Guide
How do you detect decarburization on hardened steel parts, and what limit should the drawing set?
Decarburization is the loss of carbon from the surface layer of steel while it is hot, because carbon at the surface reacts with oxygen, water vapor, carbon dioxide or hydrogen in the surrounding atmosphere and leaves as gas. The skin that is left cannot harden as fully as the core, so a hardened part shows a low surface hardness reading, wears early on a cam, bearing or sliding surface, or starts a fatigue crack at the surface, even though the core and the heat treat record look correct.
By Saola Manufacturing engineering team · Published · Updated
Decarburization is one of the few heat treat defects that a buyer can prevent almost entirely on paper. Where the soft skin came from, whether it is still on the finished surface, how it is measured and how much is allowed can all be written on the drawing and the purchase order. When none of that is written, the part can pass a core hardness check, ship, and fail in the field for a reason nobody tested for. This guide covers how to recognize it, what causes it, how it is measured under the published test methods, how to fix a lot that has it, and the note that keeps it off the next lot.
On this page
How do you recognize decarburization on a hardened part?
The classic sign is a hardness reading on the as-received surface that is below the drawing range, while a reading taken after lightly grinding the same spot, or on a cross section, comes back in range. In service it shows up as fast wear, scuffing or pitting on a surface that should have been hard, or as fatigue cracks that start at the surface.
Several symptoms point at a soft skin rather than a wrong heat treatment overall. Hardness rises as you test deeper: a light-load reading is lower than a heavier-load reading on the same surface, because the lighter indent samples more of the skin. A file bites the surface of a part that should resist it. Parts heat treated in the same load disagree with each other, because the ones near a furnace door or an air leak lost more carbon. And the problem sits on surfaces that were never machined after heat treatment, while ground faces on the same part read correctly.
Field failures follow the same logic. A decarburized skin is softer and weaker than the core, so it wears through first and is where fatigue cracks tend to begin. A study of leaf spring material published by SAE examined how the ferrite layer formed by decarburization affects fatigue resistance, alongside surface hardness, and traced the start of the problem back to the hot-rolled raw material (SAE technical paper 2013-01-0392). If a hardened shaft, pin, gear or spring fails early at the surface, ask for a decarburization check before blaming the load case.
Do not confuse decarburization with two look-alikes. A hardness reading taken through a plated or coated surface, or on a curved surface without a correction, also reads low; the guide to hardness scales covers those errors. An incomplete quench or the wrong tempering temperature also lowers hardness, but a traverse on a cross section then shows hardness that stays low, or falls further, toward the core, instead of a thin soft skin over a fully hard core.
What causes decarburization on heat-treated steel parts?
In rough order of how often it explains a problem part: a decarburized skin left over from hot-rolled or forged stock that was not machined off, a furnace atmosphere that oxidizes carbon during hardening, and too much time at temperature, including repeat heat treatment of rejected parts.
1. Skin carried in from the raw material. Hot-rolled bar and forgings come out of the mill or the forge with scale and a decarburized layer under it. If the machinist leaves a surface at the as-rolled size, or takes too little stock off it, that layer is still there when the part is hardened, and hardening cannot add the carbon back. The hot-rolled vs cold-rolled guide explains why the bar choice matters. Cold-finished bar has had its surface worked, but that is not a guarantee that it is free of decarburization, so the bar specification still needs to say so where it matters.
2. Furnace atmosphere. At hardening temperatures, carbon at the surface reacts with oxygen in the furnace gas and leaves as carbon monoxide; furnaces without a controlled atmosphere are the main source, and inert-gas furnaces can also decarburize when settings are wrong or seals leak (Fastener + Fixing, controlling decarburization). Moisture, an air leak or a carbon potential set below the steel’s own carbon content all push the reaction the wrong way.
3. Time and temperature. Carbon diffuses out faster at higher temperature, so long soaks, high austenitizing temperatures and slow heat-up in an open atmosphere all deepen the layer. Re-hardening a rejected lot adds a second exposure, which is why a rework hardening cycle can leave a part with a deeper soft skin than the original.
4. Hot working and forging heats. Parts forged or hot formed before machining pick up the same kind of layer. On a forging the decarburized skin is uneven, deepest where the part sat longest in the furnace or where scale was thick.
| Method | What it shows | When to use it |
|---|---|---|
| Screening | Separates samples with little or no decarburization from those with significant decarburization | Quick check of a suspect lot before sectioning |
| Microscopical | Ferrite-rich or carbon-poor layer on a polished, etched cross section | Needs a polished cross section; ASTM E1077 names its rigorous quantitative or lineal analysis version as the referee method in a dispute |
| Microindentation hardness | Depth to a specified minimum hardness, or to where hardness becomes uniform | Most suitable for hardened specimens with a reasonably uniform microstructure; matches a drawing limit written as a hardness |
| Chemical analysis | Carbon content of the surface layers, measured step by step inward; usually shows a greater depth than microscopical methods | Simple, uniform shapes only; generally reserved for research |
How do you measure decarburization depth on a finished part?
