Creep-Resistant Steels: P91, P92, and Grade 91 — Microstructure, PWHT, and Type IV Cracking

P91 (Grade 91) and P92 are the defining materials of modern ultra-supercritical (USC) and advanced ultra-supercritical (A-USC) power generation. These modified 9Cr–1Mo ferritic–martensitic steels replaced the older 2.25Cr–1Mo (P22) and 5Cr–0.5Mo (P5) grades from the 1980s onwards, enabling steam temperatures of 565–625 °C at pressures of 25–35 MPa with wall thicknesses 30–50% thinner than equivalent P22 construction. Their creep strength is not a simple consequence of alloying — it emerges from a precise hierarchical microstructure of tempered martensite laths, sub-grain boundaries, and nanoscale precipitate populations that took two decades of alloy development to optimise. This article covers the physical metallurgy in full, from the phase transformations during heat treatment through the mechanisms of creep deformation, long-term microstructure degradation, Type IV weld cracking, and remaining-life assessment practice in operating power plants.

Alloy Design: From 2.25Cr–1Mo to P91 and P92

The development of P91 and P92 was driven by the thermodynamic and engineering imperatives of supercritical steam power generation. Higher steam temperatures and pressures improve Rankine cycle thermal efficiency (Carnot theorem: higher T_hot increases efficiency). Every 10 °C increase in main steam temperature above 540 °C corresponds to approximately 0.4–0.5 percentage points in cycle efficiency — representing millions of tonnes of coal or gas saved annually across a fleet of power stations. The barrier was material creep resistance.

Development History and Grade Families

The evolution of ferritic creep-resistant steels for power plant follows a clear alloy engineering logic. The original P11 (1.25Cr–0.5Mo) and P22 (2.25Cr–1Mo) grades, developed in the 1950s–1960s, were limited to approximately 540 °C by their rupture strength. Increasing Cr content to 9–12% dramatically improved oxidation resistance and allowed both solid solution and precipitation strengthening mechanisms to be more aggressively exploited. The addition of vanadium and niobium to the 9Cr–1Mo base (producing modified 9Cr–1Mo, ASTM A335 P91, registered in the ASME Code in 1983) was the critical step that tripled the 100,000-hour rupture strength at 600 °C compared with P22.

Role of Each Alloying Element in P91

P91 (ASTM SA-335 P91 / ASME SA-182 F91) Specified Composition:

  %C:   0.08–0.12   — Controls M₃C₂ and MX carbide volume fraction;
                       too low → insufficient carbide; too high → weldability issues
  %Mn:  0.30–0.60   — Mild solid solution; assists hardenability
  %Si:  0.20–0.50   — Deoxidiser; max limited to avoid delta-ferrite
  %Cr:  8.00–9.50   — Oxidation/corrosion resistance; M₃C₂ stability; hardenability
  %Mo:  0.85–1.05   — Solid solution creep strengthening; M₃C₂ stabilisation
  %V:   0.18–0.25   — MX (VN, VC) precipitation; sub-grain pinning
  %Nb:  0.06–0.10   — MX (NbC, NbN) precipitation; grain refinement; thermal stability
  %N:   0.030–0.070 — VN, NbN precipitation; solid solution at low T; critical for MX
  %Ni:  ≤0.40        — Toughness; hardenability; must be low to avoid delta-ferrite
  %Al:  ≤0.02        — Strictly limited: AlN ties up N needed for VN formation
  %P:   ≤0.020       — Temper embrittlement risk; keep low
  %S:   ≤0.010       — MnS inclusion control

CRITICAL INTERACTIONS:
  N must be >8×V  for adequate VN precipitation (mass stoichiometry)
  Al must be <0.02% to prevent AlN consuming the free N before VN forms
  Ni + Mn must be controlled to prevent delta-ferrite (>1% Δ-ferrite
  degrades creep strength by >20%)
  Nb/V ratio 0.3–0.5 provides optimum MX thermal stability

Microstructure of P91 in the Normalised and Tempered Condition

P91 is supplied in the normalised (austenitised and air-cooled) and tempered condition per ASTM A335 and ASME Section II material specifications. Understanding the microstructure produced by this treatment — and why every feature in it matters for creep — is the foundation for all subsequent design, welding, PWHT, and life assessment decisions.

