Some notes on sodium silicate.
Nowadays sodium silicate, or waterglass, is principally employed in foundries as a glue for sand-casting of metals, as a concrete sealant against water, and as a grouting agent to solidify soft soils prior to construction projects. Such composites can, at best, be several times stronger than ordinary concrete made with portland cement, and they don’t suffer from the grey discoloration of portland cement or, possibly, its carbon dioxide emissions. I’m interested in its possible uses for digital fabrication.
On Mercado Libre nowadays, companies like Geese Química are selling it for AR$140 per kg of “Silige” solution, which is US$1.13 at the current AR$124/US$1 price, and is probably about 400 g of sodium silicate, thus working out to about US$2.80/kg. This compares to AR$630 for 50 kg of portland cement, US$5.10, or 10.2¢/kg. Pure white portland goes for about 50% more, and hydraulic slaked lime is AR$220 for 20 kg, 3.5¢/kg. Portland cement is about 20% of the weight of the final concrete, and lime cement is about 25% of the weight of the final mortar, while for a similar strength sodium silicate can be 5% or less of the weight of the final solid; these numbers work out to 0.88¢/kg for lime concrete, 2.04¢/kg for portland concrete, and 14¢/kg for sodium-silicate-bonded concrete. The price of the aggregate closes the gap a little bit: construction sand costs about 5¢/kg and gravel costs about 3¢/kg, though both are usually sold by volume rather than weight. So the total materials cost might be 5¢/kg for lime concrete, 6¢/kg for portland concrete, or 20¢/kg for sodium-silicate concrete.
So, sodium-silicate-bonded concrete is about three or four times pricier than portland-cement-bonded concrete when they are the same strength. This probably explains why portland is widely used as a binder and waterglass is not. But I think waterglass may have some interesting advantages that can come into play with digital fabrication.
If simply allowed to dry, sodium silicate takes a substantial amount of time, and so it’s common to cure it with curing agents — in foundry practice typically CO₂ gas, which can harden it within a few seconds, but in other cases by mixing it with a curing agent, such as calcium chloride or calcium hydroxide.
A lot of the existing literature on using waterglass as a binder focuses on how to slow down the curing to minutes or hours, in order to give it a long “pot life”. But for digital fabrication, I think it might be more interesting to explore how to speed up the curing, ideally into the milliseconds to hundreds-of-milliseconds range. Then you could use it to “print” structures rapidly and with great freedom, without having to wait hours for each part of the structure to solidify before putting the next part in place. But is this feasible? How do we know the structures would be strong? Would it be resistant to weathering? What would it look like — would it suffer from the brutal, grim, gray appearance of typical portland concrete? Can you stick it to regular glass?
It turns out that they probably would be strong and resistant to weathering, and they can have a wide variety of appearances, from glass to sandstone and a variety of matte or glossy colors. The waterglass itself is transparent, although commonly a bit greenish due to iron contamination. And the possibility of structuring it at the millimeter scale under digital control should make it possible to achieve both stiffness and resilience dramatically better than that of traditional concrete.
High-water-content waterglass is used as an intumescent firestop — when heated above about 450°, the glass softens and its water expands to steam, converting the solid, transparent, glassy waterglass into a solid glassy opaque white foam.
Waterglass is commonly used in pottery as a deflocculant, reducing the viscosity of clay slips.
The tensile strength of waterglass-cemented composites can significantly exceed that of ordinary portland concrete, and it has been used as a binder for demanding applications like grinding wheels.
Chemical gardens grow in a waterglass medium; this suggests the speed with which waterglass can be solidified if exposed to the right reagents.
