I heated up the tip of my zirconia knife to orange heat yesterday. To my surprise, much of the blade turned black; I guess the knife had some oil on it, though I thought it was clean. In the hotter parts of the blade, this carbon deposit burned away, but in the cooler parts it remained. Steel wool and brass wool proved ineffective at removing the deposits.
This led me, as most things do, to thinking about 3-D printing. Suppose instead of depositing a hot liquid onto a cool workpiece which then freezes the liquid in place, as FDM printers do, we deposit a cool liquid onto a hot workpiece, which pyrolyzes it into a solid? Sort of like chemical vapor deposition, but from a liquid so you can deposit it selectively?
So, for example, you could deposit warm bitumen and pyrolyze it to carbon at around 350°.
Almost any organic substance would work; so, for example, vegetable oil, sugar, starch, and dissolved gelatine would all work, but possibly other things would work better. Small molecules tend to volatilize before carbonizing (though any cook can tell you that even light vegetable oils will carbonize before volatilizing completely, though perhaps not before migrating to cooler parts of a surface where you don’t want them), molecules with a lot of oxygen tend to produce more bubbles, and highly saturated molecules (like the ones in bitumen) and aromatic molecules are more resistant to pyrolysis, so perhaps the ideal substance would be a high-molecular weight, highly unsaturated aliphatic hydrocarbon. (However, aromatic groups tend to promote cross-linking, which helps to prevent volatilization and melting before pyrolysis.)
Moreover, it would at least remain viscous at pyrolysis temperatures, like polycaprolactone (not to be confused with polycaprolactam, which is common nylon 6 and not viscous at all), though polycaprolactone itself is saturated and contains oxygen. Something that solidifies before reaching pyrolysis would be even better (see below about polymer-derived silicon-based ceramics.) Somewhat polymerized linseed oil is a good possibility. Nylon 6,6 and nylon 6 are not very viscous or unsaturated but are a good possibility; their amide bonds could play the role of the unstable unsaturated bonds. The urethane groups of polyurethanes contain both double bonds and amide bonds, making them especially promising, though the popular polyurethanes are highly aromatic. Polyisoprene, especially if vulcanized into a thermoset with sulfur, would work perfectly.
Decomposition can be accelerated with additives; PET notoriously takes up water from humid air and then hydrolyzes rapidly at melt temperatures if not dried, so ordinary water may be a viable option, despite its tendency to produce large bubbles. Acids and bases may also help to accelerate such decomposition — ideally for this process the additive would itself volatilize or decompose, leaving only carbon. Ammonia, sulfuric acid, nitric acid, acetic acid, formic acid, and hydrogen cyanide are possible degradation-enhancing additives. Cellulose acetate has a well-known autocatalytic degradation reaction with acetic acid, but this produces a goo which may be too liquid at pyrolysis temperatures.
Additives like alkali metals and halogens might accelerate decomposition as well, but would probably remain in the final product.
Using thermosets such as the aforementioned vulcanizing polyisoprene has the advantage that you don’t have to worry about the feedstock melting and running off the workpiece before charring, so you don’t need to prefer unsaturated aliphatic compounds. Normally thermoset polymerization is tightly controlled to reduce the risk of heat degradation of the material produced, but in this context that ceases to be a problem. So any of the common thermosets — phenolics, epoxies, polyisocyanurates, urea resins, thermosetting polyesters like Lucite, melamine resin — should be fine. Thermosetting is the approach universally taken for preceramic polymers used for producing silicon-based ceramics.
People have already done this since the 1970s, though, until recently, mostly to produce fibers such as Nicalon and Tyranno, and coatings. The technique is called “preceramic polymers”, “precursor ceramics”, or “polymer-derived ceramics”, though typically in that technique the polymer shape is fully formed before pyrolysis begins — an approach that can be taken for all of the materials discussed in this note, including carbon and the metal oxides discussed below.
Large-molecule silicones are usually thermosets. Polydimethylsiloxane ((SiO(CH₂)₂)ₙ) has a 2-to-1 carbon-to-silicon ratio, which is twice the ideal for producing silicon carbide, so polymers that have been used instead include carbosilazane resin, poly(methylsilazane), poly(methylchlorosilane), and poly(carbosilane), which pyrolyze in nitrogen to silicon carbide, yielding a ceramic whose mass is some 60–75% of that of the original polymer (the “ceramic residue yield”).
An excess of silicon is preferable to an excess of carbon for producing high-temperature ceramics, since silicon melts at “only” 1414°, (XXX Cold Plasma says 1460°) while carborundum is stable to 2830° and graphite is stable to 3642°. Poly(carbosilane), (H₂SiCH₂)ₙ, pyrolyzes in nitrogen to essentially pure carborundum, but in other precursors some carbon is lost as methane or carbon monoxide during pyrolysis.
