Heat Treatment of Steel: The Complete Engineering Guide to Annealing, Quenching, Tempering, Hardenability, and Case Hardening

A dense equation-driven engineering deep-dive into the heat treatment of steels — the Fe-C phase diagram with lever-rule worked examples, TTT/CCT kinetics with the Avrami equation, martensite start temperature prediction via the Andrews equation, hardenability with Grossmann H-values and Jominy end-quench curves, quenching media physics, tempering with the Hollomon-Jaffe parameter, carburizing diffusion math with a worked erf() case-depth calculation, nitriding and induction hardening (skin-depth math included), the complete alloy-grade recipe table from C45 to D2, decarburization and quench-crack defect taxonomy, Indian heat-treatment shop economics with indicative ₹/kg rates, and two fully worked specifications for a 4140 shaft and a case-hardened gear.

·

Heat Treatment of Steel: The Complete Engineering Guide to Annealing, Quenching, Tempering, Hardenability, and Case Hardening

Heat treatment is the step that turns cheap carbon steel into a spring, a bearing race, a gear tooth, or a die that will stamp a million parts. Steel is the only structural material whose properties you can tune across a two-decade range — from 150 HB annealed softness to 65 HRC martensite — using nothing but heat, time, and cooling rate. India makes roughly 150 million tonnes of crude steel a year, and a meaningful fraction of it passes through one of the thousands of commercial heat-treatment shops clustered in Pune, Rajkot, Ludhiana, Faridabad, and Coimbatore. The global commercial heat-treatment market is estimated at around US$100 billion. Every topic we've covered on this blog — casting, CNC machining, gears, fasteners, welding — terminates in a furnace.

This guide covers the physics, the kinetics, the process recipes, the defects, and the Indian economics, ending with two fully worked heat-treatment specifications. If you design parts, machine parts, or buy parts, this is the vocabulary your heat treater speaks.

1. Why Steel Answers to Heat: The Allotropy Engine

Iron's crystal structure changes with temperature, and that single fact is the entire industry.

1.1 Two Lattices, One Metal

That 100× difference in carbon solubility is the whole game: heat steel above ~727°C and carbon goes into solid solution in austenite; cool it fast enough and the carbon gets trapped in a supersaturated lattice that is ferrite no more.

1.2 The Fe–Fe₃C Phase Diagram: The Map

The invariant points that matter for heat treatment:

Feature · Temperature · Carbon

Eutectoid (A1) · 727°C · 0.77% (often rounded to 0.8%)

Eutectic · 1148°C · 4.30%

Max C in austenite · 1148°C · 2.11%

Max C in ferrite · 727°C · 0.022%

Cementite (Fe₃C) · — · 6.67%

Pearlite is the eutectoid product: alternating lamellae of ferrite and cementite, ~88 wt% ferrite and ~12 wt% cementite. Hypoeutectoid steels (<0.77% C) form proeutectoid ferrite before pearlite; hypereutectoid steels (>0.77% C) form proeutectoid cementite.

Lever-rule worked example. Take a 0.40% C steel (EN8/C45) cooling slowly. At 728°C, just above the eutectoid, the fraction of proeutectoid ferrite is:

The remaining 50.5% austenite transforms to pearlite at 727°C. So an annealed C45 is roughly half ferrite islands, half pearlite colonies — which is why its annealed hardness is only ~160–180 HB and why it machines so well. The A3 temperature for 0.40% C is ~810°C; below A3, ferrite starts appearing. Every austenitizing temperature in every recipe in this guide traces back to these lines.

2. The Annealing Family: Resetting the Metal

Annealing is heat treatment's "factory reset": heat, hold, cool slowly. But there are five distinct resets, and using the wrong one wastes furnace hours or ruins a part.