Normalising: Austenitisation and Martensite Formation

Normalising is performed at 1,040–1,080 °C, where the steel is fully austenitic (above Ac₃ ≈ 900 °C for P91). During the austenitising soak, all carbides and carbonitrides dissolve into the austenite matrix (except for a small fraction of Nb-rich MX which pin austenite grain boundaries and limit grain growth to ASTM size 5–8, typical grain diameter 20–60 μm). Air cooling from 1,060 °C at approximately 5–15 °C/s fully suppresses the bainite and pearlite transformations (M_s ≈ 420 °C for P91), producing a fully martensitic microstructure. As-normalised hardness is typically 350–450 HBW — far too hard and brittle for any application. PWHT is mandatory before any testing or service use.

Tempering (PWHT): Recovery and Precipitate Formation

The standard PWHT at 730–775 °C for P91 simultaneously accomplishes several microstructural changes that collectively produce the tempered martensitic microstructure responsible for creep resistance:

1. Martensite lath recovery: High dislocation density within the as-quenched laths (approximately 10¹⁴–10¹⁵ m⁻²) rearranges into lower-energy configurations. Dislocation tangles become organised sub-grain boundaries (low-angle boundaries of 0.5–2 ° misorientation). Lath boundaries partially dissolve where adjacent laths share the same crystallographic packet orientation.
2. M₃C transition carbides dissolve: Metastable cementite-type carbides (M₃C) that precipitated during initial martensite tempering below 300 °C dissolve and reprecipitate as the thermodynamically stable M₂₃C₆ chromium carbide.
3. M₂₃C₆ formation: Cr-rich M₂₃C₆ (approximate composition Cr₂₁Fe₂C₆) precipitates on prior austenite grain boundaries (PAGB), martensite lath boundaries, and sub-grain boundaries. Typical size at completion of PWHT: 50–200 nm. These particles pin boundary migration during creep deformation — the primary mechanism preventing sub-grain coarsening (lath dissolution) that would reduce the boundary area available for dislocation annihilation.
4. MX carbonitride precipitation: Fine VN, VC, NbC, and mixed (V,Nb)(C,N) particles nucleate in the lath interiors on dislocations and in the matrix. Size range after PWHT: 5–30 nm. These particles pin individual dislocations by the Orowan bypass mechanism and, critically, pin sub-grain boundary migration from within the interior of each sub-grain.

Key Precipitate Phases in P91/P92

Creep Mechanisms and the Three-Stage Creep Curve

Creep is time-dependent plastic deformation under constant stress at temperatures above approximately 0.3 T_m (absolute melting temperature). For P91 with T_m ≈ 1,500 °C (1,773 K), significant creep begins above 0.3 × 1,773 ≈ 530 K (257 °C), but engineering relevance starts above approximately 450 °C where diffusion-controlled mechanisms become rate-limiting.

Primary, Secondary, and Tertiary Creep

Creep Strain Rate — Norton Power Law (secondary creep):

  ἡ_s = A × σ^n × exp(−Q_c / RT)

  Where:
    ἡ_s = minimum (secondary) creep strain rate  [s⁻¹]
    A   = material constant
    σ   = applied stress  [MPa]
    n   = stress exponent (P91: n ≈ 5–12 at 600°C)
    Q_c = activation energy for creep  [J/mol]
    R   = gas constant = 8.314 J/mol·K
    T   = absolute temperature  [K]

  P91 at 600°C (873 K) typical constants:
    n ≈ 8–10 (dislocation climb-controlled regime, stress < 100 MPa)
    Q_c ≈ 300–400 kJ/mol (above ~300 kJ/mol implies sub-grain
    boundary controlled mechanism, not simple dislocation climb)

Monkman-Grant relationship (links minimum creep rate to rupture life):
  ἡ_min × t_rᴕ = C  (Monkman-Grant constant)

  Where t_r = rupture time; m ≈ 0.7–1.0; C = ~0.1 for P91
  Enables rupture life prediction from short-term creep rate measurements.

