Stainless Steel Castings from China: CF8M Is a Grade, Not a Guarantee

A pump importer in Rotterdam sent me a failed casting last spring. It was supposed to be CF8M, the workhorse austenitic stainless casting grade, and on paper it was. The mill certificate said so. The problem was that the casting had been poured from a heat where the molybdenum had drifted down and the ferrite content sat in the wrong band, and the part had cracked at a section change after about fourteen months in slightly chlorinated cooling water.

Nothing in the paperwork was falsified. The grade was met on the elements the certificate lists. The casting was still wrong. That is the recurring theme when buying cast stainless from China, and it has very little to do with price per kilogram.

CF8M and CF3M in plain terms

CF8M is the cast equivalent of 316, with roughly 18 to 21 percent chromium, 9 to 12 percent nickel and 2 to 3 percent molybdenum, and carbon capped around 0.08 percent. CF3M is the low-carbon version, carbon at 0.03 percent max, which matters when the part will be welded and cannot be properly solution-annealed afterwards. If your part has field welds or a heavy weld repair, CF3M is the safer call even though it costs a little more and is sometimes harder to source from smaller foundries.

The two numbers that decide service life are not on the customer drawing. The first is the ferrite content in the as-cast microstructure. Too little ferrite and the weld zones become susceptible to hot cracking and stress corrosion cracking; too much and you lose toughness and corrosion resistance. A ferrite content somewhere around 5 to 15 percent measured on the casting is the usual target for these grades, and a foundry that cannot tell you the measured value is guessing.

The second is the solution anneal. Castings should be annealed at roughly 1040 to 1120 degrees Celsius and then water quenched to put carbides back into solution. Small foundries sometimes air-cool or skip the step for thin parts to save a heat cycle. The casting still passes a hardness check and still machines. It just has a sensitised microstructure that will fail at the grain boundaries in a chloride environment. Ask for the heat treatment record with furnace chart for the batch, not a generic procedure document.

Investment versus sand casting

Investment casting gives you thin walls, tight tolerances and a good surface, and it is what most Chinese shops quote for valve bodies and pump parts under a few kilograms. Sand casting handles larger, heavier shapes and is cheaper per kilo above a certain size, but the tolerances are looser and the surface needs more machining allowance.

Where buyers get caught is the pattern and the tooling. A sand-cast part needs a pattern, and a cheap pattern in soft tooling will drift over a few thousand shots. Investment casting needs a wax die, which is expensive but lasts. Ask which process the quote assumes, what the tooling cost covers, and who owns the tooling. The last one matters: if the tooling is the foundry's, switching suppliers later means paying for it again or starting from drawings.

Defect types and how they are found

The defects that reach your customer are almost always internal. Gas porosity from inadequate venting, shrinkage cavities at isolated heavy sections, and cold shuts where two metal fronts met without fusing. All three are invisible to a visual inspection and all three show up under radiography or, at the surface, under dye penetrant.

For a pressure-retaining part, specify radiography on a defined percentage of castings or on defined critical sections, and specify the acceptance level rather than the word "radiographed". For non-pressure parts, dye penetrant on machined surfaces is usually enough. The trap here is that many Chinese foundries do the NDT through a third-party lab and the report arrives after the parts are packed; agree that the parts are held at the foundry until the report is accepted, otherwise you will be arguing about a container already on the water.

A chemistry check on the finished part is worth doing periodically. A handheld XRF on a machined surface takes seconds and will catch a heat that was poured from remelt scrap, which is a real practice at the cheaper end of the market. Full chemical analysis by spectrometer from a drilled sample gives you carbon and sulphur, which handheld XRF cannot measure properly.

Machining, passivation and the finishing gap

Most Chinese foundries either machine in-house or work with a machining shop next door. That is convenient and it hides a real risk: the part was designed for cast-then-machined, and the machining fixture was made by someone who never saw the drawing's datum callouts. Dimensional reports on first articles are standard practice; insist on them for every dimension marked critical, with the actual measured value, not a tick.

After machining, stainless castings should be passivated — typically a nitric or citric acid treatment — to restore the passive film on freshly cut surfaces and remove free iron contamination from tooling. Skipping it leaves a casting that rusts on the machined face while the cast face stays clean, which looks like a material problem and is not. Ask for the passivation method and a test method, and cheapest of all, ask for a ferroxyl test result on a sample.

What the document package should contain

For each shipment: heat number traceability from the casting to the finished part, the mill or foundry chemical certificate, the heat treatment record, NDT reports with the acceptance level, dimensional report on the first article, and any third-party inspection report. If a foundry can produce this without being pushed, they are worth the extra 5 percent. If they cannot produce it at all after you push, the price they gave you was for a different product than the one you think you ordered.