One crucial question here for construction purposes is whether waterglass can survive weathering — it’s no
The Keim company in Germany, founded by Adolf Wilhelm Keim, has sold a line of silicate-based “mineral paints” for over a century, and the Bleeck company in the UK has recently begun selling a similar line in the UK. Keim has expanded to the UK and USA. These paints are principally based on potassium silicate as a binder, which is very similar to sodium silicate, the principal difference being that solid potassium silicate can be conveniently redissolved in water at room temperature, while sodium silicate requires strong heating. (Some Keim paints instead use sodium aluminum silicate.) These paints are notable for their durability — 15 years is a common lifespan, but Keim claims that they have lasted over 130 years on the Stein Am Rhein building, and that, although “they will normally give 20–30 years satisfactory performance before redecoration is required,” it is also the case that “There are many examples of Keim Mineral Paints performing satisfactorily on lime render substrates for periods in excess of 100 years.”. I’m not sure whether these examples are interior or exterior.
Their Soldalit brochure claims, “Color shades will not change for decades,” and even recommends painting on top of acrylic or latex paint to protect it from weathering “for decades”; Soldalit, unlike their other paints, incorporates silica nanoparticles.
Wikipedia says, “The city hall in Schwyz and “Gasthaus Weißer Adler” in Stein am Rhein (both in Switzerland) received their coats of mineral paint in 1891, and facades in Oslo from 1895 or in Traunstein, Germany from 1891.”
Although sodium silicate itself is water-soluble and will thus redissolve in water, these paints “silicify” in contact with concrete or masonry, forming covalently-bonded water-insoluble hydrophobic products.
So all of this suggests that, in contact with calcite and quartz, these soluble silicates form insoluble materials that will weather at the rate of about the thickness of a coat of paint every 20 to 130 years. This compares favorably to portland cement.
Some sources talk about how calcium (hydr)oxide reacts slowly with waterglass because of its low solubility in water (1.7 g/ℓ), and magnesia (6.4 mg/ℓ), litharge (17 mg/ℓ), and minium (undetectably low) do not excel it in this, though Vail (see below) reports that they all cause “immediate precipitation”. If we want to speed it further, since the cations are apparently the active element here, more highly soluble salts might be preferred — calcium chloride (750 g/ℓ) is evidently standard, but other possibilities include magnesium chloride (540 g/ℓ); Epsom salts, magnesium sulfate (270 g/ℓ); Norwegian saltpeter, calcium nitrate (1200 g/ℓ); magnesium nitrate (710 g/ℓ); aluminum hydroxide (100 mg/ℓ); aluminum acetate (soluble); alums such as potassium aluminum sulfate (140 g/ℓ) or sodium aluminum sulfate (210 g/ℓ); and neat aluminum sulfate (360 g/ℓ). I’d rather not deal with salts of lead, barium, strontium, cobalt, and so on, although iron might be okay.
I guess these polyvalent cations displace the sodium cations, increasing the degree of connectedness of the waterglass and thus rapidly precipitating it. It took me an embarrassingly long time to figure this out. (I’m preeetty sure aluminum will work for this too.)
What would be super awesome for this would be getting boron to form soluble divalent or trivalent cations, but borate is of course an anion; boron really likes to make covalent bonds, and most of the compounds you’d hope would be soluble salts are instead found in List of highly toxic gases.
The various mineral species that ought to be formed include the following. The Mohs hardness of the minerals can be taken as some kind of indication of the strength of bonding in the material, but since the materials being formed here are actually amorphous, it is technically incorrect to refer to them as being these minerals; the amorphous glass will have different characteristics, including hardness, density, thermal behavior, and perhaps even color.
Calcium silicates: in the 2:1 Ca:Si ratio, this is the “belite” giving Portland cement its late strength, or “larnite” (Mohs hardness 6) in the wild. This is also called “lime olivine”, although properly speaking olivine varies from forsterite (Mg₂SiO₄, Mohs 7, including peridot, a refractory melting around 1900°) to fayalite (Fe₂SiO₄, Mohs 6.5–7). Halfway-lime olivine is the rare monticellite (CaMgSiO₄, Mohs 5.5). [Tricalcium silicate], with a 3:1 Ca:Si ratio, is alite, which I think is weaker and tends to revert to belite and lime; in the 1:1 Ca:Si ratio we have wollastonite (CaSiO₃, Mohs 4.5–5, melting at 1540°), noted for its whiteness and used as a filler in plastics, paint, and ceramics; it tends to form long acicular crystals when allowed to crystallize.