To get silicon nitride instead, an ammonia atmosphere is required both to supply more nitrogen than can be jammed into the polymer and to cleave off unwanted methyl groups. It is helpful but not necessary for the original polymer to contain nitrogen; in fact, ammonia pyrolysis can convert Nicalon to silicon nitride.
An attractive aspect of these processes is widely reported to be that low temperatures, in the 1100°–1300° range, are sufficient to produce these ceramics by pyrolysis, while the standard sintering processes require 1800° or more, and additionally contaminates the ceramic produced with “sintering aids”, in order to avoid even higher temperatures. So not only can polymer-derived ceramics withstand higher temperatures than are required for their production, they can even withstand higher temperatures than the same ceramics when they’re processed conventionally!
Some of these processes require a “curing” step in between plastic forming, such as spinning, and the pyrolysis step, in order to keep the plastic from melting before pyrolysis is complete. This curing may happen by way of cross-linking, as in rubber vulcanization, or by evaporation of solvents and other plasticizers. This approach is also applicable to pyrolytic carbon production described above.
A problem that commonly afflicts these processes is structural damage due to pyrolytic gas production during pyrolysis, which is a major reason for requiring high ceramic residue yields. As with traditional fired-clay ceramics, this is a bigger problem with thicker material sections (nonexistent below a few hundred microns), and it is to be expected that an incremental additive process in which the material is pyrolyzed before more material is laid on top of it should enable the fabrication of thicker cross-sections.
Another problem that commonly afflicts these processes is dimensional inaccuracy due to shrinkage during pyrolysis, and deposition during pyrolysis will also reduce this problem, since the shrinkage will affect individual beads as they are laid down, not the fabricated article as a whole, whose dimensional precision will be determined grossly by the positioning precision of the end-effector and only finely by shrinkage and wiggle.
Of course, there are certain practical difficulties attending the construction of a “hotend” and manipulator that can function in an environment that keeps the workpiece at 1100°–1300°. A combination of liquid-cooling and refractory insulation for a manipulator arm would probably be necessary. The thermal gradient near the deposition point poses additional difficulties: the cooler material being deposited onto the hot workpiece will locally cool and contract the workpiece, inducing stresses that could produce cracks.
Boron nitride, aluminum nitride, boron carbonitride, silicon oxycarbide, silicon carbonitride, SiCNO, SiBCN, SiBCO, SiAlCN, and SiAlCO have also been synthesized by this route. Some of these ceramics cannot been synthesized in any other known way.
Exposure to reactive substances has been used instead of or in addition to heating to remove the unwanted moieties. Examples include ammonia, nitrogen dioxide, reactive plasma, and highly alkaline solutions. These approaches could likely also be used with the other materials discussed in this note; incremental deposition of the feedstock, as by fused deposition modeling, would give the reactive environment access even to material that is ultimately buried inside the part.
Of special note here is HRL Laboratories’ high-density stereolithography resin, which produces almost-fully-dense silicon oxycarbide when pyrolyzed at 1000° in argon (“Additive manufacturing of polymer-derived ceramics”, Science, January 2016, many authors and Tobias Schaedler). Their recipe is mercaptopropyl methylsiloxane and vinylmethysiloxane (in a 1:1 molar ratio of thiol to vinyl groups), plus the usual cocktail of stereolithography additives; pyrolysis resulted in “42% mass loss and 30% linear shrinkage” to amorphous SiO₁.₃₄C₁.₂₅S₀.₁₅ but apparently no porosity or surface cracks. To reduce porosity and cracking, they limited feature size to 3 mm and heating to 20°/minute (or, according to their supplemental materials, 1°/minute), but it is not clear to me what the crucial factors were.
(For the purpose of the following, consider “metals” to include boron, silicon, and aluminum as well as the usual metals.)
A number of metal oxides form minerals with desirable properties, and it might be desirable to form them into particular shapes; many of these metal oxides are themselves refractory and chemically resistant, so casting or dissolving them is difficult. In particular, the oxides of aluminum, zirconium, silicon, titanium, chromium, thorium, and uranium are all hard, refractory ceramics, most occurring naturally as minerals.
But perhaps salts of the same metals can be formed into the right shape, whether as an solution (for example in water), a gel, a paste, or as solid particles; then calcined to yield the oxides? As the preface to the IUPAC–NIST Solubility Data Series volume on formates said in 2001:
Bivalent metal formates could be used as precursors for the production of catalysts because they show excellent miscibility in the solid state, i.e., they form mixed crystals that dissociate at relatively low temperatures (about 300 °C) to form the respective oxides and mixed oxides. Catalysts for the decomposition of alcohols have been prepared by the thermal decomposition of Ni and Mg formate mixed crystals, from Cu and Mg formate mixed crystals, and from the double salts CuSr₂(CHO₂)·8H₂O and CuBa₂(CHO₂)₆·4H₂O. ...
