Why the Microstructure of T92 Steel Is Treated as a Design Variable
Grade T92 is a 9% Cr tungsten-bearing martensitic steel used for superheater and reheater tubing in supercritical and ultra-supercritical boilers. Its room-temperature tensile figures are not remarkable on their own; what separates T92 from earlier 9Cr grades is how its microstructure behaves after thousands of hours at metal temperatures between 580 °C and 620 °C. For a purchaser, the practical consequence is that two tubes carrying the same grade stamp can have materially different creep lives if their prior austenite grain size, tempering response and precipitate distribution differ. That is why ASTM A213/A213M and the corresponding ASME SA-213 specification, together with EN 10216-2 grade X10CrWMoVNb9-2, place as much weight on heat treatment records and product analysis as on the tensile test certificate.
This article examines the microstructure of T92 feature by feature - tempered martensite matrix, M23C6 grain-boundary carbides, MX carbonitrides, dislocation substructure and grain boundaries - and translates each feature into the property or failure mode that a boiler designer or inspection engineer actually cares about.
The Matrix: Tempered Martensite and Prior Austenite Grain Size
After normalizing from about 1040 °C to 1080 °C and cooling, the austenite in a 9Cr-2W steel transforms almost completely to martensite; the low carbon content (0.07% to 0.13%) keeps the martensite dislocated rather than heavily twinned. Subsequent tempering at roughly 730 °C to 780 °C does not recrystallize this structure. What remains is a hierarchy of prior austenite grains, each subdivided into packets and blocks of laths a few hundred nanometres wide.
Two consequences follow directly:
The prior austenite grain size sets the coarsest scale of the structure. Finer prior austenite grains shorten the free path for dislocation motion and, more importantly, distribute creep damage over a larger number of boundaries, which delays the onset of boundary cavitation.
The lath width sets the scale on which M23C6 particles sit. Laths that are too wide, or that have been allowed to recover during an excessive tempering hold, leave long precipitate-free segments where dislocations can glide and pile up without obstruction.
A normalizing temperature that is too low leaves undissolved or incompletely dissolved precipitates and produces an uneven, banded structure; a normalizing temperature that is too high coarsens the prior austenite grains and permanently costs creep life, because that grain size cannot be restored by tempering. The heat treatment record - not the mill test certificate - is the document that reveals which case applies.
M23C6 Carbides: Boundary Pinning and the Limit of Coarsening
Chromium-rich M23C6 carbides precipitate preferentially on prior austenite grain boundaries, packet boundaries and lath boundaries during tempering. In the correctly heat-treated condition they form a semi-continuous, sub-micrometre chain rather than coarse globules. Their function is twofold: they stabilize the lath and boundary network so that the structure resists recovery, and they act as obstacles to boundary sliding.
The M23C6 population is also the weakest link in long-term service. At 600 °C the particles coarsen by Ostwald ripening, the spacing between them increases, and boundary pinning weakens. Coarsening is accelerated by higher tempering temperature, longer tempering time and by any excursion above the design metal temperature. This is why the ASME and EN specifications fix a minimum tempering temperature rather than a range of convenience, and why a tube that has been re-tempered to correct hardness cannot be assumed to have the creep strength of a correctly treated tube.
MX Carbonitrides: the Real Source of T92's Creep Advantage
The addition of niobium (0.04% to 0.09%), vanadium (0.15% to 0.25%) and nitrogen (0.030% to 0.070%) produces a second, far finer precipitate family: MX carbonitrides of the type Nb(C,N) and V(C,N), typically a few to a few tens of nanometres in size. These particles are thermodynamically much more stable than M23C6 and coarsen very slowly at 600 °C to 650 °C. Their role is to pin dislocations inside the laths and to retard the recovery of the dislocation substructure.
