Thermal Shock and Surviving Temperature Change

The crack that comes from a temperature change

You already have every piece you need for this from the last two guides, so let us assemble them fast. In the expansion guide you saw that a material forced to change temperature while it is not free to change size pays for it in thermal stress — clamp a bar and heat it and the stress it feels is roughly E times alpha times the temperature change, where E is the stiffness and alpha is the expansion coefficient. Thermal shock is simply what happens when that stress arrives suddenly and unevenly, because a temperature change cannot reach every part of a body at once. Splash cold water on a hot glass dish and the surface chills and tries to shrink in an instant, but the still-hot interior underneath holds it stretched out — so the surface is yanked into tension, and if it is brittle enough, it cracks.

So the whole mechanism is a chain you can recite: fast temperature change makes a temperature gradient (hot inside, cold surface), the gradient makes different parts want to be different sizes, and "want to be different sizes but stuck together" is exactly the recipe for internal stress. The key move is to see the body fighting itself — no external clamp is needed, because the hot core is the clamp on the cold skin. The steeper the gradient and the bigger the mismatch it forces, the bigger the stress; when the local stress beats the material's strength, a crack starts. Thermal shock is a temperature story on the outside and a stress story on the inside.

Why it is a ceramic's problem, not a metal's

Here is the honest heart of the matter: thermal shock is really a fracture problem, and it hunts brittle materials. Take a metal through the same brutal quench and the surface stress climbs, hits the metal's yield strength, and then the metal simply yields — it deforms plastically, its dislocations glide, and the stress is bled off harmlessly as a permanent little strain. The metal shrugs. A ceramic has no such escape valve: it cannot glide dislocations at room temperature, so it has no way to relax the stress except to keep loading up elastically until, at the worst flaw, a crack runs and the piece fails by brittle fracture. The very property that makes ceramics so useful at high temperature — that they do not go soft and flow — is exactly what makes them shatter under sudden temperature change.

Do not conclude that metals are immune, only that they fail differently and more forgivingly. A metal that is cycled hot-cold-hot-cold thousands of times can still crack, but that is thermal fatigue — a slow, cumulative cousin of the fatigue you met in the failure rung, where each cycle's small plastic strain gnaws a crack forward over many cycles, not a single catastrophic snap. And a genuinely brittle metal like grey cast iron can fail by true thermal shock in one go. But the classic "dropped it in cold water and it went ping" failure — the mug, the kiln shelf, the furnace tube, the turbine coating — is overwhelmingly a ceramic and glass affair, precisely because those materials have strength but no ductility to spend.

The figure of merit: how big a jump can it take?

We can turn all of this into one number an engineer can rank materials by. Ask the sharpest possible question: for the harshest imaginable quench — plunge a hot body into cold water so its surface reaches the bath temperature instantly — what is the largest temperature drop, delta-T, the surface can survive before it cracks? The surface stress in that instant is about E times alpha times delta-T (softened by a factor of 1 minus Poisson's ratio, because the surface is squeezed in two directions at once). Set that equal to the material's fracture strength, sigma_f — its flexural strength is the right number for a brittle solid — and solve for the critical drop. You get the thermal-shock resistance parameter:

  R  =  delta-Tc  =  sigma_f (1 - nu) / (E x alpha)
         (the largest sudden temperature drop before the surface cracks)

  Want a big R?  ->  high strength (sigma_f),  LOW stiffness (E),  LOW expansion (alpha)

  material          alpha        E        sigma_f      delta-Tc
                   (10^-6/K)   (GPa)      (MPa)       (approx, K)
  --------------   ---------   -----      -------     -----------
  soda-lime glass      9         70          50           ~60
  alumina              8        380         350           ~90
  silicon carbide      4.5      410         400          ~180
  borosilicate         3.3       64          60          ~220
  fused silica         0.5       73          50         ~1100

  Expansion dominates: fused silica beats soda-lime ~20x on alpha alone.
  For a finite (not instant) quench, multiply R by conductivity k:
     R' = k x R    -> a good conductor spreads the gradient and survives more.