Process · Temperature · Cooling · Purpose · Typical result (0.45% C steel)

Full annealing · 30–50°C above A3 (~850–880°C for C45) · Furnace cool, ~20–100°C/h · Soften for machining, relieve cast/forged structure · ~150–170 HB, coarse pearlite

Normalizing · 30–50°C above A3/Acm · Still air · Refine grain, homogenize, reset after forging/welding · ~180–210 HB, finer pearlite

Process (subcritical) annealing · 550–700°C · Air · Recrystallize cold-worked low-carbon steel, stress relief between draws · Soft, recrystallized ferrite

Spheroidizing · ~700°C for 10–30 h (or cyclic ±20°C around A1) · Slow · Globularize carbides for high-C tool steels · D2: 207–255 HB; O1: ~190–210 HB

Stress relieving · 150–200°C below A1 (typically 500–650°C) · Slow, uniform · Remove machining/welding residual stress without phase change · Hardness unchanged

The engineering logic: full anneal when you need maximum machinability on a low/medium-carbon part; normalize when a forging or weldment has a coarse, non-uniform grain structure and you need it homogenized before hardening (normalizing also works on castings — see our casting guide for why as-cast structures need it); spheroidize when the carbon content is high enough that lamellar pearlite would eat your end mills — D2 and O1 tool steels are spheroidize-annealed at the mill precisely so you can machine them. Try to full-anneal a 1.5% C steel instead and you'll get hard, brittle pearlite plus a carbide network that shatters cutters.

3. Austenitizing, TTT Diagrams, and the Three Daughters of Austenite

3.1 Austenitizing

Heat hypoeutectoid steel 30–50°C above A3; hypereutectoid steel just above A1 (going to Acm dissolves the proeutectoid cementite and grows the grain — both bad). Soak long enough to dissolve carbides and homogenize carbon — roughly 30–60 min per 25 mm of section for carbon steels. Overshoot and grain grows; austenite grain size (ASTM E112, N = 2^{G-1} grains/in² at 100×) is a controllable variable that feeds straight into toughness.

3.2 The TTT Diagram: Time, Temperature, Transformation

Cool austenite below A1 and hold, and it decomposes by a diffusion-controlled nucleation-and-growth process with sigmoidal kinetics — the Avrami equation:

where f is fraction transformed, t is time, and k, n are temperature-dependent constants. Plot the 1% and 99% transformation lines against temperature and you get the C-curve — the TTT (isothermal transformation) diagram. The nose of the curve sits where the driving force (undercooling) and atomic mobility balance, typically 500–600°C for plain carbon steels.

Three daughters of austenite, ordered by undercooling:

Pearlite (above the nose, ~550–727°C). Ferrite + cementite lamellae. Interlamellar spacing scales inversely with undercooling, S \propto 1/\Delta T: coarse pearlite (~0.5 μm spacing, 230–280 HV) at low undercooling, fine pearlite (~0.1 μm, 300–380 HV) near the nose. The spacing, not the composition, sets the hardness.

Bainite (below the nose, ~250–550°C). Upper bainite (feathery, cementite between ferrite laths, 300–400 HV) and lower bainite (acicular, fine carbides inside the laths, 350–450 HV). Lower bainite is genuinely attractive: hardness approaching martensite with better toughness and far less distortion — the entire rationale for austempering.

Martensite (below Ms, if cooling is fast enough to miss the nose). A diffusionless military transformation: the FCC austenite shears cooperatively into BCT martensite, carbon trapped in solution. No time for carbon to escape — it's all kinetics. The martensite start temperature depends on composition; the Andrews (1965) linear regression for low-alloy steels is the workhorse:

all in wt%. For 1040 (0.40 C, 0.75 Mn): M_s = 539 - 169 - 23 \approx 347°C. For 1095 (0.95 C): M_s \approx 116°C — martensite doesn't even start until ~115°C, and the transformation finishes below room temperature, leaving 5–10% retained austenite (more in high-C tool steels). The austenite→martensite reaction carries a ~4% volume expansion (for a 1% C steel) — the root cause of both quench distortion and quench cracking.

Martensite hardness is set almost entirely by carbon: ~40 HRC at 0.2% C, ~50 HRC at 0.4% C, plateauing at 64–65 HRC around 0.6–0.8% C.

