Delta ferrite and hot cracking in welding stainless steels: ferrite number, Schaeffler, WRC

By Robert Lahnsteiner, International Welding Engineer (IWE) · Updated: November 2026

Because a small proportion of delta ferrite protects against hot cracks. Weld metal that solidifies fully austenitic cracks easily while it is still liquid or mushy between the grains. This is why most filler metals for austenitic steels are alloyed so that a few per cent of ferrite remain in the weld metal, typically around 5 %.

The cracks are caused by accompanying elements, above all sulphur and phosphorus. They are hardly soluble in austenite, accumulate in the residual melt between the dendrites and form low-melting films there. Nickel sulphide, for example, melts at around 645 °C. As the weld cools, it contracts and the grain boundaries tear open.

Ferrite absorbs sulphur and phosphorus much better and interrupts the continuous grain boundaries. If the weld metal solidifies as ferrite first, the residual melt stays cleaner and hot cracks become rare. This intended ferrite content can be recognised by the slight magnetism of a stainless steel weld.

What matters is the phase with which the weld metal starts to solidify. Susceptibility to hot cracking decreases markedly from fully austenitic to fully ferritic solidification.

Ferrite number and solidification mode both depend on the composition, but not always unambiguously. In practice, the ferrite number is nevertheless specified and measured, because it is standardised and easy to check.

Order of the solidification modes according to ISO 3581, Annex B. The FN ranges are guide values.

There are two kinds of hot cracks. Solidification cracks form in the weld metal itself, usually along the weld centreline or in the end crater. Liquation cracks form in the heat-affected zone or in the underlying bead when the next pass remelts the low-melting grain boundaries. The mechanism is the same in both cases.

The ferrite number (FN) is a standardised characteristic value measured magnetically according to ISO 8249, not a volume fraction. At low contents, FN and per cent ferrite are close together; as ferrite increases, the FN is noticeably higher. Anyone who equates FN with per cent overestimates the ferrite content.

In the past, ferrite was counted in the metallographic section. Because it is very fine, irregular and unevenly distributed, the results scattered widely. A common scale was therefore agreed: measuring instruments are calibrated with standards to which a ferrite number is assigned. In round-robin tests, different laboratories determine the same specimen to within about ± 1 FN, in the range from 0 to 28 FN.

For duplex steels, the scale extends to 100 FN. For his range of 0 to 18 FN, DeLong gives about 0 to 14 % ferrite.

All three combine the ferrite-forming elements into a chromium equivalent and the austenite-forming ones into a nickel equivalent. For the ferrite content of austenitic and duplex weld metal, the WRC-1992 diagram is the most accurate. The Schaeffler diagram remains indispensable for joints with carbon steel, because only it shows martensite.

Contents in % by mass. Accuracy according to DVS 0946 and ISO 3581, Annex B. The diagrams apply to undiluted weld metal in arc welding.

• Schaeffler was derived from tests with 5 mm covered electrodes. It shows the regions of austenite, ferrite, martensite and their mixtures and is therefore suitable for determining dilution and the appropriate filler metal for dissimilar joints with carbon steel. Nitrogen is missing from the formula.
• DeLong is an enlarged section of the Schaeffler diagram for low ferrite contents and includes nitrogen, which lowers the ferrite number considerably.
• WRC-1992 no longer counts manganese but includes copper, and extends to 100 FN. It therefore also covers duplex weld metal.

The three diagrams can be calculated directly:

• Schaeffler, DeLong, WRC-1992 with two base materials and filler metal, including dilution
• Schaeffler diagram without filler metal, for example for TIG or laser welding without wire
• Schaeffler diagram for the range of the standard analysis: where a grade can lie within its analysis limits

The ferrite content is determined not only by the wire but also by dilution, shielding gas, arc length, cooling rate and heat treatment. If the measured value differs greatly from the data sheet, the cause usually lies here.

• Nitrogen pick-up: Nitrogen is a strong austenite former. An arc that is too long or disturbed gas shielding draws atmospheric nitrogen into the weld pool and can lower the ferrite number considerably. With duplex, on the other hand, nitrogen is deliberately added to the shielding gas.
• Shielding gas: Oxygen in the gas burns off chromium, CO₂ introduces carbon. Both shift the microstructure towards austenite, see Shielding gases.
• Dilution: The melted base material changes the analysis. With duplex, the ferrite content usually rises as a result; with dissimilar joints with carbon steel, it falls.
• Cooling rate: The faster the weld cools, the less ferrite transforms into austenite in duplex. In laser beam welding, on the other hand, Schuster found less ferrite in austenitic weld metal than in arc welding; there the diagrams no longer apply, and alloys close to the boundary can change their solidification mode.
• Multi-pass technique and heat treatment: Each further pass anneals the one below and lowers the ferrite content there. Post-weld heat treatment can reduce the measured ferrite number to zero.

