The Master Variable: Viscosity from Melt to Rigid Solid
The four guides before this one taught you what a glass is: a liquid caught mid-freeze. Cool a melt fast enough to dodge crystallization and, at the glass transition temperature Tg, the supercooled liquid stops flowing and locks into a rigid, disordered solid — a kinetic freeze, not a sharp melting point. You met Zachariasen's random network, the formers that build it and the modifiers that snip its bridges. This last guide turns all of that into a craft: how we cool, stress, strengthen, and even deliberately crystallize glass, using one dial that runs the whole show — viscosity.
Viscosity is a glass's resistance to flow, and across the working range it changes not by a little but by a factor of a trillion. A molten batch pours like honey; the same glass at room temperature is stiffer than steel. Because there is no single melting point to aim at, glassmakers navigate by fixed points on the viscosity curve — temperatures defined not by what the glass looks like but by the exact viscosity it reaches. Four matter most: the working point (soft enough to blow and press), the softening point (sags under its own weight), the annealing point (internal stress relaxes in minutes), and the strain point (below it the glass is effectively rigid). Notice that Tg sits right around the annealing point — the same freeze, read two ways.
VISCOSITY (Pa s) FIXED POINT WHAT HAPPENS THERE soda-lime log scale (~deg C) ------------------------------------------------------------------------- 10^1 runny | melt melt & homogenize ~1400 10^3 | WORKING point blow, press, draw, mould ~1000 10^6.6 | SOFTENING point sags under its own weight ~700 10^12 | ANNEALING point stress relaxes in minutes ~550 ~Tg 10^13.5 | STRAIN point below here it is rigid ~510 10^18+ stiff v room-temp glass a true solid; no real flow 25 ------------------------------------------------------------------------- up = hotter & runnier (low viscosity) down = cooler & stiffer (high viscosity)
Residual Stress and Annealing
Here is a problem hiding inside every hot glass object: it cannot cool evenly. The surface, touching cooler air, chills and stiffens first while the inside is still hot and mobile. Later the interior cools and tries to contract, but the surface has already frozen solid and will not let it — so the finished piece is left with the outside squeezed and the inside stretched, a permanent tug-of-war frozen in place. This locked-in thermal stress, called residual stress, is dangerous: it adds to any load the piece carries in service, and a thick, badly cooled object can crack on the shelf days later or fly apart at the first scratch.
The cure is annealing, and the viscosity ladder tells you exactly how to do it. Reheat the piece to its annealing point — the temperature where viscosity is about 10^12 Pa s — and hold. There the glass is still just fluid enough that atoms shuffle and the internal stresses quietly relax away in minutes, yet not so soft that the shape sags. Then cool slowly, especially through the strain point (viscosity ~10^13.5 Pa s), below which the glass is too stiff to relax any further — so any temperature gradient you allow down there would freeze fresh stress right back in. Factory glass rides a long, temperature-controlled tunnel oven called a lehr that does exactly this: soak at the annealing point, then cool gently through the strain point.
So annealing has one goal: to erase residual stress and leave the glass relaxed and even. It is the gentle, careful cooling that makes glass safe and predictable. But there is a bolder idea lurking here. If cooling a glass unevenly is what puts stress in, what if we deliberately cooled it unevenly, and left the stress in on purpose — arranged so that it protects the glass instead of endangering it? That is thermal tempering, and it is the exact opposite of annealing.
Thermal Tempering: Freezing In a Compressive Skin
Thermal tempering takes the very mechanism annealing fights and weaponizes it. Heat a finished pane above its softening point until it is uniformly soft, then blast both faces with jets of cold air. The surfaces chill and freeze rigid at once while the core is still hot and fluid. Then, as that trapped hot core slowly cools and contracts, it drags the already-solid surfaces inward — pulling the skin into permanent compression and leaving the core in balancing tension. The pane ends up wrapped in a squeezed outer layer, typically stressed to 100 MPa or more, over a stretched interior.
Why does a squeezed skin make glass strong? Recall from the fracture rung that glass fails from surface flaws: a Griffith flaw is a nick at the edge of a sheet of paper, and it only tears open under tension. The Griffith criterion says a crack grows when tension pulls its tip apart. A compressive skin clamps every surface flaw shut, so an applied bending load must first spend itself cancelling that built-in compression before any flaw feels a scrap of tension. That is why tempered glass is four to five times stronger than the same annealed pane — not because the glass is different, but because its flaws are held closed. And when it finally does fail, the stored energy dices it into a shower of small, blunt cubes instead of long daggers, which is why it is the safety glass in car side windows, shower doors, and phone screens.
Tempering by air-quench needs a fairly thick pane; it cannot toughen the paper-thin glass of a phone. Its cousin, chemical strengthening, builds the same compressive skin by chemistry instead of heat. Bathe a sodium-bearing glass in a molten potassium salt held below its strain point, and the small Na+ ions in the surface trade places with the larger K+ ions from the bath. The oversized newcomers are crammed into holes meant for smaller ions, and that crowding squeezes the surface into compression — often 700 MPa or more, far higher than air-tempering reaches. This is how the tough cover glass on phones is made. The honest catch: the compressed layer is thin, only tens of microns deep, so a scratch or drop that drives a flaw past it still finds the tension underneath and breaks.
