Melt In, Shape Out: Why Freezing Is the Whole Story
Casting is disarmingly simple to describe: melt a metal, pour it into a hollow mold shaped like the part you want, and wait for it to freeze solid. Humans have done this for six thousand years, and it is still how we make engine blocks, jewellery, and jet-engine blades. But do not let the simplicity fool you. The processing-structure-property chain you met at the very start of this ladder is nowhere more naked than here: the split second when the liquid turns to crystal writes the part's grain size, its texture, its porosity, and its internal stresses — and those in turn set how strong, how tough, and how long-lived the finished part will be. Casting does not just give the part its shape; it gives the part its microstructure, and you are stuck with whatever the freezing chose.
The first surprise is that most metals shrink when they freeze — the opposite of water, which is the weirdo that expands into ice and cracks your pipes. Picture a crowd milling about loosely in a hall (the liquid); when the music stops they snap into neat, close-packed rows (the crystal), and neat rows take up less floor. That contraction is real and large: aluminum drops about 6.6 percent in volume the instant it solidifies, cast iron a few percent, and nearly every metal falls somewhere between 3 and 7 percent. If nothing feeds that missing volume, the last liquid to freeze leaves a hollow — a shrinkage cavity — right in the heart of the casting. Half of foundry craft is a running battle against this single fact.
How a Liquid Becomes a Crystal: Nucleation and Growth
Freezing does not start everywhere at once. It begins at tiny seeds and grows outward — the same nucleation-and-growth two-step you met in the phase-transformation rung, now happening straight out of the melt. And nucleation has a catch you already know: building the very first speck of solid costs surface energy, so a speck below a certain critical radius is unstable and simply re-melts, like trying to start a snowball on a smooth wet slope — too small and it slumps back into slush. To make stable seeds the liquid has to be cooled below its true melting point, a lag called undercooling, and the larger the undercooling, the smaller the critical radius, so seeds finally survive.
Left utterly clean, a pure metal is stubborn about this: homogeneous nucleation, where seeds appear unaided in the bulk liquid, may need the melt supercooled by hundreds of degrees (pure nickel can sit liquid nearly 250 degrees C below its melting point). Real castings never do this, because the mold wall and stray specks of oxide or dirt offer ready-made ledges for the solid to grab onto — heterogeneous nucleation — the way a raindrop condenses on a dust mote instead of forming from thin air. On those surfaces solid appears within a few degrees of undercooling. Foundries lean into this on purpose: they stir in grain refiners (inoculants, like titanium-boron particles in aluminum) that flood the melt with nucleation sites, so it freezes into a great many small grains rather than a few coarse ones.
Why fight so hard for small grains? Because of a payoff you have banked before: the Hall-Petch relationship, where yield strength rises as one over the square root of grain diameter, because more grain boundary means more barriers a gliding dislocation cannot cross. Halve the grain size and the Hall-Petch term grows by the square root of two, about 1.4 — free strength, and, better still, finer grain usually lifts toughness too. This is the one lever in metallurgy that pushes strength up without paying in ductility, and casting reaches it through cooling rate: cool faster and you both nucleate more grains and give them less time to coarsen. That is the deep reason a chilled die-cast skin is stronger than a slowly-cooled sand-cast core.
A COOLING INGOT -- three grain zones (cross-section, schematic)
mold wall (cold) mold wall
|####|============ | | | ============|####|
|####|============ | | | ============|####| <- CHILL zone:
|####|============ | | | ============|####| tiny random grains
|####|============ | | | ============|####| quenched on the wall
^^^^^^^^^^^^^ ^^^^ ^^^^^^^^^^^^
COLUMNAR zone: | COLUMNAR zone
long grains grow| along the heat-flow
direction (inward) -> a TEXTURE, so anisotropy
|
EQUIAXED zone (center):
slow, coarse, roughly equal-sided grains;
last to freeze -> where shrinkage + gas hide
Faster cooling widens the fine chill zone; slower cooling grows
the coarse equiaxed core. Grain size is NOT uniform across a part.The Casting Toolbox: Sand, Die, Investment, Continuous
The four big casting families are really four answers to one question: how do I hold the liquid while it freezes, and how fast do I let it cool? Sand casting packs damp sand around a pattern, then pulls the pattern out to leave a cavity. Sand is dirt-cheap, takes any size from a doorknob to a ship's propeller, and shrugs off high melting points — but it is a single-use mold with a rough finish, and sand is a good insulator, so the metal cools slowly and freezes coarse-grained. It is the go-to for large, complex, or low-volume parts like engine blocks. Die casting is its high-speed opposite: molten metal is rammed under high pressure into a reusable steel die. The metal mold sucks heat out fast, giving a fine-grained, smooth, precise part at breakneck rates — but the steel die only survives low-melting alloys (aluminum, zinc, magnesium), and the violent fill can trap air as gas porosity.