Physical Mechanism: Sub-grain Boundary Creep in P91

At engineering stress levels relevant to power plant design (40–100 MPa at 600 °C), creep in P91 is primarily controlled by dislocation climb along and across sub-grain boundaries, not by simple power-law dislocation glide. This mechanism, often called the “sub-structure hardening” model, explains why:

• The stress exponent n is anomalously high (8–12) compared with the theoretical value of ~5 for dislocation climb — the sub-grain boundary network provides additional back stress that makes the material appear stronger than the applied stress alone would predict.
• Creep strength degrades gradually as sub-grain boundaries coarsen (M₂₃C₆ particles coarsen and become less effective at pinning boundary migration, allowing sub-grains to grow and reducing the boundary area available for dislocation annihilation).
• The transition from primary to secondary creep reflects the establishment of a steady-state sub-grain size where boundary migration rate equals the rate of new dislocation generation from applied stress.

This understanding has a critical practical implication: any process that coarsens M₂₃C₆ (excessive PWHT temperature, over-tempering, service at temperatures above design) or dissolves MX particles (service at temperatures approaching 900 °C) will irreversibly degrade creep strength. See the martensite formation guide for the crystallographic basis of the lath microstructure.

Type IV Cracking: The Critical Weld Life Limitation in P91

Type IV cracking is the premature creep failure of P91 weld joints occurring in the fine-grained intercritical heat-affected zone (ICHAZ) — not in the weld fusion zone and not in the base metal — and it is the dominant in-service failure mode that limits the life of P91 headers, crossover pipes, and main steam lines in supercritical power plants worldwide. Understanding its mechanism is prerequisite to writing welding procedures, designing inspection intervals, and conducting fitness-for-service assessments.

The HAZ Microstructure Gradient

During welding, the base metal adjacent to the weld fusion line experiences a continuous temperature gradient from near-melting at the fusion boundary to ambient temperature at the outer edge of the thermal cycle. This gradient produces a corresponding microstructure gradient through the HAZ:

Type IV Crack Initiation and Propagation Mechanism

Under high-temperature creep loading, the stress concentration at the weld toe (geometric discontinuity) and the constrained deformation of the weak ICHAZ surrounded by stronger base metal and weld metal combine to focus creep strain in the intercritical zone. The sequence of damage accumulation is:

1. Void nucleation at coarsened M₂₃C₆–ferrite matrix interfaces in the ICHAZ. M₂₃C₆ particles have coherency stresses and debond from the matrix at modest creep strains (ε_c < 0.5%).
2. Void growth by vacancy condensation driven by the tensile stress component normal to the prior austenite grain boundaries in the ICHAZ. Growth rate depends on local creep strain rate and vacancy diffusivity (hence strongly temperature-dependent).
3. Void linkage along the PAGB network of the ICHAZ, producing a connected crack path approximately 2–5 mm from the weld fusion line, following the position of maximum ICHAZ softening.
4. Macroscopic crack formation appearing as a circumferential crack at the weld toe, often discovered during inspection as surface-opening or detected by phased array UT as a subsurface reflector at a characteristic standoff distance from the fusion line.

Post-Weld Heat Treatment Requirements for P91

PWHT of P91 weld joints is the most technically demanding and code-critical heat treatment operation in modern power plant fabrication. The tight temperature window, mandatory preheat and interpass controls, and post-PWHT verification requirements make it qualitatively different from PWHT of conventional carbon or low-alloy steels.

ASME B31.1 and B31.3 Requirements

ASME B31.1 (Power Piping) PWHT for P91 (P-No. 15E, Group 1):

  PWHT temperature:  730–775°C
  Hold time:         60 min minimum + 1 min/mm above 25 mm thickness
  Heating rate:      max 55°C/hr above 300°C (for T > 300°C)
  Cooling rate:      max 55°C/hr above 300°C (cool slowly through
                     M₃⁷ range at 420°C to minimise residual stress)
  Below 300°C:      cool freely in still air

  Preheat (before welding):  200°C minimum, 400°C maximum
  Interpass temperature:     300°C maximum (do not allow to cool below
                             200°C during welding)
  Post-weld bake (optional): 200–250°C × 2–4 hr before PWHT for
                             heavy walls >50 mm

  Alternative: ASME Code Case 2781 allows PWHT at 760°C minimum
  for specific applications — reduces minimum to single point.

