AISI 321 vs 321H: The H is a carbon floor, not a carbon ceiling

The 30-second answer

Both grades are titanium-stabilised 18Cr-10Ni austenitics, and both resist intergranular attack after welding. Where they diverge is in what the alloy is being asked to do with its carbon. Understanding that single point makes every other difference between them predictable — which is the argument the rest of this article makes.

If your application is below 538 °C and sensitisation resistance is the design driver, the grade you actually need is standard AISI 321 (UNS S32100) forging parts — product forms, size envelope, ASTM A182 / EN 10222-5 certification scope and enquiry form are all on that page.

Why H grades specify a carbon minimum

Every other requirement an engineer meets on a stainless steel certificate treats carbon as a contaminant to be suppressed. L grades cap it at 0.03 %. Standard 321 caps it at 0.08 %. The H grades do something structurally different: they put a floor under it at 0.04 %.

The reason is that above roughly 538 °C, failure stops being a strength problem and becomes a rate problem. The component does not yield; it slowly deforms under a load it would carry indefinitely at room temperature. That deformation happens by dislocations climbing through the lattice, and by grains sliding past one another at their boundaries. Both mechanisms are slowed by hard, thermally stable particles sitting in the way.

Carbon is what supplies those particles. In service, dissolved carbon precipitates as fine intragranular carbides that pin dislocations, and as boundary carbides that resist grain-boundary sliding. Strip the carbon out and you remove the obstacle field. This is why a 321 heat that happens to run at 0.02 % C — perfectly acceptable, arguably superior for a welded flange at 300 °C — is a genuinely inferior material in a 620 °C header.

The titanium ratio inversion: 5×(C+N) becomes 4×C

Here is the part that catches experienced engineers, and the reason plenty of “321H” material in circulation is quietly non-compliant.

ASTM A182 requires Grade F321 to carry titanium of not less than 5 × (C + N), to a maximum of 0.70 %. For Grade F321H the requirement drops to not less than 4 × C, same 0.70 % ceiling. So the grade with more carbon is required to carry proportionally less titanium.

That is not a drafting inconsistency. It is deliberate. Full stabilisation at 5×(C+N) is designed to convert essentially all available carbon into titanium carbide, so that none remains to form chromium carbide at grain boundaries. That is exactly the right objective for 321. It is the wrong objective for 321H, because carbon locked into coarse primary TiC formed during solidification and hot work contributes very little to creep resistance. What contributes is carbon that stays in solution through the solution anneal and re-precipitates as a fine, well-distributed population during the first thousands of hours of service.

Over-stabilise a 321H heat — say, run titanium at 0.62 % against 0.05 % carbon — and the chemistry passes every line of the certificate while the material has been quietly stripped of the thing you bought it for.

Clause numbering, note references and exact ratios differ between standards and between editions of the same standard — several product specifications state 4×(C+N) rather than 4×C for the H grade. Always verify against the edition named in your purchase order.

The ASTM No. 7 grain size rule, and why it is looser than 304H

ASTM A182 requires Grades F321H, F347H and F348H to have a grain size of ASTM No. 7 or coarser. This is mandatory, and it is the requirement most often missing from certificates offered as 321H.

Since higher ASTM numbers mean finer grains, “No. 7 or coarser” means the measured grain size number must be 7 or lower. A forging that comes out at No. 8 or No. 9 has failed, regardless of how good the chemistry looks. Fine grains mean more grain-boundary area per unit volume, and grain-boundary sliding is a primary creep mechanism — so the code will not extend elevated-temperature allowable stresses to material it cannot confirm is coarse enough.

Now the detail almost nobody explains. The same table requires F304H, F309H, F310H and F316H to be No. 6 or coarser — a stricter coarseness requirement than the stabilised H grades get. On first reading that looks backwards.

It is not. Titanium and niobium carbides are thermally stable particles that sit on austenite grain boundaries and physically pin them during the solution anneal. That pinning is precisely what makes these grades resistant to sensitisation — and it also makes their grains genuinely difficult to grow. Demanding No. 6 from a titanium-stabilised heat would force annealing temperatures high enough to start dissolving the stabilising carbides, which defeats the purpose of buying a stabilised grade at all. The code accommodates the metallurgy instead of fighting it.

