Author: Site Editor Publish Time: 2026-10-20 Origin: Site
A shaft drawing arrives with three lines that matter for heat treatment: "4140, Ø60, surface 55 HRC min, case 2–3 mm". The buyer assumes those lines are complete. The heat treater reads them and immediately has four questions the drawing does not answer — which case depth definition, measured to what hardness limit, at what frequency, and what happens at the keyway corner. Those four questions decide whether the part comes out of the quench at 57 HRC or in two pieces.
This article is about induction hardening of round bar as it is actually bought and processed: which grades respond, how frequency and power density set the case, how the depth is defined and measured, and the specific geometry and metallurgy conditions under which a case cracks. References are to ISO 2639, ISO 3754, ASTM A304, ASTM E384, ASTM E3 and ASTM A29 where the numbers come from a standard. Where a figure is shop practice rather than a standard value, it is labelled as such.
Induction hardening heats a surface layer by eddy currents, then quenches it. Only the skin transforms to martensite; the core stays as supplied. The case depth you get is set by frequency, power density, heat time and the steel's hardenability — in that order of adjustability, but in the reverse order of importance. A 0.40 % C chromium-molybdenum bar will take a 2 mm case at 10–20 kHz; a 0.15 % C mild steel bar will not take a useful case at any frequency, because there is not enough carbon to harden.
| Question from the drawing | What actually decides it | Where it goes wrong |
|---|---|---|
| Case depth 2–3 mm | Frequency, power density, heat time, and the hardenability of the grade | Depth quoted as a single number without a hardness limit or a measurement method |
| Surface 55 HRC min | Carbon content of the grade, then quench severity | Grade with 0.35 % C cannot reach 55 HRC as-quenched, whatever the coil does |
| Core stays tough | Case-to-core ratio; the core must not be austenitised | Case deeper than about a third of the radius starts to harden the core too |
| No cracks | Section transitions, pre-hardening microstructure, tempering right after quench | Sharp shoulder, banded centre, or a part left untempered overnight |
An induction coil around a bar is a transformer with a single-turn primary and the bar as a shorted secondary. Alternating current in the coil induces current in the bar surface; resistance turns that current into heat. Heating is therefore confined to the layer where the current flows, and that layer is thin. Its thickness falls as frequency rises, which is the whole reason the process can put 0.5 mm of martensite on a 20 mm bar and 4 mm on a 200 mm roll.
There is a discontinuity in that simple picture, and it is the one that catches people new to the process. Below the Curie temperature the steel is magnetic, current penetration is shallow, and heating is concentrated in a thin skin. Once the surface passes roughly 720–770 °C it becomes non-magnetic, penetration depth jumps by a large factor, and heat starts to flow inward much faster. That is why induction heating is not linear in time: the first seconds heat a skin, and the next seconds push the austenite front inward. A recipe copied from a similar part with a different power supply will land in a different place, because what matters is the energy delivered per unit area per second, not the frequency alone.
The consequence for the buyer is that case depth is a process result, not an intrinsic property of the grade. Two suppliers using the same 4140 bar can deliver 1.8 mm and 3.2 mm cases and both be "correct" against a vague drawing. If the depth matters to fatigue life or to subsequent grinding, it has to be specified with a tolerance and a definition, and it has to be verified on a sectioned part rather than on the surface with a portable hardness tester.
Frequency picks the ballpark; power density and time place the case inside it. A 10 kHz supply run at low power for a long time gives a shallow case with a wide heat-affected zone underneath. The same supply at high power for two seconds gives a sharper transition and less growth of the core temperature. That trade — sharp case with high residual stress, or graded case with lower stress — is the real process decision, and it is made at the machine, not in the purchase order.
The table below is our own starting range for single-shot and scanning coils on round bar. It is shop practice, not a standard, and the numbers move with coil coupling gap, bar diameter and the grade's hardenability. Use it to sanity-check a supplier's proposal, not to write a specification.
| Frequency | Typical case depth (shop practice) | Typical bar diameter | Where it is used |
|---|---|---|---|
| 1–3 kHz | 3–8 mm | 80–300 mm | Large rolls, mill pinions, heavy shafts where deep case is required |
| 3–10 kHz | 2–5 mm | 50–200 mm | General shaft work on medium and large bar |
| 10–30 kHz | 1–3 mm | 25–100 mm | Most automotive and general engineering shafts; the common middle ground |
| 30–100 kHz | 0.6–2 mm | 12–60 mm | Small shafts, pins, splines, wear surfaces with limited grinding allowance |
| 100–400 kHz | 0.2–1 mm | 4–25 mm | Thin sections, small pins, edges that must not grow |
Two rules of thumb survive contact with the shop floor. First, case depth and bar diameter are linked: a 4 mm case on a 30 mm bar means the hardened layer is more than a quarter of the radius, the core gets hot, and the part loses the toughness gradient that induction hardening was chosen for. Second, the case never ends abruptly. Under the martensite there is a transition zone of partially transformed material, and that zone is where hardness traverses look untidy and where fatigue cracks initiate if the transition sits at a stressed fillet.