For this we need metal salts that decompose on heating, but ideally are soluble in water (IUPAC-NIST database); moreover, they probably need to be soluble together so they don’t precipitate in the nozzle. Basically these are either metal cations with anions that decompose on heating — nitrate, sulfate, or organic anions — or ammonium or hydronium with metal-complex anions. Tetramethylammonium is a possible alternative cation but for now I’m going to ignore it. Here’s a list of candidates.
| cation | anion | g/100g | decomposition |
| | | water | temperature |
| | | (20° if | |
| | | possible) | |
|------------------+---------------+-------------+----------------|
| aluminum | nitrate | 74 | 150° |
| | sulfate | 36 | 900° |
| | formate | 6 | |
| chromium(III) | nitrate | 81 | 100° |
| | sulfate | “readily” | 700° (to acid) |
| ammonium | dichromate | high | |
| | paratungstate | high? | 600° |
| (hydronium) | boric acid | low | |
| | chromic acid | 169 | |
| | alumic acid | | |
| | tungstic acid | low | |
| | titanic acid | | |
| | zirconic acid | | |
| calcium | nitrate | | 500° |
| | acetate | | 160° |
| | formate | | 200°? |
| | sulfate | ≈0 ☹ | |
| magnesium | acetate | 53 | |
| | oxalate | low | 620° |
| | chromate | 137 | |
| | formate | 14 | |
| | nitrate | 69.5 | |
| | sulfate | 35 | |
| zirconium | sulfate | 52.5 | |
| | nitrate | yes | 100° |
| | acetate | ? | |
| | formate | ? | |
| | tungstate | low | |
| titanium | sulfate | ? | |
| | nitrate | ? | |
| | formate | ? | |
| | acetate | ? | |
| cobalt | nitrate | 84 | |
| | sulfate | less | |
| copper(II) | nitrate | 83.5 | |
| | sulfate | very | |
| | formate | 7 | |
| ferrous ammonium | sulfate | 27 | |
| iron(II) | nitrate | 134 | |
| | sulfate | 29 | 680° |
| | oxalate | poor | |
| iron(III) | nitrate | 138 | |
| | sulfate | slight | |
| | oxalate | slight | |
| | chromate | decomposes | |
| lead(II) | acetate | 44 | |
| | nitrate | 54 | |
| | sulfate | ≈0 ☹ | |
| lead(IV) | acetate | “reversible | |
| | | hydrolysis” | |
| nickel | acetate | high? | |
| | nitrate | 94 | |
| | sulfate | 44 | |
| | formate | low? | |
| tin(II) | sulfate | 19 | |
| | nitrate | ? | |
| yttrium(III) | acetate | 9 | |
| | formate | 26 (at 50°) | |
| | nitrate | 123 | |
| | sulfate | 7 | |
| zinc | formate | 5.2 | |
| | nitrate | 98 | |
| | sulfate | 54 | |
| | acetate | 30 | |
| thorium(IV) | nitrate | 191 | |
| | sulfate | ≈0 ☹ | |
| uranyl | nitrate | 122 | |
| | sulfate | 21 | |
| | acetate | 8 | |
I can’t find any concrete information about ammonium aluminate; I suspect it doesn’t exist, although a number of chemical suppliers have it in their catalogs. Ammonium silicate apparently does exist but is too finicky to be of any practical use. Ammonium borate also seems to exist, but information about it is rare.
Tetraethyl orthosilicate is commonly used in a way similar to this to produce silica gel, but it is itself liquid rather than being water-soluble, and its decomposition is driven by exposure to water, not to heat.
Halogen complexes might be another thing to check out: titanium and zirconium both complex with halogens, and it may be possible to drive off the halogens with enough heat.
Glasses (frits) of metal oxides melt at lower temperatures; may be suitable fillers
Titanium, zirconium, aluminum, magnesium, chromium
Aluminum: nitrate (74%, decomposes at 150°) and sulfate (36%, decomposes below 900° to SO₃ and cubic γ-alumina) are highly soluble. Also occurs in soluble aluminates, but there is no aluminate of ammonia, so you can’t get alumina by calcining it; strontium aluminate is a glow-in-the-dark pigment and a refractory cement good to 2000°.
Chromium: ammonium dichromate is fairly soluble. Chromium(III) nitrate and especially sulfate are highly soluble; hexavalent chromium oxide too, but we don’t want that.
