Annealing Heat Treatment

Why metals need to be softened on purpose

Cold-rolled steel that comes off a 20-stand tandem mill is dimensionally exact and mirror-bright, but it is also a brick. Yield strength has roughly doubled, ductility has collapsed from 30 % elongation to under 5 %, and the sheet snaps rather than bends. The reason is that every pass through the rolls drove the lattice into plastic strain. Slip systems activated, dislocations multiplied, and what started as a clean equilibrium crystal at 10⁹–10¹⁰ dislocations per square metre now carries 10¹⁵ to 10¹⁶ per square metre — a tangled forest that pins itself.

You cannot keep cold-working that sheet. It will tear in the next drawing die or fracture under the first sharp bend. The metal has to be reset. Annealing is the reset: heat the material until atoms can diffuse, hold long enough for the dislocation tangle to dissolve and fresh grains to grow in its place, then cool slowly so no new stress is locked in. What comes out is metallurgically as if it had never been cold-worked at all. Production lines depend on this cycle — roll, anneal, roll, anneal — and a quarter of all heat-treating in the world is dedicated to it.

The three stages of annealing

Annealing is not a single event but three overlapping processes that take over as temperature climbs. Each is driven by the same fundamental currency — stored elastic energy in the dislocation field — but uses a different mechanism to spend it.

Recovery

Below about 0.3 of the absolute melting temperature, atoms diffuse only short distances. Dislocations cannot disappear but they can move. Like-signed dislocations climb and glide into walls — low-angle sub-grain boundaries — that reduce their elastic strain energy without changing the overall grain shape. The metal loses internal residual stress and electrical resistivity, but yield strength and grain size are nearly unchanged. For a cold-rolled low-carbon steel this stage runs roughly 200–400 °C. Visually you cannot see anything happen under a microscope; the dislocation density has only dropped by a factor of two or three.

Recrystallization

Cross roughly 0.4 T_m and a new mechanism switches on. In regions of locally high strain — typically near old grain corners and shear bands — small strain-free nuclei appear and begin to grow. Each nucleus is a fragment of essentially perfect lattice whose boundary, a high-angle grain boundary, sweeps outward through the deformed matrix. As it passes, every dislocation it encounters is absorbed. Within minutes to hours the entire deformed structure is consumed and replaced by new equiaxed grains with dislocation densities back at 10¹⁰/m². Hardness and yield strength drop sharply; ductility comes back. This is the transformation people usually mean when they say "the metal has annealed."

Grain growth

Once recrystallization is complete, no more strain energy is available to drive change. But there is still a smaller driving force: the surface energy of all those new grain boundaries. To reduce total boundary area, large grains grow at the expense of small ones; boundaries migrate toward their centres of curvature. The mean grain diameter d grows as d² = d₀² + k·t — Burke and Turnbull's classical relation. Coarse grains are usually unwanted (they weaken the metal and cause orange-peel surface texture), so industrial schedules are tuned to stop the hold time just after recrystallization completes.

The numbers behind the dislocation story

condition                ρ (dislocations / m²)   yield strength (MPa)
annealed copper          ~10¹⁰                   ~70
heavily cold-rolled Cu   ~10¹⁵–10¹⁶              ~350
recovered Cu (300 °C)    ~10¹⁴                   ~310
recrystallized Cu        ~10¹⁰                   ~70

The Taylor relation σ_y ∝ √ρ explains the strength change quantitatively: a five-decade rise in dislocation density doubles to triples the yield strength, and annealing collapses both numbers back to the starting point.

Industrial annealing variants for steel

"Annealing" on a heat-treat traveller is not one process but a family, chosen by what the part needs.

Variant	Temperature	Cool	Result	Used for
Full anneal	30–50 °C above A3 (~830–900 °C low-C steel)	Furnace cool, < 30 °C/hr	Coarse pearlite + ferrite, hardness 120–180 HB	Reset the steel for maximum softness/ductility
Process anneal	Below A1 (~650–700 °C)	Air or furnace	Recrystallized ferrite, hardness 110–150 HB	Inter-pass softening during cold rolling/drawing
Stress relief	500–650 °C	Slow air or furnace	No microstructural change; residual stress drops 80 %+	Weldments, large machined parts, castings
Spheroidize anneal	~700 °C (just below A1)	Furnace, 8–24 h	Spheroidal cementite in ferrite	Pre-machining high-C tool/bearing steel
Isothermal anneal	Austenitize, hold at ~650 °C	Air after hold	Uniform pearlite, fast and reproducible	Heat-treat shops with cycle-time constraints
Bright anneal	800–1100 °C in H₂ or vacuum	Furnace	No scale, mirror surface	Stainless steel sheet, electrical strip

The temperatures all live below the alloy's solidus by a wide margin. The "A" lines refer to the iron-carbon phase diagram: A1 = 727 °C eutectoid, A3 = the upper boundary of the (α + γ) field, which depends on carbon content. A1 is the temperature at which pearlite begins to dissolve; A3 is where any leftover ferrite finally disappears into austenite. Full annealing crosses both; process annealing crosses neither.

Worked example: how long does a 1020 steel coil take to anneal?