3.3 Critical Cooling Rate and Why Alloying Matters

Continuous cooling doesn't follow the TTT diagram exactly — CCT diagrams shift the curves right and down. The critical cooling rate is the slowest cooling that produces full martensite. A plain 0.8% C steel must streak through the 700–500°C band at on the order of 100°C/s — the nose gives it under a second. Add chromium, molybdenum, and nickel (4340/EN24), and the pearlite/bainite nose shifts dramatically right: the same steel hardens through at ~10°C/s and will air-harden in small sections. This time window is the entire concept of hardenability — and the entire reason 40NiCrMo costs more than C45.

4. Hardenability: It's About Depth, Not Hardness

Hardenability is the depth to which a steel hardens under a given quench — a property of the steel, not the furnace. Two steels can quench to the same surface hardness while one is martensitic to 40 mm and the other to 4 mm.

4.1 Quench Severity: Grossmann H-Values

The H-value rates how fast a medium extracts heat from a 25 mm bar:

Quenchant · H-value

Still air · ~0.02–0.05

Still oil · ~0.25–0.35

Agitated oil · ~0.4–0.5

Still water · ~1.0

Agitated water · ~1.5

Brine (10% NaCl) · ~2.0

4.2 The Jominy End-Quench Test (ASTM A255)

The standard hardenability fingerprint: austenitize a 25.4 mm × 100 mm bar, mount it vertically, and blast the bottom face with a controlled 24°C water jet from a 12.7 mm orifice at 65 mm free-jet height. Cooling rate decays monotonically with distance from the quenched end; measure hardness at J1.5, J3, J5, J10, J25, J50 mm.

Typical numbers: AISI 1040 reads ~55 HRC at J1.5 mm but collapses to ~28 HRC by J13 mm — it's a surface-hardening-only steel in anything bigger than a screwdriver. AISI 4340 holds ~55 HRC past J25 mm. Same quench, same bar, thirty times the hardenable depth. That's why a 150 mm diameter 1040 shaft can never be through-hardened no matter how aggressive the quench — the core cooling rate is set by conduction, and the core will always fall into the pearlite nose.

The ideal diameter (DI) concept extends this: the largest bar diameter that would quench to 50% martensite at the center in an ideal (H = ∞) quench, computed by multiplying Grossmann factors for each alloying element and grain size. Mn, Cr, and Mo are the cheapest hardenability per rupee; B (boron, 0.001–0.003%) is the most potent per gram.

5. Quenching: The Fastest Is Not the Best

5.1 The Three Stages of a Liquid Quench

Every liquid quench runs the same sequence:

  1. Vapor blanket (film boiling): a steam/vapor jacket insulates the part — cooling is slow, non-uniform, and unpredictable. Agitation and additives (salt, polymer) collapse it faster.
  2. Nucleate boiling: violent bubble formation at the surface; peak heat extraction, often 200–600°C/s.
  3. Convection: below the boiling point, cooling slows to the quenchant's convective rate.

The hazard is non-uniformity: the vapor blanket lingers longest in pockets, bores, and corners — classic soft-spot generators.

5.2 Media Choice, Distortion, and Cracks

Quenchant · Severity · Distortion risk · Use for

Brine · ~2.0 · Highest · Low-C plain steels, small sections

Water · ~1.0 · High · C ≤ 0.4% plain carbon, medium sections

Polymer (PAG, 5–20%) · ~0.5–1.0 · Medium · Between water and oil, tuned by concentration

Oil (fast/marquench) · ~0.25–0.4 · Low · Alloy steels, tools, anything C > 0.4%

Forced air / gas · ~0.05 · Lowest · Air-hardening steels (D2, A2), vacuum quenching

The distortion physics: the part cools surface-first, so thermal contraction puts the surface in tension — then the surface transforms to martensite first, and the ~4% expansion flips the stress state. The net warpage is the difference between two large, opposing effects, which is why prediction is hard and why the slower quench that still reaches full hardness is always the right one. Quench cracking happens when transformation stresses plus thermal stresses exceed the (already brittle) as-quenched strength — classic in water-quenched high-carbon parts with sharp corners, keyways, or abrupt section changes. Rules of thumb: above ~0.4% C, use oil; radius every corner; avoid water on parts with keyways cut before hardening.