Compared with Grade T91, the T92 composition adds 1.50% to 2.00% tungsten and reduces molybdenum to 0.30% to 0.60%. Tungsten partitions between the matrix and the precipitates and, in combination with the MX population, delays the loss of substructure strength at the top of the temperature range. This is the metallurgical reason T92 appears in boiler designs where T91 would have to be replaced by austenitic tubing above roughly 580 °C to 590 °C metal temperature.
| Microstructural feature | Typical scale | Property contribution | Process control that governs it |
|---|---|---|---|
| Tempered martensite laths | 0.2-0.5 micrometre width | Yield and tensile strength, low-cycle fatigue resistance | Normalizing temperature and cooling rate; tempering temperature |
| Prior austenite grain size | ASTM grain size number typically 7 or finer | Creep ductility, cavitation resistance, toughness | Normalizing temperature and hold time; austenitizing history of the billet |
| M23C6 on boundaries | 0.1-0.5 micrometre | Boundary pinning, resistance to boundary sliding, hardness | Tempering temperature and time; carbon and chromium balance |
| MX carbonitrides | 5-50 nanometres | Creep strength, substructure stability, temper resistance | Nb, V and N content; solution treatment; avoided overheating |
| Dislocation density | Order of 10^14 per square metre after tempering | Initial strength, work-hardening capacity | Martensite finish temperature, tempering response |
Dislocation Substructure: Strength Now, Softening Later
Tempered martensite in T92 retains a high dislocation density - of the order of 10^14 m^-2 - arranged in tangles and cell walls inside the laths. This substructure is the immediate reason the grade reaches a specified minimum tensile strength of 620 MPa with a minimum yield strength of 440 MPa in the ASTM A213 condition, while elongation remains at 20% or better. The same substructure is metastable. At service temperature it gradually annihilates, and the rate of annihilation is controlled by how effectively the MX particles pin the dislocations.
For the plant owner, this means hardness alone is an unreliable acceptance criterion. A hardness reading inside the specified range indicates that tempering occurred, but it does not prove that the MX population is correct. Product analysis for Nb, V and N, plus a review of the solution-treatment temperature, is what actually closes that gap. Where a creep test is required by the purchase order, it is the only direct evidence of the balance between substructure strength and precipitate stability.
Heat Treatment Window and Its Microstructural Consequences
The heat treatment of T92 is narrow and unforgiving. Normalizing in the range of approximately 1040 °C to 1080 °C dissolves the M23C6 and most of the MX particles so that they can reprecipitate in a fine, uniform dispersion on tempering. Tempering at approximately 730 °C to 780 °C then relieves the martensite stresses, allows M23C6 to form on the boundaries and reprecipitates the MX family at nanoscale. Under-tempering leaves a hard, brittle structure with poor toughness and a high risk of cracking in service; over-tempering coarsens M23C6 and initiates recovery of the substructure, which permanently reduces creep strength.
Because the window is narrow, the following documents should be requested for any T92 heat of tubing destined for a pressure part:
Continuous furnace chart or batch furnace record showing charge geometry, set temperature and hold time for both normalize and temper.
Hardness survey across a tube cross-section, not a single reading on one wall.
Product analysis for Nb, V, N, B and W, cross-checked against the ASTM A213 limits for the grade.
For the weld procedure qualification, a post-weld heat treatment record with the actual thermocouple trace.
What Happens in the Heat-Affected Zone
The heat-affected zone of a T92 weld contains a narrow coarse-grained region that reaches temperatures above Ac3 and transforms to fresh, untempered martensite on cooling. If that region is left untempered it is hard and crack-sensitive; if it is tempered at the wrong temperature the M23C6 distribution there differs from the parent tube. Post-weld heat treatment in the range of approximately 750 °C to 770 °C, with controlled heating and cooling rates and verified thermocouples, is therefore mandatory rather than optional. Filler metal selection follows the matching 9Cr-2W route typical of the ER90S-B9 classification in AWS A5.28, and the procedure should be qualified with hardness traverses across the joint, not only with tensile tests.
Field repairs deserve particular attention: a localized temper with a heating blanket often fails to hold the whole wall thickness for the required time, producing a soft or hard band that later becomes the initiation site. Ultrasonic examination to ASTM E213 or the ISO 10893 series, together with hardness mapping, is the practical way to confirm that the repaired zone matches the parent metal.
Degradation Mechanisms Visible Only on the Microstructural Scale
Three damage processes govern the long-term performance of T92 tubing, and all three are microstructural:
Precipitate coarsening. M23C6 particles grow, spacing increases and boundary pinning is lost. The visible symptom is accelerating creep rate long before any dimensional change is measurable.
Formation of Laves phase. Iron-tungsten-molybdenum intermetallics precipitate during long exposure at 600 °C to 650 °C, removing tungsten from solid solution and reducing the solid-solution strengthening contribution. Their appearance is normal; the problem is the associated depletion of the matrix.