Magnesium silicate: as mentioned above, in the 2:1 Mg:Si ratio, this is forsterite olivine.
I worry somewhat about olivines’ vulnerability to weathering, since in an amorphous gel they will be even more exposed to reactions. But the way olivines weather is by incorporating water, as with iddingsite (Mohs 3). If hydroxyls are just incorporated into the olivine structure, you may get humite (Mohs 6–6.5), norbergite (Mohs 6–6.5), chondrodite (Mohs 6–6.5), and clinohumite (Mohs 6).
Manganese silicate: this is the heavy mineral tephroite, Mohs hardness 6, which exists in a continuum with forsterite and fayalite.
Aluminum silicate: this occurs naturally as topaz, Mohs hardness 8, although I’m not sure whether you can make topaz without fluorine, but also as several other minerals.
Topaz (Al₂SiO₄(OH,F)₂)has a 2:1 Al:Si ratio; other aluminum silicate minerals with the same ratio include andalusite, kyanite, and sillimanite, which are polymorphs of Al₂SiO₅. Kyanite, commonly used as a refractory, is the thermodynamically favored form at STP, and it's highly anisotropic, with Mohs hardness of 4.5–5 along one crystal axis and 6.5–7 perpendicular to it; it can be cooked into mullite and vitreous silica at 1100°. Sillimanite is Mohs 7 and andalusite, also commonly used as a refractory, is 6.5–7.5.
Kaolinite (Al₂Si₂O₅(OH)₄) has a 1:1 Al:Si ratio; it is a phyllosilicate clay, with almost negligible strength. Heating it above 550° converts it to metakaolin, a tranformation that is complete at 900°: Al₂Si₂O₇; this is used as an excellent pozzolan for pozzolanic cement, but it is still fragile. Further heating converts it into Si₃Al₄O₁₂ + SiO₂, quartz and a sort of spinel, above 950°; to platelet mullite 2(3 Al₂O₃ + 2 SiO₂) and cristobalite; at to acicular mullite (contaminated with the cristobalite) above 1400°, which remains solid up to 1840°.
Mullite itself — the key to the alchemists' famous Hessian crucibles — can also form at 3:2 or 2:1 ratios, but I suspect that isn’t what you’ll get by treating sodium silicate with aluminum salts.
Sodium silicate is a bit of a tricky beast to find good engineering data about, because it exists as a continuous spectrum between pure lye and pure fused silica, with a highly variable amount of water, and additionally can react with gases from the air as it hardens.
“Behavior of a sodium silicate grouted sand” by Gonzalez and Vipulanandan, 2007. Mixed “N-Sodium Silicate” (“Na₂SiO”(!!)·3H₂O) with “dimethyl ester” (which ester? “C₁₀H₁₀O₄” — clearly these are not organic chemists — “a byproduct of the nylon industry” — oh, apparently it’s a random mixture of succinate, “gluterate” (glutarate?), and adipate?) and injected it into “medium dense sand” to grout it in a mold. Compressive strength of the sand was 300–1900 kPa, Young’s modulus 200–500 MPa, but it had creep. No explanation is given as to why they thought adding dimethyl esters would be interesting, but apparently they sped up the gelling, maybe as a source of CO₂, but weakened the final product. Strain at failure was 0.4%–2%. No samples without DME were included. No tensile or flexural strengths were recorded, I guess because they were interested in grouting sands for civil engineering purposes.
I have zero faith in Gonzalez and Vipulanandan; the formula they give for “sodium silicate” would actually be a metallic silicon-sodium alloy which would be at the very least violently reactive with water and possibly pyrophoric. The absence of a DME-free control is particularly glaring (for my purposes) and they don’t talk at all about their CO₂-control measures.