5.3 Interrupted Quenching: Martempering and Austempering

Martempering (marquenching): quench into hot oil/salt at just above Ms (~200–250°C), hold until temperature equalizes through the section, then air cool. The transformation happens simultaneously everywhere, near-isothermally — distortion drops dramatically at a small hardness cost.

Austempering: quench into salt held in the bainite range (250–450°C), hold until the Avrami curve completes bainite, air cool. No martensite → no quench cracks, minimal distortion, and lower bainite at 350–450 HV. Applied to springs, clips, and — with ductile iron — austempered ductile iron (ADI), which reaches 900–1600 MPa tensile (ASTM A897) and replaces forgings in gears and suspension parts.

6. Tempering: Trading Hardness for Toughness

As-quenched martensite is a supersaturated, internally stressed, brittle mess. Tempering is the controlled decomposition: heating to 150–650°C precipitates transition carbides (ε, then Fe₃C), decomposes retained austenite, and coarsens carbides — hardness falls, toughness rises, and the tradeoff is tunable.

The time-temperature equivalence is captured by the Hollomon–Jaffe parameter:

with T in kelvin and t in hours. A tempering cycle is a line of constant P: 1 hour at 600°C (P = 873 \times 20 = 17{,}460) is equivalent to 16.4 hours at 550°C (823 \times (20 + 1.215) = 17{,}460) — useful for deep sections where the core needs time to reach temperature.

Tempering curve for 4140/EN19 (1 h cycles, oil quenched from 850°C):

Temper temperature · Hardness

As-quenched · ~55–57 HRC

200°C · ~52–53 HRC

300°C · ~48 HRC

400°C · ~43–44 HRC

500°C · ~37 HRC

550°C · ~33–35 HRC

650°C · ~28–30 HRC

Avoid 250–400°C for impact-loaded parts — tempered martensite embrittlement (TME) dips the Charpy notch toughness right where the hardness looks tempting. Bearings and tools that need full hardness temper low: EN31/52100 is tempered at 150–180°C for 60–64 HRC. Hot-work steels do the opposite: H13 tempered at 550–600°C shows secondary hardening — fine Mo/V carbide precipitation pushes hardness back up to 44–52 HRC while the matrix softens, exactly what a die that runs at 500°C in service needs (double or triple temper to decompose retained austenite fully).

Cryogenic treatment (−80 to −196°C, after quenching, before tempering) converts retained austenite to martensite — standard practice on precision gauges, D2 dies, and carburized gears where a few percent of retained austenite means dimensional drift and ~1 HRC loss.

7. Case Hardening: Hard Skin, Tough Core

Gears, shafts, cams, and pins need a 58–62 HRC wear surface on a 30–40 HRC shock-absorbing core. Four process families deliver it:

7.1 Carburizing

Carbon diffuses into the austenite surface at 900–950°C. Gas carburizing runs in a furnace with an endothermic carrier gas (~20% CO, 40% H₂, 40% N₂) plus methane enrichment, with the carbon potential (Cp, typically 0.8–1.0%) controlled by oxygen probe or dew point. Diffusion in is governed by Fick's second law, with the classic semi-infinite error-function solution:

with the carbon diffusivity in austenite following D = D_0 \exp(-Q/RT), D_0 \approx 2 \times 10^{-5} m²/s, Q \approx 142 kJ/mol — about 1.5 \times 10^{-11} m²/s at 950°C.

Worked example. Cp = 1.0% at the surface (C_s), core C_0 = 0.2% (20MnCr5). After 4 h at 950°C, where does the carbon reach 0.4%?

So 4 hours yields a ~0.75 mm enriched layer — matching industrial boost/diffuse practice for a 0.6–0.8 mm effective case depth (the depth to 550 HV, per ISO 2639). Typical carburizing steels: 20MnCr5, 16MnCr5, EN353 (3% Ni), SAE 8620, 9310. After carburizing: harden from 830–860°C (oil), temper 150–200°C → surface 58–62 HRC, core 30–40 HRC. Pack carburizing (charcoal + energizer in a sealed box, 8–12 h for ~1 mm) survives in small Indian shops for one-off gears; vacuum carburizing (low-pressure acetylene, 900–1000°C) is the modern answer — faster, no intergranular oxidation.

