kind of instant hardening trick you can do with waterglass, only with bases rather than acids.
Another interesting thing he points out, when he tried mixing up some different phosphate mortars, is that at 1500° the magnesium phosphate mortar is stronger than the aluminum phosphate mortar, because the aluminum phosphate mortar hasn’t started fusing yet!
Ultimately, he says, the monoaluminum phosphate ends up as aluminum metaphosphate, Al(PO₃)₃ (p. 25), which is different from the AlPO₄ orthophosphate of which berlinite consists. But above 830°, in the absence of oxygen, as in a gasworks, the metaphosphate gradually becomes the orthophosphate! And above 1220°, again in the absence of oxygen, even that berlinite gradually calcines to sapphire. This all happens in a few hours when it’s in a very thin layer.
Oh, fantastic! His Appendix G on p. 79 has cost data. US$1.10 per 100 lb. (45 kg) of 40° sodium silicate, US$2 for MgCl₂, US$5.50 for 85% H₃PO₄, US$17 for Al(H₂PO₄)₃.
Ooh, interesting, “Ceramic properties of kaolinitic clay with monoaluminum phosphate (Al(H₂PO₄)₃) addition” at UNLP.
In isolation the phosphates that dissociate sometimes produce phosphoric acid, and sometimes pyrophosphates or polyphosphates. A “metaphosphate” is a polyphosphate chain that’s either cyclic (as in the trimetaphosphate and hexametaphosphate of sodium, rings of respectively three and six phosphate radicals) or extremely long.
Phosphates, ordered by the decomposition or melting point of their crystalline structure as a rough guide to how stable it is:
So much for phosphates; what about the cations — boron, calcium, magnesium, and aluminum?
Boron is the simplest, and borates (not to be confused with Borat) have also been used to make ordinary methyl vinyl silicone into a “ceramizable composite”. It has the advantage of having a very amphoteric oxide, so you can get your boron as a borate. Indeed, nearly all the non-borate compounds of boron are extremely exotic, highly toxic, very unstable, or all three.
It’s worth mentioning that, if I’m remembering correctly, borates are commonly used to slow down the setting of magnesium phosphate cements.
Calcium of course has many compounds, but most of them are dismayingly stable for this sort of purpose, like larnite (melts at 2130°) and quicklime (melts at 2613°), though Kingery’s experiments quoted above suggest that quicklime reacts with phosphoric acid with great violence even at room temperature. Less heat-stable compounds include:
How about magnesium? Magnesium phosphate is an awesome refractory cement. As with calcium, [the oxide] is absurdly stable, melting at 2852°, though it’s quite soft. But there are some convenient magnesium compounds that could provide magnesium to react with phosphate if heated:
If anything, most aluminum compounds are even more obnoxiously stable than those of calcium and magnesium, but there are exceptions. The obvious candidate sources for aluminum cations are aluminum trihydroxide and metallic aluminum; also, the highly water-soluble (and violently acidic) aluminum chloride sublimes at 180°.
(Aluminum hydroxide flour might also be an interesting material for an Oogoo made with hardware-store silicone, as an alternative to cornstarch; it might contribute strength, but also possibly contribute hydroxyls to speed the uniform setting of the silicone, and then, after setting, might make it possible to calcine the silicone into a silicoaluminate or carborundum-alumina composite.)