Read the formula like a recipe and it tells you exactly what to want. You want high strength (sigma_f on top — a stronger surface tolerates more stress before cracking), low stiffness (E on the bottom — a floppier material generates less stress for the same forced strain, a genuinely counter-intuitive win where being less stiff helps), and above all low thermal expansion (alpha on the bottom — if it barely wants to change size, there is barely any mismatch to fight). That last term is the giant: alpha varies by more than a factor of twenty across common ceramics, so it dominates the ranking. There is a second, subtler figure of merit R' equal to k times R, for real quenches that are fast but not truly instant: here a high thermal conductivity helps, because a material that hurries heat inward flattens the gradient and never lets a steep mismatch build. That is a big reason silicon carbide, which conducts heat well, resists thermal shock better than its modest R alone suggests.

Designing to survive the swing

The figure of merit is not just for grading materials — every term in it is a lever a designer can pull. The most powerful lever is material choice: reach for the low-alpha family. Fused silica and borosilicate glass for labware; cordierite, whose expansion is nearly zero, for the honeycomb of a car's catalytic converter that must ride red-hot exhaust and cold starts twice a day for a decade; silicon carbide and silicon nitride for kiln furniture, diesel glow plugs, and rocket-nozzle throats. Then there are the levers that do not change the material at all: slow the temperature change so the surface and interior stay closer in temperature (preheat the kiln shelf, warm the engine gently), and keep sections thin so heat crosses them quickly and no steep gradient can form — a thick block shocks far worse than a thin wall of the same material.

There is a cleverer trick still: since cold-shock cracks the surface by pulling it into tension, pre-load the surface into compression so a quench merely relaxes it back toward zero instead of into tension. This is exactly the secret of tempered glass — its outer skin is frozen in permanent compression, giving it far more thermal-shock (and impact) headroom than ordinary glass, though at the price of shattering into dice all at once if a crack ever does breach that layer. And a fourth lever attacks not crack starting but crack stopping: a tougher material, one with higher fracture toughness, can arrest a thermal-shock crack after it initiates so the damage is a harmless network of tiny cracks rather than one part-splitting rupture. Zirconia's transformation toughening does exactly this, which is why toughened ceramics degrade gracefully under repeated shock instead of failing all at once.

1. Pick a low-alpha material first — it is the term the figure of merit rewards most (fused silica, cordierite, silicon carbide over soda-lime glass or alumina).
2. Slow the temperature change: preheat, ramp gently, avoid dunking a hot part into cold liquid, so surface and core never diverge far.
3. Keep sections thin and conductive so heat crosses fast and no steep gradient can build.
4. Relieve constraint: let the part expand freely where you can, so no clamp turns the strain into stress.
5. Put the surface in residual compression (temper it) so a quench relaxes tension instead of adding it.
6. Remove the worst flaws and round off sharp corners — the crack starts at a stress concentration, so the finish quality sets the real strength.

Honest edges: a strength number that lies

Before you trust that neat delta-Tc, meet its weaknesses head-on. The formula puts sigma_f on top as if a ceramic had one honest strength, but it does not: a brittle solid fails at its worst flaw, so nominally identical pieces scatter widely, and the strength you should feed the formula is a low-probability value from Weibull statistics, not a rosy average. Two mugs from the same kiln can have quite different real thermal-shock limits simply because one hides a slightly meaner pore. This is the same lesson from the ceramics rung wearing thermal clothes: never trust a single "strength" for a brittle material.

Two more caveats keep you honest. First, delta-Tc is derived for the infinitely fast quench — the worst case where the surface hits bath temperature instantly. Real quenches are gentler, governed by how fast heat actually crosses the surface relative to how fast it spreads inside (the Biot number, if you meet it later); that is why the R' = k times R version exists, and why a real part often survives a bigger nominal drop than the raw R predicts. Second, do not confuse thermal-shock resistance with high-temperature capability — a refractory brick can happily sit at 1600 degrees C yet crack the moment you cool one face too fast. Surviving heat and surviving a change in heat are different talents, and thermal shock is entirely about the change.