Ductile iron is a promise: cast-iron economics with steel-like ductility. The promise holds only if the graphite actually solidified as spheres. Nodularity is the number that says whether it did — and it is the first thing a customer's metallurgist checks on a ductile iron certificate.
Why shape is everything in cast iron
Graphite is mechanically weak. In grey iron it forms interconnected flakes, and every flake tip is a stress concentrator — which is why grey iron is stiff and damps vibration beautifully, but breaks rather than stretches. Treat the melt with magnesium and the same carbon crystallizes as isolated spheroids. Spheres have no sharp tips; the iron matrix carries load around them, and elongation jumps from almost nothing to double digits. Between those extremes lies compacted (vermicular) graphite — a controlled intermediate with its own uses and its own specification battles.
The ISO 945-1 form classes
ISO 945-1 classifies graphite particles by shape into six reference forms:
| Form | Morphology | Typically seen in |
|---|---|---|
| I | Lamellar (flake) | Grey iron |
| II | Spiky, crab-like | Degenerate structures |
| III | Vermicular (compacted) | CGI; Mg-fade in ductile iron |
| IV | Aggregated, temper-carbon-like | Malleable iron |
| V | Imperfect, partly spheroidal | Borderline ductile iron |
| VI | Regular spheroids | Good ductile iron |
ASTM A247 covers the same territory with its own reference charts of graphite type, size and distribution. In day-to-day foundry language, "nodular" means forms V and VI.
How the percentage is computed
Nodularity is measured on the unetched polished section — graphite reads dark against the bright matrix, so no etching is needed and none should be used. Conceptually the measurement is simple:
The work is in the classification behind that ratio. Each detected particle is assessed for how closely its outline approaches a sphere's section — image analysis uses shape descriptors (roundness/compactness) for exactly this — and assigned a form. Very small particles below a minimum size are excluded so that polishing debris and micro-porosity don't masquerade as nodules, and particles cut by the image edge are treated consistently. Then the areas are summed and the ratio falls out.
Specifications commonly require nodularity of 80% or higher for ductile iron grades, with premium and safety-critical applications asking for more. Alongside the percentage, certificates usually quote nodule count (nodules per mm²) — a high count of small, well-formed nodules generally signals good inoculation and a fast-cooled, sound structure.
What makes the number move
- Magnesium fade. Mg boils off with time after treatment. Pour too late and forms III and V creep in — the classic cause of a failing heat that "looked fine at treatment".
- Section thickness. Slow-cooling heavy sections give graphite time to degenerate; the same heat can pass in a test bar and fail in the casting's thermal center.
- Inoculation practice. Weak inoculation lowers nodule count and worsens form.
- Where you cut the sample. Nodularity is a local property. Surface, center and riser-adjacent metal can differ meaningfully; agree the sampling location with your customer up front.
- Measurement settings. The minimum-size cutoff, edge-particle handling, and how strictly form V is scored all shift the result by real percentage points. Whatever software you use, keep those settings fixed and auditable between heats — a nodularity number without its settings is only half a number.
Reading a certificate critically
When a report claims "Nodularity 92%, Form VI, 250 nodules/mm²", the informed reading is: the graphite is overwhelmingly well-formed spheres, at a healthy count, measured — not estimated. If instead you see a bare percentage with no form breakdown, no count and no field information, ask how it was measured. The difference between an eyeball estimate against a wall chart and a per-particle classified measurement is exactly the difference between hoping and knowing.
Classify a heat you already know
Bring your reference samples to a demo and compare the module against your bench call.