Z-phase formation. In the most highly alloyed 9% Cr steels, chromium niobium nitride can replace the fine MX particles after very long exposure, removing the most effective strengtheners and causing an acceleration of creep rate that laboratory short-term tests do not reveal.
Steam-side oxidation is a parallel concern. The chromium oxide scale that forms on 9% Cr steels at 600 °C and above is thin and protective in the early stage, but it thickens over time and can exfoliate, sending hard oxide flakes downstream into the superheater outlet headers and turbine. Where metal temperatures exceed roughly 600 °C, designers frequently specify a fine-grained variant or an internal surface treatment, and the purchase specification should state which surface condition is required rather than leaving it to the mill.
Inspection and Acceptance Rules That Verify the Microstructure
Microstructure is not directly specified in ASTM A213, so acceptance has to be built from indirect but verifiable evidence. A practical acceptance sequence for T92 tubing is:
Verify grade identity and heat number traceability on the mill test certificate, and confirm that the product analysis sits inside the A213 limits for C, Cr, W, Mo, V, Nb, N and B.
Confirm tensile properties meet the specified minima of 620 MPa tensile and 440 MPa yield with 20% elongation, and that hardness does not exceed the grade limit of about 250 HBW.
Confirm the heat treatment condition recorded on the certificate is normalizing plus tempering within the specified ranges, and that no re-tempering for hardness correction has occurred.
Perform dimensional and wall-thickness checks against the ordered sizes, since eccentricity reduces the effective creep life of the thinnest wall sector.
Carry out non-destructive testing - ultrasonic or eddy current as permitted by the product specification - and, for critical headers, a metallographic replica on a sample tube to confirm grain size and spheroidization.
Common Misunderstandings About T92 Microstructure
Claiming that a higher tensile result proves better creep performance. Room-temperature tensile strength mostly reflects the as-tempered dislocation density; creep strength at 600 °C reflects the MX dispersion and its stability. A heat that meets 620 MPa can still have an inferior creep life if its nitrogen or niobium content sits at the bottom of the range.
Treating T92 as a drop-in replacement for T91 without re-checking the design. The two grades share a chromium level but differ in tungsten, in the tempering response and in the allowable stress at temperature. Weld procedures, post-weld heat treatment parameters and dissimilar joints all need to be re-qualified.
Assuming that machining or bending is harmless. Cold work introduces dislocations that interact with the existing substructure; cold bending beyond modest strain requires a subsequent stress-relief heat treatment, and the temperature of that treatment must stay below the range that would over-temper the parent metal.
Reading the source literature as if fine grains alone guarantee performance. Grain refinement helps cavitation resistance, but it also increases the boundary area available for M23C6 precipitation; the balance between grain size, precipitate distribution and tempering condition is what the specification is trying to control.
FAQ
Q: What is the primary microstructure of T92 steel after heat treatment?
Tempered martensite with a high dislocation density, decorated by chromium-rich M23C6 carbides on prior austenite grain boundaries, packet boundaries and lath boundaries, and by nanoscale MX carbonitrides of niobium and vanadium inside the laths.
Q: Which precipitates control the creep strength of T92 at 600 °C?
The MX carbonitrides, because they are far more stable than M23C6 and continue to pin dislocations after the carbides have coarsened. Tungsten in solid solution and in Laves phase adds a secondary contribution.
Q: Why must T92 be normalized and tempered rather than only stress relieved?
Normalizing at approximately 1040 °C to 1080 °C is required to dissolve the carbides and produce a uniform martensitic structure, and tempering at approximately 730 °C to 780 °C is required to develop the M23C6 and MX dispersions. Stress relief alone leaves an untempered, brittle martensite.
Q: Does a finer prior austenite grain size always improve T92 performance?
Finer grains improve creep ductility and resistance to boundary cavitation, but the effect must be achieved through controlled normalizing rather than through excessive cold work followed by low-temperature heat treatment. Grain size is judged together with precipitate distribution, not in isolation.
Q: What is the specified minimum tensile strength of T92 tubing to ASTM A213?
620 MPa tensile strength, 440 MPa yield strength and 20% elongation in the specified test direction, with hardness typically limited to about 250 HBW.
Q: How is a T92 weld restored to the parent-metal microstructure?
By post-weld heat treatment at approximately 750 °C to 770 °C with verified thermocouples across the joint, controlled heating and cooling rates, and confirmation through hardness traverses and ultrasonic examination that the heat-affected zone has been uniformly tempered.