Two Everyday Glasses: Soda-Lime and Borosilicate
Nearly all the glass you touch — windows, bottles, jars, drinking tumblers — is soda-lime-silica glass, about 72% SiO2, 14% Na2O, and 10% CaO by weight. Pure silica makes a magnificent glass but has to be worked near 2000 degrees C, which is ruinously expensive. The Na2O is there as a flux: as guide three explained, each Na+ is a modifier that snaps a Si-O-Si bridge into two non-bridging oxygens, cutting the network into shorter pieces so the melt softens and can be worked near a practical 1000 degrees C. But soda alone leaves a glass so full of broken bridges that water slowly attacks it — the stuff of 'water glass'. The lime, CaO, is the stabilizer that ties the network back together enough to resist water. Cheap, clear, and easily worked, soda-lime's one real weakness is a high thermal expansion of about 9 x 10^-6 per degrees C.
Pour boiling water into a cold soda-lime tumbler and it may crack — the sudden heat expands the inner surface while the outer stays put, and the mismatch tears it apart. The fix is borosilicate glass (Pyrex, laboratory ware), where boric oxide, B2O3, joins SiO2 as a second network former and the modifier content is cut right back. The result is a tighter, better-bridged network with a thermal expansion near 3.3 x 10^-6 per degrees C, roughly a third of soda-lime's. At the far extreme, fused silica — pure SiO2 with no modifiers at all — expands a mere 0.5 x 10^-6 per degrees C, so a red-hot piece can be plunged into water unharmed; that near-perfect network is also exactly why it costs so much to melt and shape.
The through-line is simple: low expansion means low thermal shock. When one part of a piece is heated it wants to expand; if the rest holds it back, the difference becomes expansion-driven tensile stress — and tension is exactly what cracks a flawed glass. Shrink the expansion coefficient and the same temperature jolt makes a far smaller stress. That is the whole reason a borosilicate dish goes from oven to counter unfazed while a soda-lime one shatters. Be honest, though: low expansion only raises the threshold, it does not grant immunity. Hit any glass with a big enough or fast enough temperature change and the stress will still find its worst flaw.
Devitrification and Its Controlled Cousin, the Glass-Ceramic
Glass is a liquid that got away without crystallizing — but it never entirely gives up the urge. Hold a glass too long at the wrong temperature, in the danger zone between Tg and the melting point where crystals can both nucleate and grow, and it will slowly crystallize after all. This unwanted crystallization is devitrification (literally 'un-glassing'). It is a working hazard: the growing crystals cloud the once-clear glass, and because they usually have a different thermal expansion from the surrounding glass, they build in stress and can crack the piece. A glassblower keeps the work moving through that danger band precisely to avoid it.
Now flip the hazard into a triumph. What if, instead of letting a few large crystals grow at random and ruin the glass, we forced a huge number of tiny crystals to grow everywhere at once, uniformly and under control? The result would not be a spoiled glass but a new material: a glass-ceramic, a dense, fine-grained polycrystalline solid shaped as an easy-flowing glass and then crystallized on purpose. The trick is a nucleating agent — a pinch of TiO2, ZrO2, or fine platinum dissolved into the melt — that makes crystals sprout at millions of sites throughout the body rather than at a scattered few. Seed the whole volume densely and every crystal stays tiny, because its neighbours crowd it, giving an extremely fine grain size.
- Melt and form as a glass. Melt the batch — nucleating agent and all — and shape it while it is still a workable glass: blow it, press it, roll it into sheet. This is the easy, pore-free forming that powder-and-fire pottery can only envy.
- Nucleate. Reheat and hold just above Tg, at the nucleation temperature, so the nucleating agent seeds a dense, even population of crystal nuclei throughout the whole body — the goal here is many seeds, not big ones.
- Grow. Raise to a higher crystal-growth temperature (still below the softening point) and hold, letting fine crystals grow on those countless seeds until most of the glass has converted to an interlocked, fine-grained crystalline solid.
- Cool. Because it began as a pore-free glass and crystallized in place, the finished glass-ceramic is dense and flaw-poor, and its properties are tuned by which crystals you chose to grow.
The payoff can be spectacular. The most famous family, the lithium aluminosilicate glass-ceramics, grow crystals of beta-quartz or beta-spodumene that actually shrink slightly when heated. That negative expansion can be tuned to almost exactly cancel the positive expansion of the leftover glass, giving a net thermal expansion near zero — a material that barely notices temperature at all. This is the glass-ceramic behind the black cooktop panels you can heat red-hot and splash with cold water, and the ultra-stable blanks of giant telescope mirrors. Two honest closing notes: a glass-ceramic is never fully crystalline, since a little residual glass always remains between the grains; and none of this follows what an equilibrium phase diagram alone would predict, because the whole art lies in controlling the kinetics of nucleation and growth. Turning glass's own crystallizing urge from a defect into a design tool is a fitting place to close the glassy-state rung — and it hands you the idea of nucleation and growth to carry forward into the firing and sintering rungs ahead.