The Glass Transition and the Viscosity Curve

The Elbow in the Cooling Curve

In guide 1 you met a glass as a liquid caught mid-freeze: cool a melt fast enough to sneak past crystallization and you get a supercooled liquid that keeps thickening until it is, for all practical purposes, a solid — glass formation by dodging the crystal. This guide zooms in on the exact moment that liquid stops behaving like a liquid and starts behaving like a solid. That moment has a name and a temperature: the glass transition, at the glass transition temperature Tg.

The cleanest way to see Tg is a plot of volume against temperature as you cool. A crystal follows a gentle downhill line as it contracts, then at its melting point Tm the line drops off a cliff — the atoms snap into a dense ordered lattice all at once, a sharp, fixed transition. A glass does something quite different. Cooled past Tm, it ignores the cliff entirely: the supercooled liquid keeps sliding down the same steep liquid line, contracting freely. Then, over a narrow band of temperature, the line quietly bends to a shallower slope — the slope of a solid. That bend is Tg. It is an elbow, not a cliff.

Tg Is Kinetic, Not a Fixed Point

Here is the single most important honest fact about the glass transition. A melting point Tm is a fixed thermodynamic constant — ice melts at 0 degrees C whether you warm it in one second or over a week. Tg is nothing of the sort. As the supercooled liquid thickens on cooling, its atoms need more and more time to shuffle into their equilibrium arrangement. Tg is simply the temperature where they run out of time: the network becomes so sluggish that, on the timescale of your experiment, the atoms freeze in place mid-shuffle. The glass transition is a kinetic event — a traffic jam — not a true phase change.

Because it is a race between cooling and rearranging, Tg depends on how fast you cool. Cool quickly and the atoms run out of time sooner, at a higher temperature — Tg rises. Cool slowly and they keep up longer, so Tg drops, by very roughly a few degrees for every tenfold change in cooling rate. So Tg is not one sharp number but a short band, and it belongs as much to the process as to the glass. That single fact — a transition temperature you can move just by changing the clock — is what 'kinetic, not thermodynamic' really means.

Viscosity: The One Number That Runs Everything

What is actually changing across all of this is viscosity — a fluid's resistance to flow, measured in pascal-seconds (Pa.s; one Pa.s equals ten poise, the older unit). Water is a runny 10^-3 Pa.s; honey is around 10 Pa.s; a hot glass melt sits near 10 Pa.s too. As that melt cools toward room temperature, its viscosity does not jump — it climbs, smoothly and relentlessly, through more than fifteen orders of magnitude. That astonishing, continuous climb, with no break at any melting point, is the whole physical story of a glass.

This is why glassblowers talk about viscosity, not temperature. A metal has a sharp melting point, so you work it by temperature; a glass has no such point, so you work it by viscosity. Every operation lives at a characteristic viscosity: you gather a workable gob at one value, blow it at another, let it set at a third. Tg itself turns out to sit at a fixed viscosity — right around 10^12 Pa.s — so the glass transition and a particular stiffness of the melt are two names for the same thing. Land the glass on the right rung of the viscosity ladder and it will do what you want.

VISCOSITY LADDER of a soda-lime glass   (viscosity in Pa.s)

  log10(Pa.s)  named point       what the glass does         T ~ (deg C)
  -----------  ----------------  --------------------------  ----------
      ~1       melting point     thin, pourable melt            1450
      ~3       working point     gather + shape a gob           1000  \
    3 - 6.6    WORKING RANGE     press, blow, draw, cast     1000-750  | form
      ~6.6     softening point   a rod sags under its weight     700  /
     ~12       annealing point   rigid, yet stress relaxes       540  <- ~Tg
     ~13.5     strain point      below here = a solid            505
     ~19       room temperature  NO flow -- ever                  25

  for scale:  water ~10^-3     honey ~10     cold pitch ~10^8

The Fixed Points: Working, Softening, Annealing, Strain

Read the ladder from warm to cool. At the working point, about 10^3 Pa.s, the glass has the stiffness of thick taffy — soft enough to gather, press, draw, or blow, and the top of the working range where all forming is done. Keep cooling and you reach the softening point near 10^6.6 Pa.s (fixed by a standard test: a thin rod begins to sag under its own weight), where the shape sets. Far colder is the annealing point at about 10^12 Pa.s, where the glass is a rigid solid to the eye yet can still creep just enough to relax internal stress in about fifteen minutes. Just below sits the strain point, near 10^13.5 Pa.s — below it, flow is so slow the glass is a solid for every practical purpose.