“Nanoindentation and Brillouin light scattering studies of elastic moduli of sodium silicate glasses” by Zhao et al., 2011. Talks about a “large discrepancy” in Young’s modulus measured by different methods (and offers an explanation). They prepared their sodium silicate with varying amounts of sodium (8, 20, 30, and 40 mol%) from Na₂CO₃ and SiO₂ (presumably crystalline) mixed in an agate pestle and then melted at 1500° or, for the 8mol%-Na glass, 1700°, and compared to fused silica. The idea is, I guess, that the sodium carbonate converts to Na₂O when you heat it up.
A thing I’m not clear about with these mole percentages is whether the metals are 8 mol% Na — thus, two sodium atoms per 23 silicon atoms — or whether the oxides are 8 mol% Na₂O — thus, two Na₂O units per 23 SiO₂ units, and therefore four sodium atoms per 23 silicon atoms. I’m pretty sure it isn’t two sodium atoms per 23 silicon or oxygen atoms.
Astonishingly, they got plastic deformation out of the glasses by indenting it with a diamond-tipped “Hysitron TI 900 TriboIndenter”, which they then measured with an AFM. The whole methods section of the paper is equipment porno.
They got a 72 GPa Young’s modulus for fused quartz with all four measurement methods, down to 67 GPa at 8%, 61 GPa at 20%, 61 GPa at 30%, and about 59 GPa at 40%. There’s a bunch of stuff in there about correcting the figures because at the higher sodium contents they give significantly different results, up to 64 GPa for nanoindentation for the 40%.
They also give a “hardness” value in GPa, ranging from 8 GPa for fused quartz down to 4–4.5 GPa for the 40% sodium glass. I’m guessing that this is the compressive yield stress, although I am surprised to learn that these glasses have a yield stress; I thought they would just deform elastically until they broke. But I guess in a small enough area you wouldn’t have enough energy to propagate a crack, and so even if the glass there powdered, you’d squish it back into the glass surface (“indentation-induced densification”, although it’s not clear that there was any powdering going on). I don’t know. The AFM images make it look pretty fucking rough, and in the glasses with larger amounts of sodium, there’s a “pile-up” of plastically deformed material around the outside of the four-micron-wide triangular craters. But in the lower-sodium glasses, the surface is totally flat outside the craters.
No tensile strength figures are given.
“The Effect of Microstructure on the Physical Properties of Glasses in the Sodium Silicate System”, by Redwine and Field 1967. It’s not a survey paper — it focuses on changes in physical properties that can be obtained by heat-treating glasses within a metastably-miscible concentration range — but it still gives a broader overview of the field. It gives values of Young’s modulus E from 8.38–9.36 million psi (57.8–64.5 GPa in non-medieval units) depending on temperature, composition, and heat treatment, as well as measured values of shear modulus G (25–27 GPa), bulk modulus B (33–36 GPa), and Poisson’s ratio μ (0.18–0.20). Linear TCE ranged from 4.64 ppm/° to 10.15 ppm/°. No strength of any kind is measured. Most of the paper is concerned with how these vary by temperature.
They don’t seem to say how they made the glasses.
It suggests that at low temperatures Na₂O and SiO₂ are miscible at below about 77 mol% Si₂O and above about 97 mol% SiO₂, but between these limits there is a regime where the two materials spontaneously separate into different phases, presumably a sodium-rich phase and a silicon-rich phase. This immiscibility persists up to about 825°, above which they are miscible in all proportions. (The plot only goes down to 500°, though, perhaps because below that temperature the separation processes are too slow to observe.)
Mostly they focus on glasses of 7.2 mol% to 18.4 mol% Na₂O, which is to say, between 92.8 mol% SiO₂ and 81.6 mol% Si₂O, thus covering much of the range where this immiscibility occurs. Within the “unstable” region, they report that heat treatment resulted in phase separation into “two independently interconnected phases”, while in the “metastable” region it resulted in “classical nucleation and growth of particles”.