Investment casting — lost-wax casting — is the jeweller's and the turbine-maker's art. You build the part in wax, dip it repeatedly in ceramic slurry to grow a hard shell, melt the wax out, and pour metal into the vacated cavity. It captures exquisite detail and needs almost no machining, which is why it makes both wedding rings and nickel-superalloy jet blades; the price is a slow, one-shot, expensive mold. Continuous casting is the industrial workhorse behind almost all steel: liquid metal is poured nonstop through a short water-cooled mold and emerges as a never-ending solid strand, sheared to length downstream — efficient, uniform, and the source of the slabs and billets the next guide will forge and roll.
Defects: Shrinkage, Porosity, and Coring
Recall that 3-to-7-percent shrink on freezing. The cure is a riser (or feeder): a reservoir of extra liquid, deliberately placed to freeze last, that trickles down to top up the shrinking casting as it solidifies — like keeping a bottle poised over an ice-cube tray, topping it up while the water sinks. Get the riser too small or badly placed and the last pocket of liquid, starved of feed, collapses into a shrinkage cavity. A second, sneakier villain is gas porosity: hot metal dissolves far more gas than cold solid can hold (aluminum is notorious for soaking up hydrogen), so as it freezes the excess gas boils out as bubbles trapped in the solid. And a casting cooling unevenly builds internal thermal stress as hot and cold regions pull against each other — enough, while the metal is still weak and part-liquid, to tear it apart in a defect called hot tearing, and enough, once solid, to leave locked-in residual stress.
- Feed the shrink: add a riser that freezes LAST and sits above the heaviest section, so liquid can flow downhill into every pocket as the casting contracts.
- Chase the gas: degas the melt (and avoid a violent, air-churning fill) so dissolved hydrogen or trapped air cannot bloom into porosity as it freezes.
- Cool it evenly: taper thick-to-thin sections and add chills so the part freezes progressively toward the riser, not into isolated hot spots that hot-tear or leave residual stress.
- Even out the chemistry: if the alloy cored, give it a homogenization anneal so diffusion can flatten the composition before the part goes into service.
Reading the Fingerprint — and Why Blades Are Single Crystals
Every casting is born with a microstructural fingerprint, and now you can read it: fine or coarse grain (from cooling rate), a directional columnar texture (so the part is anisotropic — stronger along the grains than across, like wood), porosity and shrinkage voids that act as ready-made stress raisers where a fatigue crack will start, coring, and residual stress. That fingerprint is not an accident to be ashamed of — it is information. This is why cast iron earns its own name: it cannot be forged, so its entire personality, from grey iron's soft graphite flakes to the tough graphite spheres of ductile iron, is decided at the moment of freezing and never re-worked.
The most beautiful twist closes a loop from the creep rung. At room temperature grain boundaries strengthen a metal — that is the whole Hall-Petch gift we chased above. But run a metal red-hot under load for thousands of hours and those same boundaries become the weak seams where the metal slowly slides and voids open up: they promote creep. A jet-engine turbine blade lives exactly there, glowing under enormous stress. So the most advanced blades are cast by investment casting with such carefully steered heat flow that they solidify as a single columnar structure with the boundaries running only lengthwise — or, at the pinnacle, as one single crystal with no grain boundaries at all. The same casting-and-solidification physics that gives a cheap sand casting its coarse grain, pushed to its limit, gives a superalloy blade eternal life at 1000 degrees C. Same physics, opposite goal — and that is the theme this whole rung will keep proving: every process leaves a fingerprint, and mastery is choosing the fingerprint you want.