7.2 Carbonitriding

Carburize at a lower temperature (820–870°C) with ammonia added: nitrogen stabilizes austenite and boosts hardenability. Case depth 0.1–0.5 mm, quench directly from the furnace (or gas quench) — cheaper, lower distortion, ideal for small stampings and fasteners.

7.3 Nitriding

No quench at all: diffuse nitrogen into ferrite at 500–570°C in ammonia. Surface hardness comes from alloy nitride precipitation, so the steel matters: nitriding steels contain Al, Cr, Mo (Nitralloy 135M / EN41B: ~1% Al, 1.4–1.8% Cr, 0.1–0.25% Mo) and reach 900–1100 HV at the surface — harder than any martensite. Case depth 0.1–0.75 mm in 24–96 h. The brittle white layer (ε-Fe₂N and γ′-Fe₄N, 5–25 μm) is usually ground off or suppressed. Because the temperature stays below A1 and there's no quench, nitriding distorts nothing — the standard finish for precision spindles, extrusion screws, and injection-molding screws that can't afford a single micron of warpage.

7.4 Induction (and Flame) Hardening

Surface heating by eddy currents, instantly quenched — a martensitic case in seconds, no long furnace cycle. Depth control comes from frequency: skin depth

with resistivity \rho in μΩ·m and frequency f in Hz. Steel just above its Curie point (768°C) has \rho \approx 1.2 μΩ·m (permeability collapses to ~1): at 10 kHz, \delta = 503\sqrt{1.2/10{,}000} \approx 5.5 mm; at 300 kHz, \delta \approx 1.0 mm. So 3–10 kHz hardens crankshaft journals and axle shafts to 2–6 mm, while 100–500 kHz contour-hardens gear teeth to ~1 mm. Materials: 0.35–0.55% C steels (1045, EN8, 55C8 per IS 1570). Flame hardening is the torch-and-quench cousin — cruder control, but a ₹5,000 setup on a lathe bed can case-harden a ₹50,000 shaft.

Case-hardening comparison:

Process · Temperature · Quench · Case depth · Surface hardness · Distortion · Cost

Carburizing · 900–950°C · Oil (after reheat) · 0.5–2.5 mm · 58–62 HRC · Moderate · Medium

Carbonitriding · 820–870°C · Direct oil/gas · 0.1–0.5 mm · 58–63 HRC · Low · Low

Nitriding · 500–570°C · None · 0.1–0.75 mm · 900–1100 HV · ~Zero · High (long cycle)

Induction · Surface only · Water/polymer · 1–8 mm · 55–62 HRC · Low · Low–medium

8. Furnaces, Atmospheres, and the Equipment Question

The furnace decides the process envelope:

Temperature uniformity is the quality gate: AMS 2750 pyrometry, ±5°C uniformity surveys, and calibrated thermocouples separate a certified commercial heat treater from a charcoal forge. Ask for the furnace chart, not the brochure.

9. Defects, Inspection, and How to Write a Heat-Treatment Spec

9.1 The Defect Taxonomy

Defect · Mechanism · Prevention

Decarburization · C oxidizes out of the surface above ~700°C in air — 0.1–0.3 mm can vanish in an hour at 900°C · Controlled atmosphere, vacuum, copper plating, machining allowance

Scale/oxidation · FeO/Fe₂O₃ layer in air furnaces · Atmosphere, salt, vacuum

Quench cracks · Transformation + thermal stress > as-quenched strength; water on high-C steel, sharp corners · Oil quench, radii, marquenching

Soft spots · Vapor blanket trapped in pockets; scale acting as insulator; non-uniform agitation · Agitation, cleanliness, fixture design

Distortion · Thermal + transformation strain imbalance · Slowest adequate quench, fixtures, press quenching, nitriding for precision parts

Overheating · Austenitizing too hot → grain growth → coarse brittle fracture · Control, normalize to recover

Burning · Near-solidus incipient melting — grain-boundary oxidation, permanent scrap · Control (nothing recovers it)