Cut a cross section through the suspect surface, polish it, and either read the soft layer under a microscope or run a microhardness traverse from the surface inward until the hardness reaches the core level. A standard Rockwell test on the surface can raise the alarm, but it cannot tell you how deep the layer is.
ASTM E1077 covers four families of procedure for estimating decarburization depth in any steel: screening, microscopical, microindentation hardness and chemical analysis, summarized in the table above. Its microindentation method can define depth either to a specific minimum hardness or to the point where hardness becomes uniform, which is why the drawing has to say which one it means. SAE J419, a recommended practice now stabilized, defines the types of decarburization with charts and micrographs and describes the measuring methods most commonly used (SAE J419), and ISO 3887:2023 defines decarburization and specifies three methods for measuring its depth (BS EN ISO 3887:2023).
Two depths are worth keeping apart. Complete decarburization is a skin with essentially all its carbon lost, which etches as plain ferrite and is easy to see. Partial decarburization is a deeper zone with less carbon than the core but not none, and on a hardened part its inner boundary is hard to judge by eye in the etched structure. ASTM E1077 rates the microindentation hardness method as most suitable for hardened specimens with a reasonably uniform microstructure, because the hardness keeps climbing until it reaches the unaffected core and the depth can be read off the curve.
Threaded fasteners have their own short method. SAE J121 covers measuring, classifying and specifying decarburization in the threaded section of hardened and tempered bolts, screws and studs (SAE J121), and for metric property classes the ISO 898-1 hardness method compares two Vickers HV 0.3 readings on a section through the thread, one near the flank surface and one in the base metal; the part passes when the reading near the surface is no more than 30 HV below the base metal reading (Fastener + Fixing). That two-point comparison is a useful model for any hardened part: compare surface and core on the same section, with the same indenter load.
Practical points for a receiving inspector: section a part, not a test coupon, if the dispute is about a specific surface; section across the surface that matters, such as the bearing journal or the cam lobe, not a convenient flat; and take the traverse on the finished surface, because grinding after heat treatment may already have removed the layer. A lab report should state the method, the indenter load, where the section was taken and how depth was defined.
How do you fix or prevent decarburization on hardened parts?
If the soft layer is shallower than the stock left on the surface, grind or machine it off after heat treatment; if it is not, the lot usually cannot be rescued by simply hardening it again. Prevention means removing the raw material skin before hardening and hardening in a protective atmosphere, vacuum or salt bath.
Use this decision table on a suspect lot before deciding to sort, rework or reject it. The controlling question is always the same: is there enough material left on the functional surface to remove the layer and still meet size?
| What you find | Likely origin | Fix for this lot | Prevention on the next lot |
|---|---|---|---|
| Soft layer only on surfaces left at bar or forging size | Raw material skin not removed | Machine or grind off if size allows; otherwise reject | Require stock removal from all functional surfaces before hardening |
| Soft layer on all surfaces, uneven across the load | Furnace atmosphere or leaks | Grind functional surfaces if stock allows; sort by hardness | Specify protective atmosphere, vacuum or salt bath; ask for atmosphere records |
| Deeper layer on reworked parts | Repeat hardening cycle | Section a reworked part before accepting | Require approval before any re-hardening |
| Layer deeper than the grinding stock | Any of the above | Material review: scrap, or controlled carbon restoration if the drawing allows it | Add finish stock to surfaces ground after heat treatment |
| Low surface reading, no layer on the section | Not decarburization | Check test method, coating and curvature; check quench and temper | Specify where and how hardness is measured |
Re-hardening alone does not cure decarburization, because the furnace cycle cannot put back the carbon that is gone, and it adds distortion and another exposure. Carbon restoration in a controlled carburizing atmosphere is possible, but it is a deliberate process with its own risk of overshooting into a carbon-rich skin, so it belongs only where the drawing or the customer approves it. Shot peening helps fatigue by adding compressive stress at the surface, but it does not restore hardness and is not a substitute for removing the layer.
The cheapest prevention is in the routing. Rough machine every functional surface from bar or forging, harden, then finish grind or hard turn the surfaces that carry wear or fatigue load. If the part must be hardened at finished size, the heat treatment must be done in a protective environment and the drawing must say so, because the default commercial process may not be. Where hardening also distorts the part, the stress relieving guide covers the sequence that keeps it straight.
