Same steel, many personalities
You ended the last rung with a warning that turns out to be the whole point of this guide: a phase diagram is an equilibrium map, so the most useful steel structures — the hard ones, the tough ones — appear nowhere on it. Composition tells you which phases a steel can form; it does not tell you how they end up arranged, and arrangement is where the properties actually live. Heat treatment is the craft of controlling that arrangement. Take a single bar of 0.8 percent carbon steel: cool it lazily and it is soft enough to bend by hand; cool it violently and it becomes glass-hard and can snap like a biscuit. Same atoms, same iron-carbon diagram, wildly different metal — and the only thing you changed was the schedule of heating and cooling.
Almost every steel heat treatment opens with the same master move: austenitizing. You heat the steel up into the austenite field — past 727 degrees C, often 30 to 60 degrees above the relevant boundary — and hold it there. Up there the iron is face-centered cubic austenite, the carbon-hungry phase that dissolves carbon happily and evenly into one uniform solid solution. This wipes the slate clean: whatever coarse, uneven structure the bar arrived with is dissolved away, and every carbon atom is homogeneously spread through the crystal. Everything interesting then happens on the way down, because how fast you cool decides whether that dissolved carbon has time to move.
Annealing and normalizing: the gentle end
Start at the slow, soft end of the cooling menu. Annealing means heat, hold, then cool very slowly — traditionally by switching off the furnace and letting the part drift down with it over hours. Slow cooling gives carbon all the time in the world to diffuse, so the steel settles into the coarse equilibrium structure the phase diagram predicts: big grains, thick lamellae of coarse pearlite. The result is the softest, most ductile, most machinable state the steel offers. There are three flavours worth knowing by name: full annealing (austenitize, then furnace-cool — the softening reset); process annealing, done below 727 degrees to restore ductility to a cold-worked sheet by letting it recrystallize without ever forming austenite; and spheroidizing, a long hold just below 727 degrees that coaxes the brittle cementite plates to ball up into rounded spheroids, giving high-carbon and tool steels their softest, most machinable possible condition before they are hardened.
Normalizing is annealing's slightly brisker cousin. You austenitize a touch hotter, then pull the part out and cool it in still air instead of in the furnace. Air is faster than a cooling furnace, so carbon has less time to diffuse and the pearlite comes out finer — thinner, more closely spaced ferrite-and-cementite layers. Finer pearlite is stronger and tougher than coarse pearlite at the exact same carbon content (arrangement, again, not composition), so a normalized steel is a bit harder and stronger than an annealed one, with a more uniform, refined grain structure. That is why castings and forgings are so often normalized: it erases the coarse, patchy grains left by solidification and gives the part a clean, even starting point. Where annealing chases the very softest state, normalizing chases uniformity and a modest strength bump.
The cooling-rate menu (same austenitized 0.8% C steel)
slower <------------------------------------------> faster
furnace cool still air oil quench water / brine
| | | |
coarse fine (miss the martensite
pearlite pearlite C-curve nose) (BCT, glass-hard)
| | | |
softest, stronger, bainite / hardest,
most tougher, mixed most brittle
ductile uniform structures -> MUST temper
ANNEAL NORMALIZE QUENCH (+ TEMPER)Quenching: outrunning diffusion
Now the dramatic end. Everything soft above relied on carbon having time to diffuse. Quenching is the deliberate act of denying it that time. From the phase-transformation rung you know the trick lives on the TTT diagram: the transformation of austenite into pearlite or other cementite-bearing structures traces a C-shaped curve with a fastest-acting "nose", and if you cool quickly enough to sweep past that nose before the curve is reached, none of those diffusion-based products can form. The carbon is trapped. With nowhere to go, the face-centered-cubic lattice cannot revert to ordinary ferrite; instead it shears, almost instantly and without any diffusion at all, into a strained, carbon-stuffed body-centered-tetragonal structure called martensite. It is not on the equilibrium diagram because it is not an equilibrium phase — it is austenite caught mid-transformation and frozen.
Martensite is the hardest thing plain steel can become — and among the most brittle. Its hardness comes from carbon jammed into interstitial sites it does not fit, distorting the lattice everywhere and studding it with obstacles. Recall the picture of plastic flow as a gliding dislocation shuffling a ruck across a rug: in martensite the rug is pinned down so densely that the ruck can barely move at all, which reads on a tester as sky-high hardness and almost no ductility. A high-carbon steel can leap from perhaps 20 on the Rockwell C scale when annealed to over 60 when quenched. One honest subtlety: martensite hardness is set almost entirely by carbon content, not by how fast you quenched — quench faster and you get more of the bar turned to martensite, not harder martensite. A low-carbon steel simply cannot be quenched to a high hardness, because there is too little trapped carbon to do the jamming.