Boric acid is fairly soluble in water at 100°, nearly half as much at 50° (13% or so)
Calcium: nitrate is highly soluble, decomposes at 500°; soluble acetate releases acetone at 160° leaving carbonate; soluble formate decomposes at 300°, maybe to CaOH and CO, or like NaCOOH to an oxalate (CaC₂O₄, insoluble) and hydrogen (at 360° for Na?), then to a carbonate releasing carbon monoxide (at 290° for Na, 200° for calcium oxalate)? Calcium will precipitate lots of things including sulfate.
Magnesium acetate 53%; chromate 137%; formate 14%; nitrate 69.5%; sulfate 35%.
Zirconium: sulfate 52.5%. Nitrate has been successfully calcined to produce zirconia: https://pubs.acs.org/doi/10.1021/cm060883e
Titanium:
cobalt? vanadium? manganese? nickel? copper? zinc? tin? bismuth? strontium? barium? lithium?
Cobalt nitrate is 84% soluble in water; sulfate a bit less so.
Copper(II) nitrate is 83.5% soluble in water, substantially more than sulfate.
Ferrous ammonium sulfate is 27% soluble. Iron(II) nitrate 134%; sulfate 28%; iron(III) nitrate 138%.
Lead acetate 44%; lead(II) nitrate 54%. Lead(II) will precipitate sulfate.
Lithium acetate 40.8%; nitrate 70%; sulfate 34.8%; tartrate 27%.
Nickel acetate “easily soluble”, nitrate 94%; sulfate 44%; everything else pyrolyzable almost insoluble. (Its highly soluble chloride is not relevant.) Hexaamminenickel chloride is soluble in anhydrous ammonia and decomposes with heat, presumably to yield nickel.
Ammonium paratungstate pyrolyzes to tungsten trioxide at 600°, which is the soft mineral tungstite; the paratungstate ion has a tendency to precipitate from aqueous solution over time. There’s also a “metatungstate” oxyanion with 12 tungstens in it which is more soluble and stable in highly acidic solution.
Tin sulfate 19%; nitrate?
Yttrium: Yttrium(III) acetate 9%, nitrate 123%, sulfate 7%.
Zinc: formate 5.2%, nitrate 98%, sulfate 54%, acetate 30%.
Uranium, thorium
Thorium: thorium(IV) nitrate 191%, sulfate almost insoluble.
Uranium: Uranyl nitrate 122%, sulfate 21%, acetate 8%.
A common use for preceramic polymers, apart from the fibers and coatings mentioned earlier, is as polymeric binders for powdered ceramic — perhaps the same ceramic that the polymer will pyrolyze to. Filled systems like this have a variety of advantages:
The point about controlled composites bears further exploration. For example, pure amorphous carbon is quite weak, but if used in small quantities to cement iron filings, the composite would achieve significant strengths. Like cancellous bone, a porous composite made by pyrolyzing a binder between fully-dense whiskers of a ceramic will tend to be far more fracture-resistant than the same material if nonporous. A mixture of different kinds of particles can provide desirable combinations of properties unachievable in a homogeneous material, such as high surface hardness along with high crack resistance — again, like bone. Anisotropic filler orientation can provide anisotropic mechanical properties — again, like bone, or wood.
Ferromagnetic fillers like powdered iron can make a ferromagnetic bulk material with low electrical conductivity, but ceramic binders can maintain dimensions at different temperatures more precisely than the usual organic binders used for powdered-iron cores; also, iron’s Curie temperature is 770°, which many ceramics can withstand easily, but organic binders can’t even approach. (And cobalt’s is 1115°!)
The cheapest possible combinations would be sugar or flour with quartz sand or glass fiber, but at least in my low-temperature, poorly-controlled experiments (up to perhaps 400°–600°) the carbon resulting from sugar pyrolysis adheres very poorly to glass, represented by the glazing of stoneware, and to quartz sand. I could scrub it off easily with steel wool, and even scratch off some with a fingernail. Surface preparation, for example with plasma (perhaps in a fluidized bed), could conceivably improve the situation. Carbon should stick well to carbon fiber, though, and many things stick well to glass. And, as I said above, to my sorrow carbon sticks beautifully to zirconia.
in ammonia?
Forsterite, including peridot, is Mg₂SiO₄, while fayalite is Fe₂SiO₄; these are the endmembers of the olivine spectrum. Calcium cation substitutions also occur, modifying the structure and making it softer, going all the way to larnite, the belite of portland cement.
Titanium carbide? Zirconium carbide (3530°)? Tantalum carbide (3850+°)? Zirconium diboride? Gallium nitrate (soluble, decomposes to GaN in flowing ammonia at 500°–1050°)?