One more idea makes the curve practical: its steepness. A glass whose viscosity changes gently with temperature is called a long glass — it stays workable over a wide span, forgiving and easy to hand-work. A glass whose viscosity shoots up over a narrow span is a short glass — it sets fast and leaves the worker little time. Soda-lime glass is relatively long, part of why it is so friendly for bottles and hand-blown ware; some borosilicate compositions are shorter and less forgiving. Long or short is nothing but the slope of the line you just read.

1. Melt and fine the batch at very low viscosity (around 10 Pa.s, roughly 1450 degrees C) so bubbles rise out and the liquid becomes uniform.
2. Gather a gob at the working point (about 10^3 Pa.s) — stiff enough to hold together, soft enough to shape.
3. Form it through the working range by pressing, blowing, drawing, or casting before it stiffens.
4. Let the shape set as it cools past the softening point (about 10^6.6 Pa.s); the part now holds its form.
5. Hold near the annealing point (about 10^12 Pa.s) so residual stress relaxes, then cool slowly through the strain point so none freezes back in.
6. Below the strain point, cool freely — the glass is now a rigid solid and no fresh stress can form.

Two Everyday Glasses, Two Curves

Composition slides the whole curve left or right, and the two glasses in your kitchen prove it. Soda-lime-silica glass — the stuff of windows, jars, and bottles, and about 90 percent of all glass made — is mostly silica with soda (Na2O) and lime (CaO) stirred in. Those additives are network modifiers: they snap the silica network's bridges into dangling non-bridging oxygens (the subject of guides 3 and 4), which loosens the melt and pulls the entire viscosity curve down to gentle temperatures — a softening point near 700 degrees C. Cheap, easy to melt, easy to work.

Borosilicate glass — Pyrex, and every lab beaker — swaps much of that modifier for extra network former (B2O3 alongside the SiO2), so its whole curve sits shifted to higher temperature, with a softening point up near 820 degrees C. The prize is a very low thermal expansion coefficient, about 3.3 x 10^-6 per degree C against soda-lime's roughly 9 x 10^-6 — which is exactly why it shrugs off the thermal shock of going straight from a hot oven to a cool counter. Push this to the limit and you reach fused silica, nearly pure SiO2: expansion around 0.5 x 10^-6 per degree C, but a curve so high it must be worked above 2000 degrees C.

Reading Stress off the Curve — and What Comes Next

The annealing and strain points are not academic — they are where a factory saves its glass from cracking. When a formed part cools, its surface passes Tg and freezes solid while the inside is still a soft liquid; the two then fight as the interior contracts, locking in residual stress that can split the part days later. The cure is annealing: hold the whole part at the annealing point (about 10^12 Pa.s) so slow viscous flow erases the stress, then cool it gently through the strain point so no fresh stress can freeze back in. The two viscosity points you just met literally bracket the safe annealing window.

Turn the same physics around and you get the rest of this rung. Thermal tempering deliberately quenches the surface into permanent compression, building a protective skin that makes the glass far stronger — the safety glass in a car window. Linger too long in the temperature band where crystals want to grow and you get devitrification, the unwanted crystallization that ruins a glass; steer that same crystallization on purpose, seeded by a nucleating agent, and you get a glass-ceramic, strong and nearly zero-expansion. Guide 5 walks all three; they all live on the viscosity curve you now own.

One last honest point, because it is the most repeated myth about glass. Old cathedral windows are thicker at the bottom because of how they were made, not because the glass has flowed. At room temperature a common glass sits at a viscosity of roughly 10^19 Pa.s or more — so stiff that it would take far longer than the age of the universe to sag by a hair's width. Below its strain point a glass is not a slow liquid in any meaningful sense; it is, for every purpose you will ever care about, a rigid solid.