A coil of cold-rolled AISI 1020 sheet, 1.5 mm thick and 60 % cold-reduced, is sent to a continuous annealing line operating at 750 °C in N₂-5 %H₂.

heating: 25 °C → 750 °C  in line over 90 s
hold:    750 °C           for 60 s
cool:    750 °C → 25 °C   over 180 s in slow-cool tower

The 60-second hold is enough because the JMAK (Avrami) recrystallization kinetics for this grade say:

X = 1 − exp[−(t/τ)ⁿ]
τ(750 °C) ≈ 12 s for 60 %-reduced 1020,  n ≈ 2.5
→ X(60 s) ≈ 1 − exp[−(5)^2.5] ≈ 1 − 1.3 × 10⁻²⁴ ≈ 100 %

So in 60 seconds the sheet is essentially fully recrystallized. The line speed is set by the slow-cool requirement (cooling too fast above ~600 °C re-quenches some retained austenite into bainite, raising hardness), not the recrystallization itself. The same coil annealed at 600 °C would need 30 minutes; at 850 °C, only 5 seconds — but at 850 °C grain growth would run away. Industrial schedules are always a temperature-time-grain-size trade-off.

Atmosphere control

Steel at 800 °C in air oxidises hungrily. A bare coil emerges with millimetres of blue-black mill scale that has to be pickled in hydrochloric acid before any downstream operation. Worse, near the surface, carbon reacts with oxygen and burns out — a decarburized layer 0.1 to 0.5 mm deep that is soft, fatigue-fragile, and worthless for a structural part. Industrial annealing therefore runs under a protective atmosphere:

• Vacuum (10⁻³ to 10⁻⁵ torr). Cleanest. Used for tool steel, titanium, stainless, and anything where hydrogen pickup is unacceptable.
• Dry nitrogen. Inexpensive, slight risk of nitriding at the surface for some alloys.
• Forming gas (N₂ + 5–10 % H₂). The H₂ scavenges residual oxygen and reduces any surface oxide back to metal. Standard for carbon steel sheet.
• Pure hydrogen. Used for bright annealing of stainless and electrical steel. Excellent surface, but it embrittles high-strength steels and many non-ferrous alloys, so it is alloy-specific.
• Dissociated ammonia (3 H₂ + N₂). A cheap on-site source of hydrogen-rich gas, common in older sheet lines.
• Endothermic gas. CO/CO₂/H₂/H₂O blend with the carbon potential set by burner mix. Allows neutral, carburizing, or decarburizing service from a single generator.

Picking the atmosphere is half the job. The other half is keeping it dry: water vapour above about 0.1 % H₂O turns a reducing atmosphere into a decarburizing one.

Annealing vs normalizing vs quench-and-temper

Three steel heat treatments cover most of the space, and they differ only in cooling rate after a hold in austenite.

Process	Cooling	Resulting microstructure	Hardness	Use case
Full anneal	Furnace, hours	Coarse pearlite + ferrite	Lowest — 120–180 HB	Maximum machinability/formability
Normalize	Still air, minutes	Fine pearlite + ferrite	Higher — 150–220 HB	Uniform starting structure before final HT
Quench + temper	Oil/water seconds; reheat 200–650 °C	Tempered martensite	25–55 HRC tunable	Service hardness / toughness for the finished part

Annealed steel is for making the part — soft, easy to cut, easy to bend. Normalized steel is the predictable middle ground used as a starting condition. Quench-and-temper steel is for using the part — a crankshaft, a gear tooth, a spring. A typical production sequence is hot-roll → anneal → cold-form → quench-and-temper.

Annealing of non-ferrous metals

• Copper. Recrystallizes near 200 °C; bright-annealed at 400–600 °C in dry H₂. Wire-drawing lines anneal between every two or three dies. The "soft" temper used in plumbing tube is fully annealed.
• Aluminium alloys. Annealing produces the "O" temper — fully soft, maximum elongation, used for deep-drawing operations and beverage-can bodies. 5xxx and 6xxx alloys anneal at 340–415 °C; the precipitation-hardening 2xxx and 7xxx require slow cooling from 415 °C to avoid retained metastable phases that would harden on aging.
• Brass (Cu-Zn). Cartridge brass (70/30) anneals at 500–700 °C, used between successive deep-draw operations on cartridge cases and lamp components. Beta-phase brass coarsens grains rapidly and is normally avoided.
• Titanium. Mill anneal at 700–800 °C in vacuum or argon. Hydrogen must be excluded — Ti picks it up and embrittles permanently.
• Nickel superalloys. Solution anneal at 950–1180 °C in vacuum/argon to dissolve carbides and prepare for aging precipitation. The result is the soft, ductile starting condition for turbine-disk forging.

Common pitfalls

• Overshooting on grain size. Holding 50 °C too hot or twice as long can double grain diameter, halve strength, and ruin formability. ASTM grain size 7 is typical for sheet; 4 is usually unacceptable.
• Decarburization from a leaking furnace door. Any oxygen or water vapour ingress drops surface carbon. Tested with a Rockwell scan: surface reads 30 HB softer than core means decarb.
• Hydrogen embrittlement of high-strength steel. Bright-annealing a UTS > 1000 MPa steel in pure H₂ produces hydrogen pickup that delays cracks for days. Vacuum or N₂-only is required.
• Incomplete austenitization in full anneal. A 1045 medium-carbon steel must reach 30 °C above A3 (~810 °C) to fully dissolve old pearlite. Hold below A3 and ghost-banding survives the cycle.
• "Quench cracking" of a slow cool. A part with thin and thick sections cools unevenly even in a furnace; residual stress can crack a heavy section. Stress-relief at 600 °C after furnace cool is standard for big castings.
• Misnaming stress relief as anneal. Stress relief does not recrystallize and does not soften appreciably. Specifying "anneal to 100 HRB" when the part is too thick to actually recrystallize at the available temperature is a chronic source of shop-floor disagreement.