What actually changes at 538 °C

At room temperature, put 321 and 321H side by side and you will struggle to tell them apart. Minimum tensile is around 515 MPa and minimum 0.2 % proof around 205 MPa for both; our solution-annealed forgings typically return 550–610 MPa and 240–290 MPa in either grade. Elongation, hardness and impact behaviour overlap almost completely. If your component never sees sustained heat, the H grade buys you nothing measurable.

Cross 538 °C and the question changes from “what load will this carry” to “what load will this carry for 100,000 hours.” Allowable stress stops being derived from yield and starts being derived from creep-rupture and creep-rate criteria. Published rupture data consistently place 321H above 321 in this regime, with the advantage widening as temperature rises — a modest margin near 550 °C, becoming the difference between a viable and a non-viable design by 650–700 °C. Long-service studies on 321H reformer-furnace tubing also show measurable creep-strength degradation after very extended exposure, which is why remaining-life assessment exists as a discipline for this grade.

The second-order effect matters too. Creep-rupture ductility falls as temperature and exposure time increase, so a component that has performed for twenty years is not the same material that was installed. This is the practical reason refinery and boiler operators track hours-at-temperature per component rather than treating H-grade material as a permanent solution.

The heat treatment conflict nobody warns you about

This is the section worth the read, and it is where 321H orders most often go wrong at the inspection stage.

Piping code practice restricts the use of 321H above 538 °C unless the material has received a solution anneal at a substantially higher temperature than the ordinary grade requires — in ASME B31.3 this appears as a note attached to the material in the allowable stress tables, calling for a minimum around 1095 °C. The logic is sound: you need enough thermal driving force to dissolve carbides into solution and let the grains grow to the required coarseness.

At the same time, the product specifications themselves carry an explicit caution that solution annealing above roughly 1065 °C may impair resistance to intergranular corrosion in TP321, TP321H, TP347 and TP348 after subsequent exposure to sensitising conditions. Push the anneal high enough and you begin dissolving the titanium carbides that make the grade stabilised in the first place; on cooling and in early service, that liberated carbon is available to form chromium carbide.

There is no universal answer, but there is a correct process. Establish which failure mode governs your component. A superheater header running continuously at 650 °C, never opened to atmosphere, is creep-governed — anneal high and accept the stabilisation trade-off. A hydroprocessing component that is creep-loaded in operation and exposed to polythionic acid during every turnaround shutdown is exposed to both, and needs the decision documented in writing before manufacture, usually with a post-fabrication stabilising treatment in the 870–900 °C range to re-commit free carbon to titanium before the first operating cycle.

What you must not do is leave it to the mill. Silence on the purchase order means the shop will anneal to whatever the base product standard requires, and the certificate can be fully compliant with that standard while the material still fails the note in your piping code. We have re-solution-annealed customer-supplied H-grade material for exactly this reason more than once.

Heavy-section reality: where grain size is actually measured

Most published guidance on 321H was written with tube and pipe in mind — thin walls, fast quench, uniform thermal history. Forgings behave differently, and three issues appear once section thickness passes about 150 mm.

Grain size varies through the section

The core of a heavy forging spends longer at temperature and cools more slowly than the surface. Grains there are coarser. A specimen taken at the outside diameter can report No. 7 while mid-radius sits at No. 5. Both may be acceptable, but only if the specification says where the measurement is taken. Our practice on H-grade forgings above 200 mm is to report grain size at surface, mid-radius and core rather than a single figure, and we recommend customers write that requirement in rather than inherit whatever the shop chooses.

Quench severity is a real limit, not a formality

Cooling from 1095 °C fast enough to keep carbides in solution is straightforward in a 6 mm tube wall. In a 350 mm forged block, the core cools through the 816–425 °C band slowly no matter how the quench is configured. Water quench with adequate tank volume, agitation and a controlled load-to-volume ratio is mandatory, and beyond a certain section thickness the honest answer is that the core cannot be treated as fully solution-annealed material. That is a conversation to have at the enquiry stage, not after the ultrasonic report.

Reduction ratio still governs soundness

None of the H-grade requirements substitute for forging work. We hold a minimum 3:1 reduction ratio on all critical AISI 321 and 321H forgings and target 4:1 or better for pressure-boundary components, with the finish-forging temperature held above 900 °C so no deformation occurs inside the sensitisation band before final heat treatment. The full process sequence from melting through CNC machining, and the available product forms — open die forgings, seamless rolled rings, forged bars, and hollow forgings — are documented on our AISI 321 open die forgings and rolled rings product page.