Surface hardness after induction hardening is capped by carbon. The as-quenched hardness of martensite rises with carbon up to about 0.6 %, and beyond that it flattens while the risk of quench cracking climbs. Practical ceilings, before tempering: a 0.45 % C grade such as 1045 reaches roughly 58–60 HRC; a 0.40 % C grade such as 4140 / 42CrMo4 round bar reaches the high 50s; a 1.0 % C bearing grade such as 52100 reaches 64–66 HRC. A 0.20 % C carburising grade tops out around the mid-40s and will never make a 55 HRC case, no matter how the coil is tuned.
Hardenability, not just carbon, decides whether the case is through-hardened at depth. Chromium and molybdenum widen the transformation range so a thicker layer converts to martensite at a given quench. That is why 4140 is the default induction grade for shafts and 1045 is the cheap alternative used where section is small and cost dominates. Nickel-bearing 4340 / 1.6582 bar hardens deeper still and keeps a tougher core, which is why it appears in highly stressed shafts where a 1045 part would snap rather than bend.
| Grade | Carbon, ASTM A29 range | Response to induction hardening | Watch out for |
|---|---|---|---|
| 1045 / C45 | 0.43–0.50 % | Good case hardness, cheapest option, needs water or fast polymer | Crack-sensitive at keyways and sharp shoulders; low core toughness at large diameter |
| 4140 / 42CrMo4 | 0.38–0.43 % | Deep, even case; forgiving on quench; the general-purpose choice | Decarburised skin from rolling will not harden — specify peeled or ground bar |
| 4340 / 1.6582 | 0.38–0.43 % | Deepest case of the three, best core toughness | Cost; needs oil or polymer, not water |
| 52100 / 100Cr6 | 0.98–1.10 % | Highest surface hardness; standard for bearing races and wear plates | Must be spheroidise annealed before hardening; high cracking risk if case is deep or tempering is late |
| 1144 / 11SMn30 | 0.40–0.48 % / ≤0.14 % | Free-machining grades harden, but sulfur raises crack sensitivity | Sulfide stringers act as crack paths; avoid deep cases and water quenching |
| 410 / 420 / 440C stainless | 0.08–0.15 / 0.16–0.25 / 0.95–1.20 % | Martensitic grades respond; 440C reaches high hardness | Ferritic and austenitic grades (430, 304, 316) do not harden by transformation at all |
"Case depth" is ambiguous by itself, and the ambiguity causes more arguments between buyer and heat treater than any metallurgical problem. Three definitions are in circulation, and on a 4140 part with a 55 HRC surface they can differ by nearly a millimetre on the same specimen.
ISO 2639 handles carburised and case-hardened parts and defines the case-hardening depth as the distance from the surface to the point where hardness falls to 550 HV1. The 550 HV1 limit is roughly 52–53 HRC, and it was chosen because it corresponds to a structure that is no longer predominantly martensite. ISO 3754 covers flame and induction hardened parts and leaves the limit hardness to be agreed, but most drawings inherit the 550 HV1 convention from ISO 2639 without saying so. Then there is total case depth, measured to the point where hardness reaches the core value — that number is always larger, sometimes by 40 %, and it is useless for fatigue calculations.
| Definition | Reference | Limit used | Typical use |
|---|---|---|---|
| Case-hardening depth (CHD) | ISO 2639 | 550 HV1 | Carburised parts; often quoted by habit on induction drawings |
| Effective depth of hardening | ISO 3754 | By agreement; commonly 550 HV1 or 80 % of surface hardness | Induction and flame hardened parts — the number to specify |
| Total case depth | Works practice | Core hardness + 50 HV | Non-destructive estimation, weld overlay work; not for fatigue design |
| Surface hardness | ISO 6507-1 / ASTM E92 | Minimum HRC or HV at 0.5–1 mm below surface | Wear resistance; always quote the depth at which it was measured |
The practical fix is one sentence on the drawing: "Effective case depth per ISO 3754, limit 550 HV1, 2.0–3.0 mm, measured on a longitudinal section at mid-length." That sentence removes three sources of dispute — the definition, the limit hardness and the location of the measurement. Without it, a supplier measuring total depth on a transverse section can report 3.5 mm for a part that fails an effective-depth check at 2.4 mm.