How fast you cool depends on the quench medium, and there is a whole menu of severities: agitated brine is fiercest, then plain water, then oil, then a blast of air, then still air — each a rung gentler than the last. Fiercer sounds better, since it more surely beats the nose, but speed has an ugly price. The surface of a quenched part contracts and transforms before the core does, and martensite is actually slightly less dense than the austenite it comes from, so it wants to expand at the wrong moment. Those clashing volume changes leave the part full of residual stresses that warp it, or, in a violent water quench of a complex shape, crack it outright. This is the quencher's dilemma: harsh enough to harden, gentle enough not to tear itself apart.
Tempering: relaxing the glass
As-quenched martensite is almost never used as it comes out of the tank. It is too brittle — full of trapped carbon and residual stress, it can shatter under a shock or even crack sitting on the bench. The cure is tempering: you reheat the quenched part to a moderate temperature, somewhere between roughly 150 and 650 degrees C (always below 727, so you never re-form austenite), and hold it there. At that warmth the trapped carbon finally gets just enough mobility to creep a short way out of the strained lattice and precipitate as a fine dispersion of tiny cementite particles. The lattice relaxes back toward ordinary ferrite, the internal stresses drain away, and the metal transforms into tempered martensite: minute cementite particles peppered through a soft ferrite matrix. This is the analogy you were promised — quench the steel glass-hard, then relax it just enough to become tough.
The tempering temperature is a dial you turn to buy back exactly as much toughness as you want, and you pay for every degree of it in hardness. A low temper near 200 degrees C sheds the worst brittleness while keeping most of the hardness — the setting for a knife edge or a ball bearing that must resist wear above all. A high temper near 550 to 650 degrees sacrifices a lot of hardness for a great deal of toughness — the setting for a spring, an axle, or a gear that must absorb pounding without cracking. This is the strength-ductility trade-off you have met all the way up this ladder, but now with a control knob in your hand: quench sets the ceiling of hardness, and the tempering temperature slides you back down that ceiling toward whatever balance of hard-and-brittle versus soft-and-tough the job actually needs.
Two honest caveats before you trust this too far. First, tempered martensite is not the same as pearlite, even though both are just ferrite plus cementite by phase — the tempered structure has far finer, rounder cementite particles and is much tougher at the same hardness, a vivid reminder that arrangement beats bookkeeping. Second, tempering is not always monotonic: several alloy steels suffer temper embrittlement, becoming brittle if they are tempered in, or slowly cooled through, a particular window (very roughly 250 to 400 degrees C for one kind, higher for another), often because trace impurities segregate to the grain boundaries. Metallurgists dodge these windows deliberately. None of this — not tempered martensite, not the embrittlement traps — appears anywhere on the iron-carbon diagram, because every last bit of it is a story about non-equilibrium kinetics, which the equilibrium map is silent about by design.
The recipe, and a hard skin on a tough core
- Austenitize. Heat the steel into the austenite field and hold, so every carbon atom dissolves into one uniform FCC solid solution — the clean slate that every heat treatment starts from.
- Quench. Plunge it into the chosen medium (oil or water) fast enough to sweep past the nose of the C-curve, freezing the austenite into hard, brittle martensite before any carbon can diffuse out.
- Pause and respect the danger. As-quenched, the part is now at its maximum hardness but also glass-brittle and full of residual stress — unusable, and liable to crack if left too long.
- Temper. Reheat to a moderate temperature below 727 degrees C and hold, letting a little carbon precipitate as fine cementite so the strained lattice relaxes and the toughness returns.
- Set the dial. Temper low to keep hardness for an edge or bearing; temper high to trade hardness for toughness in a spring, shaft or gear. Quench for the ceiling, temper for the balance.
Sometimes you want two opposite personalities in one part. A gear tooth should have a glass-hard skin that shrugs off wear, wrapped around a tough core that soaks up shock without snapping. Case hardening delivers exactly that by cheating on composition where it counts. In carburizing, you pack a low-carbon steel part in a carbon-rich atmosphere at austenitizing heat and let carbon diffuse into the surface for hours; the skin drinks up carbon while the core stays lean. Quench the whole part and the carbon-rich case transforms into hard martensite while the low-carbon core, with too little carbon to harden much, stays tough. One quench, two microstructures, precisely because carbon content — not cooling rate — sets martensite hardness, exactly as you saw earlier.
Step back and see what you now hold. Composition, from the last rung, fixes which phases a steel can have; heat treatment, from this one, fixes how those phases are actually arranged — coarse pearlite, fine pearlite, bainite, martensite, tempered martensite — and arrangement is where hardness, strength and toughness are truly decided. From dead-soft spheroidized stock to file-hard tempered martensite, it is all one bar of steel and one furnace. What this guide quietly kept bumping against was depth: a fierce quench hardens the skin but may leave a thick core soft, and only some steels harden all the way through. Why that is, how alloying changes it, and the elegant end-quench test that measures it, are the whole of the next guide.