(Interestingly, the miscibility limit in this paper seems close to the “pile-up” limit displayed in Zhao 2011 above. This might be a coincidence.)
It might be interesting to see if laser heat treatment could induce this “heat treatment” effect in very small areas very quickly, as a way of writing data; for compositions right in the middle of the “unstable” region, say around 11 mol% Na₂O, the separation might be fastest. However, in the paper, they heat-treated for 1½ hours at 770° to get phase separation at 12.6 mol% Na₂O, so that might be very challenging. However, they noted that they were not able to obtain homogeneous glasses for some compositions, presumably because they could not cool them fast enough.
They measured the “dilatometric softening point” of the glasses from 500° for the highest-sodium variants (18.4 mol%) up to 735° for a heat-treated high-silica glass (7.2 mol% Na); this is the temperature at which heating the glasses does not dilate your dilatometer any further because the viscosity is low enough that it flows instead, which is of course dependent on how much force the dilatometer is clamping with.
The linear coefficients of thermal expansion (αRT-350)ranged from 4.64 ppm/° for heat-treated 7.2-mol% Na glass up to 10.15 ppm/° for 18.4-mol% Na without heat treatment, varying linearly. These numbers barely changed with heat treatment.
“Dynamic Fatigue of Sodium-Silicate Glasses With High Water Content”, by Ito and Tomozawa, 1982. These guys were also at RPI. They measured 40–70 GPa Young’s modulus for dry sodium silicate and 3–50 GPa for glasses including a lot of water. They also measured its tensile strength but I can’t understand their results.
They slowly (over several days) dried out some commercial sodium silicate solution (8.9 wt% Na₂O, 28.7 wt% SiO₂, Na₂O·3.3SiO₂, which I guess is 23.2 mol% Na₂O) to various water contents around 25%, at which point it was solid; they sliced it into 1.7-mm-thick slips and and used four-point bending to measure its flexural strength, finding a strong dependence on speed of loading especially for higher-water-content glasses, which also had the highest Young’s modulus, which was, insanely, viscoelastic.
Unfortunately the Y-axis labels on the fracture strength plots are very difficult to understand: it says “Log Fracture Strength (kg/mm²)”, which is already ambiguous (is that a base-10 log or base-e?) but to worsen the situation, a legend helpfully explains: “log σ = (1/(n+1)) log σ̇ + log C”, only without the parentheses. Is that an empirical approximation formula or does it explain how the plotted numbers were derived? The numbers plotted, at any rate, range from about -0.1 to about 1.2, with the strongest glass typically being the one with 15.9% water, which is slightly stronger than the dry glass. If we suppose that this is a base-10 logarithm of the flexural strength, then we have a tensile strength of about 0.8–16 kg/mm², or 8–160 MPa in modern units. But I am not confident in that interpretation.
The Young’s-modulus plot in Fig. 4 is, by contrast, decently labeled — it uses a logarithmic Y-axis but with ticks labeled in real units. It gives 4–7 thousand kg/mm² (40–70 GPa) for the dry glass, with numbers ranging from 0.3–5 (3–50 GPa) for the wet glasses.
Their figure 5 also plots Young’s modulus, a theoretical Young’s modulus limit at infinite stress rate, which is some three orders of magnitude lower, ranging from 1 kg/mm² to 5.5 kg/mm². I suspect they have mislabeled their plot.
They also plotted the Knoop hardness of the samples, in the range 50–400 kg/mm² (500–4000 MPa), decreasing with higher water content.
They cite “McMillan (1982)” as giving flexural strengths for soda-lime silica glass, which looks like a paper in “Non-Crystalline Solids” by McMillan and Chelebik, 1980, I think volume 38/39, p. 509. I think that’s actually Chlebik, and the paper is perhaps “The effect of hydroxyl ion content on the mechanical and other properties of soda-lime-silica glass”. But it seems like probably that paper doesn’t cover soda-silica glass. (And they didn’t say it did, after all.)