The quench is where the case becomes martensite and where most of the residual stress is locked in. Water gives the highest surface hardness and the highest cracking risk; polymer solutions sit in between and are the usual choice for 4140; oil is slower and safer for alloy grades with deep cases. The choice is normally the heat treater's, but it becomes the buyer's problem when the drawing demands 58 HRC on a 1045 part with a keyway.
Immediately after the quench, the part is not finished. The surface is hard and brittle, and the core still holds heat. Two things can be done with that heat: let it temper the case from inside (self-tempering, used on simple shafts where 1–3 HRC of surface hardness can be given up), or quench the part out fully and temper in an oven. What must not happen is the part sitting on a bench overnight. Untempered martensite plus residual stress plus a small amount of hydrogen is a well-known recipe for delayed cracking, and the crack usually appears hours later, at a change of section, when nobody is watching.
For parts that will be ground after hardening, tempering is not optional. A 150–200 °C temper for one hour per 25 mm of section relieves enough stress to survive grinding without a network of surface cracks, and it costs far less than scrapping a finished shaft. Where the application needs toughness in the case as well as hardness — splines under shock load, for example — the tempering temperature goes up and the surface hardness requirement comes down, and that trade belongs in the discussion before the order, not after the first hardness report.
Induction cracks are rarely random. In our experience they cluster at four places, and all four are visible before the part reaches the coil if someone looks.
Section transitions. A shoulder, a keyway root or a groove concentrates both heat pattern and stress. The corner heats faster because it has less mass around it, so it austenitises deeper while its neighbours stay cold, and it is the last place to be tempered by the core. Radius the corner at 0.5 mm or more, or move the coil so the case ends before the transition. A sharp internal corner with a 3 mm case on a 40 mm shaft will crack, and no amount of quench tuning will save it.
Banding and segregation in the bar centre. Continuous cast or poorly soaked bar can carry carbon and alloy banding along the rolling direction. Bands of higher carbon transform at a different time from their neighbours, and the interface between them is a crack path. This shows up on a macro etch of the bar before heat treatment; it is a raw material acceptance issue, not a heat treatment one, and it is one reason we ask for the melt route and, on critical parts, a macro etch result.
Surface condition of the incoming bar. Decarburisation is the quiet one. A decarburised rim on hot rolled 4140 has less carbon than the core, so the case there cannot reach the specified hardness and, worse, the transition between decarburised and sound material sits inside the hardened zone. Seams are the loud one: a seam that survives turning opens under the stress of the quench. This is the main reason induction-hardened parts are usually made from peeled or ground bar rather than hot rolled bar with scale.
Grinding after hardening. A hard case ground with a wheel that is too soft or a coolant that is too lean burns the surface. The burn is a tempering effect — a thin layer softer than the case below it, in tension — and it appears as a network of fine cracks after Nital etch. The part passes a hardness check at 0.5 mm and fails a crack inspection at the surface, which is the worst combination for everyone downstream.
| Symptom | Most likely cause | Countermeasure |
|---|---|---|
| Longitudinal cracks along the bar axis | Seams or laps under the case; sulfur stringers in free-machining grades | Peeled or ground stock; avoid free-machining grades for high-hardness cases |
| Crack at shoulder or keyway corner | Stress concentration plus deeper heating at the corner | Radius the corner, relocate the case end, lower power density at the transition |
| Fine network on the surface after grinding | Grinding burn, or untempered martensite | Temper before grinding; re-dress wheel; check coolant flow |
| Crack appearing hours after quench | Delayed cracking in untempered martensite | Temper within the hour; do not stack parts warm |
| Hardness low but case depth on target | Decarburised surface; insufficient carbon in the grade | Specify decarburisation limit on the bar; check grade carbon against the hardness target |
The only honest way to verify a case is to cut a part. A hardness traverse on a polished cross-section, using a Vickers indenter at 1 kg or a Knoop indenter for thin cases, gives the curve; the depth at which it crosses the agreed limit is the answer. Microhardness testing is covered by ASTM E384 and ISO 6507-1, and the specimen preparation matters as much as the tester — a poorly polished section reads soft and you will chase a heat treatment problem that is really a metallography problem. Sectioning and mounting practice is in ASTM E3.