Dual certification 321 / 321H: when it is real

Because 321 caps carbon at 0.08 % and 321H spans 0.04–0.10 %, a heat landing between 0.04 % and 0.08 % can satisfy both. Mills like this, distributors like it more, and it is legitimate — but only when four conditions hold simultaneously:

1. Carbon falls inside the 0.04–0.08 % overlap on both heat and product analysis.
2. Titanium satisfies the 5×(C+N) rule for 321 and the H-grade ratio, while staying under 0.70 %.
3. Heat treatment is the higher H-grade solution anneal, not the standard 321 cycle. You cannot upgrade a certificate retroactively.
4. Grain size is measured and reported as No. 7 or coarser.

Condition four is where dual certification usually collapses. A large fraction of material offered as “321/321H dual certified” is dual certified on chemistry alone — a 321 heat that happened to land above 0.04 % carbon, annealed to the 321 cycle, with no grain size on the certificate. It is not 321H, and if the component is going into code service above 538 °C, accepting it transfers a real compliance problem onto your project.

Note also that condition two creates a narrow chemistry window. Meeting the 5×(C+N) titanium requirement at 0.07 % carbon while keeping titanium low enough not to over-stabilise for creep is a genuinely tight target, and it is a large part of why dual-certified heavy forgings command a premium over single-grade material.

Four situations where 321H is the wrong specification

• Sustained temperature below 538 °C. There is no allowable-stress benefit, the higher carbon marginally narrows the stabilisation margin, and you pay 8–18 % more. For service below that threshold, Grade 321 / EN 1.4541 forgings deliver everything you need at a lower cost.
• Chloride pitting is the governing corrosion mode. Neither grade contains molybdenum; PREN sits near 18 for both. Move to a molybdenum-bearing grade such as 316Ti or to duplex, and note that the H concept does not exist to solve corrosion.
• Continuous service above about 870 °C. Titanium stabilisation loses thermodynamic effectiveness and oxidation becomes limiting. Consider AISI 347 (UNS S34700) for niobium stabilisation, or AISI 310 / 310S where oxidation resistance dominates.
• Severe polythionic acid SCC with turnaround exposure. The higher carbon of the H grade is a liability here, and the elevated anneal makes it worse. AISI 347H is generally the better answer.

European projects add one more consideration. EN 1.4878 (X8CrNiTi18-10) is the closest European grade to 321H and is what usually gets supplied against H-grade enquiries in the EU, while EN 1.4541 (X6CrNiTi18-10) maps to standard 321. They are close equivalents, not clause-for-clause equals — the grain size and annealing rules written into the ASTM and ASME H-grade framework do not transfer across identically. On dual-scope projects, name both designations and state the grain size requirement explicitly.

Purchase order wording that produces the right material

Nine out of ten 321H problems we see are specification problems, not manufacturing problems. This is the minimum text that closes the gaps discussed above. Adapt the bracketed values, and delete nothing.

MATERIAL      ASTM A182 Grade F321H (UNS S32109)
              [and EN 10222-5 / EN 1.4878 where dual scope applies]

CHEMISTRY     Carbon 0.04-0.10 % on heat AND product analysis.
              Titanium per the governing edition of A182, max 0.70 %.
              Report C, N and Ti individually on the MTC.

HEAT TREAT    Solution anneal at [1095] C minimum, water quench.
              Report actual furnace temperature, soak time and
              quench medium. Justification: [ASME B31.3 note for
              service above 538 C / project spec ref].

GRAIN SIZE    ASTM No. 7 or coarser per ASTM E112. MANDATORY.
              For sections over 200 mm, report at surface,
              mid-radius and core.

FORGING       Minimum reduction ratio [4]:1 from ingot.
              Finish forging temperature above 900 C.

NDT           100 % UT per [ASTM A388 / EN 10228-3 Class 3].
              PT or MT on all machined surfaces.

IGC TEST      ASTM A262 Practice [E], sample cut from the
              production forging, not a companion bar.

CERTIFICATE   EN 10204 [3.1 / 3.2]. If 3.2, name the inspection
              body in this purchase order.

NOT ACCEPTED  Dual 321/321H certification without a reported
              grain size and the H-grade anneal record.

The last line is the one that saves projects. It costs nothing to include and it removes the single most common non-conformance in this grade.