This article has the deeply misleading title, “Water Glass as Hydrophobic and Flame Retardant Additive for Natural Fibre Reinforced Composites,” by Medina and Schledjewski, 2009. I say “deeply misleading” because waterglass is preeettty faaar from being hydrophobic! As noted above, drying the stuff out is really tough.
The article has a lot of problems like that. It describes a Si(OH)₄ moiety as “silane”, talks about "natural fibers" as if they're all equivalent of (I was assuming cellulose because the descriptions they give don’t fit chitin, keratin, asbestos, etc., but even if it’s cellulose not all cellulose is the same — finally on page 3 we find out that the fiber they tested is 70% kenaf, 30% hemp, with no source given), never describes which acrylic resin it’s using (I think, although sometimes it mentions “polyester”, so maybe it’s a polyester acrylic — although on page 8 they finally slip up and admit that it’s one of the Acrodurs, whose composition is apparently secret), never describes how much sodium is in the waterglass it’s using, uses a very crude flammability test, etc., etc.
But it’s pretty interesting. Apparently they glued together some cellulose fiber mats with various mixtures of sodium-silicate waterglass and the unspecified acrylic resin, and got some decent boards out of it, and of course the waterglass made them flame retardant.
Because of the amount of crucial data omitted, apparently intentionally (“a new water glass type specially developed as hydrophobic additive for acrylic systems”), the paper falls far short of basic reproducibility criteria.
The low-sodium endmember of the sodium silicate continuum is fused quartz, and that’s the most highly polymerized part, so we would expect all sodium silicates to have tensile strength and hardness at most that of fused quartz.
http://www.quartz.com/gedata.html agrees with https://technicalglass.com/technical_properties/ on the curiously precise tensile-strength number of 48 MPa. Marijuana paraphernalia merchant https://highlyeducatedti.com/blogs/information/thermal-shock-vs-tensile-strength gives 67 MPa for flexural strength and 50 MPa for ultimate tensile strength, apparently quoting makeitfrom. It also gives 0.5 ppm/° linear TCE.
“Studies on the Possibility of More Effective Use of Water Glass Thanks to Application of Selected Methods of Hardening”, by Stachowicz, Granat, and Nowak, 2010. They say that waterglass-bound foundry casting sand commonly has tensile strengths (RₘU) in the 0.3–0.5 MPa range; with 5% waterglass in their sand they got tensile strengths as high as 3.6 MPa, with higher-sodium waterglasses generally giving stronger bonds.
They’re concerned with binding foundry sand with small amounts (1.5–5.0%) of waterglass, and in particular with whether microwave heating can make it stronger and maybe allow you to use less than the usual minimum of 2.5%, which it apparently does. Also they were able to microwave their samples for four minutes instead of oven-drying them for two hours.
It has a helpful table of waterglass grades used in foundries, with molar ratios of SiO₂ to Na₂O anging from 3.2:3.4 (grade 137) to 1.9:2.1 (grade 150).
I’m not sure whether their 1.5% and 5% etc. refer to the weight of the dried waterglass or to its wet weight. (Grade 137 is 35% solids, with the rest being water, while the very viscous grade 140 is 42.5% solids.) Anyway, the strength continues to increase quite linearly up to the 5% they tested, which makes me optimistic that strengths several times higher are feasible with higher binder content.
The linear extrapolation of the 1.5%–5% suggests a tensile strength of something like 50–70 MPa for solid 100% waterglass, which is consonant with my tentative 8–160 MPa interpretation of Ito 1982 and the 50–70 MPa numbers given above for fused quartz.
Carbon dioxide is not mentioned.
All nine entries in their bibliography are Polish.
“Silicate Bonding of Inorganic Materials, Part I”, by MacKenzie et al., 1991.