As much as necessary against hot cracks, as little as possible for toughness and corrosion resistance. For standard austenites, a few FN have proved successful; for duplex weld metal, 30 to 70 FN.

• Standard austenites (308L, 316L): a few FN, often agreed as a range such as 3 to 8 or 5 to 10 FN.
• Fully austenitic (e.g. 310, grades such as 1.4439 and 1.4539): practically ferrite-free. Requiring zero FN is not realistic, a maximum of 0.5 FN, however, is.
• Duplex: 30 to 70 FN according to DVS 0946. Below that, resistance to stress corrosion cracking suffers, above it toughness. Up to 85 FN in TIG weld metal is generally still considered acceptable. More in Duplex and super duplex.

Too much ferrite is harmful for three reasons: it lowers toughness, especially at low temperatures. In prolonged service around 475 °C, it decomposes and embrittles (475 °C embrittlement); at higher temperatures or during heat treatment, brittle phases such as the sigma phase can form. And in some media it is attacked preferentially. This is why some codes set an upper limit for the ferrite content, for example for low-temperature applications.

Ferrite ranges should not be agreed too narrowly. According to ISO 3581, 5 to 10 FN or 40 to 70 FN are realistic, whereas 5 to 6 FN or 45 to 55 FN are not. Process and measurement scatter more.

With a magnetic measuring instrument calibrated according to ISO 8249. Measurements are taken at a smoothed spot in the middle of the weld, not at edges, on rough ripples or next to magnetic base material.

Within a weld, the ferrite content varies naturally: where beads overlap, the next pass anneals the previous one and lowers the ferrite. Individual values are therefore less meaningful than the mean of several measurements. Curved surfaces and thin sheet distort the result.

Procedure and assessment are described in Testing, health and safety.

With a filler metal that solidifies primarily as ferrite, low heat input and clean working. With fully austenitic steels, which must not form ferrite, heat control and cleanliness are particularly important.

• Choose a filler with ferrite, i.e. 308LSi for 1.4301 and 316LSi for 1.4404, not a fully austenitic wire.
• Limit heat input and observe interpass temperatures. Small weld pools solidify quickly and segregate less.
• Stringer beads instead of weave beads: they are narrower and shrink with lower stresses.
• Fill end craters and grind out crater cracks at tack welds.
• Work cleanly: Sulphur from oil, grease, marker pens or hand sweat easily gets into the weld pool. Do not touch cleaned weld areas with bare hands.
• Take dilution into account: For joints with other materials, check the ferrite of the diluted weld metal with the Schaeffler calculator.

Fully austenitic fillers such as 310 (ML 25 20) are needed for heat-resistant steels or when no magnetism is permitted. In that case, everything said above applies particularly strictly.

Why is my stainless steel weld magnetic but the base material is not?

Because the weld metal intentionally contains some delta ferrite, which is magnetic. This is not a defect but the protection against hot cracks.

Is 8 FN the same as 8 % ferrite?

Approximately. At low values, FN and per cent are close together; at higher values, the FN is above. Specifications should always state the FN, because only it is measured reproducibly.

Which diagram should I use?

For the ferrite content of austenitic and duplex weld metal, the WRC-1992 diagram; for dissimilar joints with carbon steel and the question of martensite, the Schaeffler diagram. The calculator for Schaeffler, DeLong and WRC calculates all three at once.

Do the diagrams also apply to laser welding?

Only to a limited extent. Because of the very rapid cooling, less ferrite remains than the diagrams indicate, and alloys close to the boundary can solidify differently. Only a measurement on the specimen helps here.

The series from the material via process, gas and root protection to the tested weld. For an overview, see Welding of stainless steel.

1. What makes stainless steel stainless
2. Choosing the filler metal
3. Processes at a glance
4. Shielding gases
5. Heat input and interpass temperature
6. Joint preparation and tack welding
7. Root protection and purging
8. Temper colours, pickling, passivation
9. Duplex and super duplex
10. Joining stainless to carbon steel
11. Weld defects and corrosion damage
12. Testing, health and safety