ISO 18203 — determination of the thickness of surface-hardened layers

Scope — and what it does not cover

ISO 18203 is the document that now carries the depth-of-hardening definitions for steel, in one place and under one number: case hardening depth (CHD) after carburising and carbonitriding, nitriding hardness depth (NHD) after nitriding and nitrocarburising, and surface hardening depth (SHD) after induction or flame hardening. For each it fixes what the limit hardness is, how the microhardness traverse is made on a prepared cross-section, and how the depth is read from the result. It does NOT decide what depth your part needs, does not choose a limit hardness for you when the default does not fit the material, does not cover layers whose thickness is measured metallographically rather than by hardness, and does not replace the drawing — which still has to carry the designation, the depth range, the limit hardness, the test force and the measurement location.

Standards are copyrighted documents. We explain scope, intent and practical application in our own words; obtain the current edition from ISO, ASTM, DIN or your national standards body.

What actually bites in production

• 1This is the number the case-depth definitions live under today. A drawing that cites ISO 2639 for a carburised case, or ISO 3754 for an induction-hardened one, is citing a document this one superseded. The measurement principle survived the renumbering, so a legacy callout is still workable — but the current number is the one a new drawing should carry, and the ISO 2639 entry in this reference is where the CHD traverse itself is worked through in detail.
• 2Three processes, three different limit-hardness rules — and only one of them is a fixed number. 550 HV is a convention agreed for carburised cases, not a physical boundary. NHD is taken to the core hardness plus 50 HV. SHD is taken to 80 % of the minimum surface hardness the drawing specifies. Quoting "550 HV1" on a nitrided or induction-hardened part therefore asks for a measurement the standard does not define.
• 3Two of the three limits move with the part. Raise the specified minimum surface hardness on an induction-hardened item and the SHD limit hardness rises with it, so the same physical layer is reported as a shallower depth. Temper a nitrided part softer and its core hardness drops, the NHD limit drops with it, and the same layer reads deeper. Neither is a measurement error; both are the rule working as written.
• 4The depth is interpolated, not picked. Indentations land where the spacing rules of the hardness standard allow, almost never exactly on the limit hardness, so the depth is read between the two indentations that bracket it. Rounding to the nearest indentation is a common way to lose 0.02–0.05 mm of reported depth on a shallow case.
• 5The traverse has geometry rules that decide whether the number means anything: the section is cut perpendicular to the surface, the indentations sit inside a defined narrow band rather than wandering across the face, the edge must not be rounded off during polishing, and the depth is measured from the FINAL surface — so a part that is ground or thread-rolled after hardening is measured from the surface it leaves the shop with, not the one it had in the furnace.
• 6One of the three is ours to measure; the other two arrive on drawings. We carburise and carbonitride small fasteners in bulk, so CHD is a measurement our laboratory makes on its own work. Nitriding, induction hardening and flame hardening are not processes we run — NHD and SHD are documented here because customer drawings carry them and the definitions have to be read correctly before anyone quotes. Compound-layer or white-layer thickness on a nitrided part is a metallographic measurement, a different quantity from NHD, and has to be specified separately when it matters.

What to write on the drawing or PO

• The whole string in one line: designation, limit hardness, test force, depth range — "CHD 550 HV1 = 0.15–0.30 mm". Four quantities, not one number, and the standard number beside it with the edition in force if your quality system requires the edition to be named.
• The measurement location and the section plane: thread flank, root, crest, shank, head-to-shank fillet, or a coupon that ran with the load, with a sketch when the geometry gives more than one honest answer.
• Surface hardness and core hardness as their own requirements, because NHD and SHD limits are derived from them. On an induction-hardened part the minimum surface hardness is doing double duty — it is also setting the depth criterion.
• Which surface is "the surface": as-hardened, after grinding, after thread rolling. The depth is measured from the finished surface, so any stock removed after hardening changes the number.
• Sampling and reporting: parts per lot, whether sacrificial parts or process coupons are acceptable, and whether the hardness profile itself — the traverse readings, not only the calculated depth — has to be delivered with the lot.

What goes wrong when it is cited loosely

• Writing "case depth per ISO 18203 = 0.3 mm" and stopping there. The standard defines three depths with three different limits; a bare number under this reference number is less specific, not more.
• Carrying 550 HV1 across to a nitrided or induction-hardened part because it is the figure everyone recognises. It belongs to carburised cases; the other two designations have their own rules and a supplier following the standard will not use it.
• Reporting the depth at the nearest indentation instead of interpolating between the two that bracket the limit hardness — a difference that hides inside the noise on a deep case and dominates the result on a shallow one.
• Tightening the minimum surface hardness on an induction-hardened part and treating the SHD requirement as untouched. Raising the surface requirement raises the limit hardness, which shortens the depth that gets reported, and a part that passed both requirements yesterday can fail one of them today with nothing changed in the furnace.
• Specifying a hardened-depth designation for a layer whose thickness is not defined by hardness at all — a compound layer, a coating, a plated thickness. Those are measured by other methods and belong on their own line.