What decarburization limit should the drawing and RFQ set?
Set the limit by surface: no complete decarburization and no measurable partial decarburization on surfaces that carry wear, contact or fatigue load, and a stated maximum depth elsewhere only if the design can tolerate it. Name the test method, how depth is defined, where it is measured and that it applies to the finished part.
There is no single depth that suits every part, which is why the published methods describe how to measure and classify decarburization and leave the acceptance value to the product specification. A sliding or rolling-contact surface usually allows none, because the skin is exactly where the wear happens. A non-functional surface on a hardened bracket may tolerate some. Decide surface by surface, then write it down. A note in this form is hard to misread:
- Material and condition. “AISI 4140 bar, heat treat to [hardness range] measured on the finished surface.” The 4140 vs 4340 guide covers how to state the hardness range.
- The limit. “Surfaces marked A: no complete or partial decarburization after finishing. All other surfaces: total decarburization not to exceed [depth].”
- How it is measured. “Depth by microindentation hardness traverse per ASTM E1077, defined as depth to [minimum hardness] or to uniform core hardness.” For threaded fasteners, cite SAE J121 or ISO 898-1 instead.
- Where and how often. “One part per heat treat lot sectioned through surface A; report with method, load and section location.”
- Routing controls. “Remove all mill and forging skin from functional surfaces before hardening. Harden in protective atmosphere, vacuum or salt bath. No re-hardening without written approval.”
Put the same items on the RFQ, so the quote prices the finish stock, the controlled heat treatment and the sectioned sample instead of discovering them after the first rejection. Ask for the heat treat record and the test report as part of the shipment documents, not on request after a failure.
Who does what on a Saola order with a hardened steel part?
Turning and milling run at our lead CNC plant; hardening is not on that plant’s process list. If your drawing calls for a hardened part, say so on the RFQ, and the reply will state whether and where hardening and any decarburization test can be arranged, with the production site and country of origin named on the quote.
Turned and milled features are cut on the lathes and machining centers at our lead CNC plant, as described on the CNC turning and CNC milling pages; grinding and polishing are on the plant’s equipment list, which matters when a decarburized layer has to come off after hardening. A hardness tester is on the same list, so a surface hardness check can be part of the inspection plan, but a cross-section microhardness traverse needs a metallurgical lab, and the quote will say where that test would be done rather than assume it. Any coating or plating after hardening goes to partner plants we manage through the surface finishing route, and the quote names where that work happens.
The most useful thing you can send is the drawing with the decarburization note already on it, the surfaces marked, and the routing intent: which surfaces are finished after hardening and how much stock you expect left on them. Material certification is available on request, and ISO 9001 certification is in progress, not held. Send the package through the quote form.
Frequently asked
Can a decarburized part be fixed by hardening it again?
No. Hardening cannot replace carbon that has left the surface, and a second cycle in the same atmosphere can deepen the layer. Remove the layer by grinding if stock allows, or use a controlled carbon restoration only where the drawing permits it.
Will a Rockwell C test on the surface detect decarburization?
It can flag it, because the reading comes in low, but it cannot measure depth and it can miss a thin layer under a heavy load. Compare a light-load and a heavy-load reading, then confirm on a cross section with a microhardness traverse.
Is decarburization the same as scale?
No. Scale is oxide sitting on the surface and comes off with blasting or pickling. Decarburization is a carbon-poor layer of steel under the scale, and it stays after the scale is gone.
Does carburizing or nitriding avoid the problem?
Carburizing adds carbon to the surface on purpose, so standards such as SAE J121 exclude deliberately carburized parts from their decarburization limits. Nitriding runs below the hardening temperature, but a decarburized layer left from earlier processing still lowers the result, so remove it before nitriding.
Who pays when a lot fails for decarburization?
That depends on what the drawing and purchase order said. If no limit, method or surface was specified, the supplier can argue the part met print; if the note was there, the lot is nonconforming. That is the practical reason to write the limit down.
Sources
- ASTM E1077-14(2021), Standard Test Methods for Estimating the Depth of Decarburization of Steel Specimens (scope), ASTM International, 2021
- SAE J419, Methods of Measuring Decarburization (stabilized), SAE International, 2018
- SAE J121, Decarburization in Hardened and Tempered Unified Threaded Fasteners (scope), SAE International, 1997
- BS EN ISO 3887:2023, Steels: Determination of the depth of decarburization, BSI Knowledge listing, 2023
- Controlling decarburisation (ISO 898-1 methods), Fastener + Fixing Magazine, December 2021
- Decarburization Effects on Fatigue Behavior of Leaf Spring Material, Karaagac et al., SAE Technical Paper 2013-01-0392, SAE International, 2013