Hardness alone does not show the case boundary clearly on alloy steels, so the standard companion check is a Nital etch: 2–5 % nitric acid in alcohol, a few seconds, and the hardened zone darkens against the core. That etch also reveals seams, decarburisation and grinding burn, which is why it is worth doing even when the hardness traverse looks acceptable. On a production run we section one part per batch and per setup; on a first article, we section at the smallest diameter and at the largest, because those are the two extremes the coil sees.
A portable Leeb or ultrasonic hardness tester is fine for sorting parts on the line and useless for case depth. It reads a weighted average of the material under the probe — on a 1 mm case that means it is partly reading the core. If a supplier offers a portable reading as evidence of case depth, ask for the traverse.
Three properties of the incoming bar change the outcome, and all three can be written into the purchase order.
Decarburisation. Specify a maximum depth, or specify peeled or ground stock where the drawing cannot tolerate any. On hot rolled alloy bar we report decarburisation depth on request per ASTM E1077 rather than quoting a blanket figure, because it varies with size and rolling campaign.
Microstructure. For 52100 and other high-carbon grades, spheroidise annealing is a prerequisite — a pearlitic structure will not dissolve uniformly in a two-second heat, and the result is a patchy case with retained carbide and soft spots. For medium-carbon grades, a normalised structure gives a more predictable response than an as-rolled one with banding.
Residual stress and straightness. Cold drawn bar carries drawing stress. Induction hardening a cold drawn shaft without a stress relief step gives a part that bends as the case transforms, and the bow shows up after grinding. Where straightness after hardening matters, start from annealed or peeled and straightened stock, and say so.
A bar order and a heat treatment order are two documents, and the failures happen in the gap between them. Six lines close most of that gap:
Grade with the standard edition — "4140 per ASTM A29, current edition", not "4140". Diameter and tolerance class, with the condition — hot rolled, peeled or ground. Decarburisation limit or "decarb free". Hardenability: where the case must be reliable at depth, order an H-grade to ASTM A304 limits rather than a standard grade, because the standard only controls chemistry and two heats with identical chemistry can harden to different depths. Surface and internal quality: EN 10221 surface class and, for highly stressed shafts, ultrasonic inspection per EN 10228-3 with an acceptance class. Documentation: inspection certificate to EN 10204 3.1, with the actual chemistry of the heat that made the bar.
Then the heat treatment specification, separate and explicit: effective case depth with its definition and limit hardness, surface hardness range with the depth at which it is measured, core hardness maximum, tempering temperature, and the inspection method. When those two documents agree, the arguments stop.
No. Austenitic grades do not transform to martensite on cooling, so there is no hardening mechanism to trigger — heating and quenching leaves them at roughly the same hardness. If a part needs a hard surface and corrosion resistance, the martensitic grades are the ones to use: 410 and 420 for moderate hardness, 440C where 58 HRC and above is needed. Free-machining ferritic grades such as 430F harden even less and are chosen for machinability, not wear resistance.
Less than through hardening, because only the skin transforms, and that is the main reason the process is used on finished-machined shafts. But "less" is not "none". The martensite transformation expands the case, the part grows slightly in diameter and, if the case is asymmetric around the section — one side of a flat, a keyway — it bends. Leave grinding allowance on diameters that must hold a tolerance, and expect straightening to be needed on long slender parts. Any figure quoted for growth should come from the heat treater's own trials on your geometry, not from a table.
Only by grinding, and only within the allowance. Turning or milling a 55 HRC case is possible with ceramic tooling but is slow and tends to leave tensile residual stress at the surface. The practical sequence is: rough machine, heat treat, grind, and — where the drawing allows — finish any soft features before hardening or mask them from the coil.
Where the case depth is critical and the part is made in quantity, yes. A standard 4140 heat is only guaranteed on chemistry; its hardenability can sit anywhere in a wide band, and two heats can give different case depths in the same fixture. An H-grade such as 4140H is ordered to hardenability limits under ASTM A304, so the response to the coil is repeatable from lot to lot. For one-off parts with generous tolerances, standard grade is fine.
Most induction problems we are asked about turn out to be bar problems: a decarburised skin, a seam that survived turning, banding in the centre, or a grade chosen for price that could not reach the hardness on the drawing. Those are decided before the bar is cut, at the order stage.
We supply 4140, 4340, 1045, 52100 and the martensitic stainless grades as hot rolled, peeled, cold drawn or ground round bar, with chemistry from the heat reported to EN 10204 3.1, decarburisation depth on request, and H-grade hardenability on order. If you are working through a case depth requirement, send the drawing line and the finished diameter — the grade and the bar condition usually follow from those two numbers. Enquiries: enquiry@qilumetal.com.