Temper embrittlement · TME at 250–400°C; or reversible embrittlement from slow cooling through 450–550°C in Cr-Mn steels · Avoid the window; add Mo; quench from tempering

9.2 Inspection

Hardness is the universal acceptance gate: Brinell (castings, annealed stock, <450 HB), Rockwell C (hardened parts, 20–70 HRC), Vickers microhardness (case-depth traverses, 10–1000 gf). For steels below ~450 HB, the conversion everyone uses:

A case-depth check is a microhardness traverse: effective case depth = depth to 550 HV (ISO 2639); total case = depth to core hardness. Batch control: one furnace chart + hardness readings per lot, kept with the part serial — traceability that aerospace and auto OEMs audit.

9.3 The Specification

"HT to 28–32 HRC" is an underspecification — it tells the treater nothing about which steel, which quench, or which tempering window. A complete spec reads:

EN19 (AISI 4140), Ø40 mm shaft: normalize 860°C/air; austenitize 840–860°C, soak 60 min; oil quench (agitated, 40–60°C); temper 580–620°C, 2 h, air cool. Result: 28–32 HRC, ~900–1000 MPa UTS. Decarburization ≤ 0.1 mm. Hardness test 3 points/part, 2 parts/lot.

That spec, plus a furnace chart and a signed hardness report, is the difference between a shaft that survives 50,000 km and one that shears at the first stress raiser.

10. Indian Heat-Treatment Economics

The Indian commercial heat-treatment industry is geography-first: Pune (auto and tooling — vacuum, carburizing, nitriding for die-casting tooling), Rajkot (castings and fasteners — bulk normalizing and Q&T), Ludhiana (cycle and auto components — the world's densest concentration of salt baths and belt furnaces), Faridabad/Ghaziabad (auto OEM tier-1), Coimbatore (pumps and motors). Indicative 2026 rates:

Process · Indicative rate

Normalizing / full annealing (bulk) · ₹10–20/kg

Hardening + tempering (Q&T, oil/water) · ₹15–40/kg

Carburizing + hardening + tempering · ₹40–80/kg

Carbonitriding · ₹30–60/kg

Gas nitriding · ₹100–250/kg

Induction hardening · ₹20–60/kg (volume-dependent)

Vacuum hardening (H13/D2 dies) · ₹150–350/kg

Cryogenic treatment · ₹50–150/kg

The economics lesson: for a 2 kg EN8 bracket, Q&T at ₹30/kg is noise. For a 150 kg H13 die at ₹250/kg vacuum-hardened, the ₹37,500 bill is cheap insurance against a ₹12-lakh die failing mid-run. Choose a treater like you'd choose a machine shop: certified pyrometry, atmosphere/vacuum capability matching your steel, and a willingness to show furnace charts. FabFlow's manufacturer directory includes verified fabrication and machining vendors — ask your shortlisted shops who their heat treater is before you place the order.

11. Two Worked Specifications

Example 1 — Gearbox shaft, EN19/4140, Ø40 × 300 mm. Normalize 860°C to refine the bar's hot-rolled structure. Austenitize 850°C, soak 60 min. Oil quench, agitated. Temper 600°C/2 h. Expect 28–32 HRC (UTS ≈ 900–1000 MPa) with good ductility — the standard Q&T condition for shafts, spindles, and bolts above 8.8 class.

Example 2 — Spur gear, 20MnCr5, module 3, 45 teeth. Carburize at 930°C, Cp = 1.0%, 4 h (worked example above gives ~0.75 mm enriched depth; effective case ~0.5–0.6 mm after hardening). Cool in the vestibule, reheat to 830°C, oil quench, temper 170°C/1 h. Result: 60–62 HRC flanks, 32–40 HRC core, ~15–20% retained austenite in the case (cryo or shot-peen if the application demands dimensional stability). Grind after hardening; expect 0.03–0.05 mm distortion on the bore.

The recurring theme: heat treatment is specified in temperatures, times, and cooling rates — and paid for in kilograms, microns of distortion, and points of Rockwell. Get the spec right and the furnace does the rest.

12. Further Reading

More FabFlow blog posts