XXX
“Soluble Silicates: Their Properties and Uses”, Vail, 1952. This is a thousand-page two-volume set full of valuable information.
It mentions that a major use of waterglass in the mid-1800s was “the hardening of stone to increase its weather resistance”, further allaying my concerns about weathering, and it has a whole section on using it to bond grinding wheels. It mentions that Feuchtwanger claims to have introduced the use of waterglass in the US, using it to prevent rusting of naval weaponry.
It seems that when Vail wrote his book, sodium silicate was considerably more widely used than it is today: “There are few manufacturing plants which do not make some use of [soluble silicates].” Today I think it’s kind of a niche product, despite the growing importance of avoiding phosphate runoff (silicates can substitute for phosphates as detergents). This consideration does not appear in the introductory section, although it does talk about how conservation may stimulate the use of silicates in the future.
With respect to the prospect of precipitating or “curing” waterglass, Chapter 2 (“Present Practices”) begins wih the promising note: “Most of the impurities likely to be found in sand form insoluble silicates, and even small quantities, less than one per cent, can create serious difficulties.” It has the appealing note that the old way of making it was “dissolving diatomaceous earth in caustic liquors”, which does sound much easier than the standard approach of heating sulfate or carbonate of soda to some 700° to 800° in contact with sand. On the other hand, the standard approach is considerably more legal in Argentina.
It explains that the “so-called neutral glass”, usually “pale bluish or greenish”, is 1:3.3 Na₂O:SiO₂, although IIRC the pH of the solution is still above 11, while the “alkaline” is 1:2.1. This probably explains why the pale greenish bottle I have doesn't burn my skin and was sold as “neutral”.
Astoundingly, at this time it was still not known that solid waterglass, or indeed any solid, was amorphous! Vail says the question “might be of more academic than practical value”, though he also said, “A sodium silicate is as nearly devoid of ordered structure as any known material.”
It explains that finely divided dry waterglass sometimes does get dissolved in water at atmospheric pressure 100°, but to dissolve lumps of glass, 90–100 “pounds gage” steam pressure is used (psig I guess, so 700–800 kPa absolute).
It explains that the reason sodium silicate has eclipsed potassium silicate is just that sodium is cheaper than potassium.
I find this unjustifiably amusing: “Immediately after use, hydrometers should be washed thoroughly with warm water until alkali cannot be tasted on the glass...” — clearly a pre-OSHA book.
He points out that you can blow waterglasses just like you can blow other glasses, but that it can contain varying amounts of water “without substantially altering their appearance”. This makes me wonder if they might be a particularly suitable material to attempt to 3-D print graded-index optics in.
It explains that alcohol precipitates waterglass just by removing water, which I had suspected but was not sure of. Also, he mentions doing the same with alkali metal salts or ammonia.
It includes the oldest citation I've seen: “A sodium silicate glaze is described in cuneiform records of the reign of Ashurbanipal, 668–626 B.C.: 10 mana of sand, 10 mana of alkali ash, and 1.67 mana of styrax gum were heated to white heat, cooled, crushed, and placed in a clean melting pot in a cold furnace.”
A surprising thing mentioned a couple of times in the book is that potassium silicate does not effloresce, while sodium silicate does, a fact particularly relevant for production of fake stone; this afflicted Ransome’s fake stone in 1861.
A technique frequently mentioned both in this book and in Keim's paint brochures is the inclusion of amorphous silica particles in the liquid — a sol of precipitated silica gel particles, for example, although diatomaceous earth should also work. This reduces the amount of the waterglass that must be gelled to form a solid gel, since the particles form part of the gel network. Other effects include thickening the liquid and making it colloidal and possibly thixotropic.
In Chapter 5, Vail refers to “immediate precipitation which occurs when calcium, magnesium, or lead oxides are mixed with concentrated silicate solutions”, although it's not clear what timescale he’s talking about.