Post-PWHT hardness verification:
  Weld metal:        >200 HBW and ≤265 HBW (ASME B31.1 para. 132.7)
  HAZ:               Same range (spot-check at ICHAZ position)
  Base metal:        No change from mill cert values expected

Temperature accuracy requirement:  ±14°C across the heated band
  (requires Type K or N calibrated thermocouples, not base-metal
  contact pyrometers; optical pyrometers unacceptable for PWHT)

Consequences of PWHT Non-Compliance

Historically, P91 PWHT errors have caused more failures and rejections than any other single process variable. The consequences of specific deviations:

• Temperature below 730 °C: M₃C transition carbides not fully converted to M₂₃C₆; lath sub-structure recovery incomplete; high residual dislocation density and hardness (>265 HBW); risk of hydrogen-assisted cracking and inadequate creep strength. The weld must be re-heat-treated.
• Temperature above 775 °C (approaching Ac₁ ≈ 815 °C): Partial re-austenitisation at grain boundaries produces fresh untempered martensite on cooling, dramatically increasing hardness in a heterogeneous pattern. Re-heat-treating at correct temperature partially mitigates this but cannot fully reverse grain boundary chemistry changes. If Ac₁ is substantially exceeded, the affected area must be removed and re-welded.
• Insufficient hold time: Incomplete M₂₃C₆ precipitation; inadequate sub-grain formation; properties not uniformly through-thickness in heavy-wall pipe (>75 mm). Extension of hold time in a second PWHT cycle may correct this.
• Cooling too fast below 300 °C: Increases residual stress; risk of delayed hydrogen cracking if hydrogen is present. Slow furnace cooling through the M_s range (420 °C) is preferred for all P91 welds above 25 mm wall thickness.

Remaining Life Assessment Methods for P91 Power Plant Components

P91 headers, steam pipes, and tube assemblies in supercritical power plant have design lives of 200,000–300,000 hours at 600 °C. Given the age of the first-generation P91 plants (commissioned from the late 1980s to 2000s), remaining life assessment is now a routine activity for plant operators. The assessment combines metallographic evidence (current damage state) with time-fraction calculations (consumed creep life) to predict the time to the next inspection and eventual end-of-life decision.

Metallographic Replica Examination

Non-destructive metallographic replicas are taken from the OD surface of headers and pipe at weld toes, saddle branch junctions, and bend mid-span positions. The procedure (per ASTM E1351 / EPRI TR-1004934):

1. Surface preparation: Grind to 600-grit SiC; polish with 1 μm diamond; etch with Vilella’s reagent (for P91) or Marshall’s reagent to reveal prior austenite grain boundaries and carbide distribution.
2. Cellulose acetate replica: Dissolve acetate film in acetone, apply to etched surface; allow to dry; peel off — the replica captures the surface topography at ~1 nm resolution.
3. SEM examination: Replicas mounted on SEM stubs and examined at 500–5,000× magnification. Count creep voids per unit area at the ICHAZ position; classify against EPRI A–E damage scale.
4. Carbide coarsening assessment: M₂₃C₆ mean equivalent diameter from image analysis provides an independent indicator of thermal exposure (time at temperature) via coarsening kinetic models.

Hardness Surveys and PMI Requirements

In-service hardness surveys (Leeb rebound, Vickers portable, or Brinell impression) at weld joints provide rapid screening for over-tempering or mis-heat-treatment. Hardness below 200 HBW at any weld position requires investigation; hardness below 170 HBW at HAZ positions is a Fitness-for-Service red flag requiring formal assessment before continuing operation.

EPRI Creep Damage Classification

EPRI void density classification for creep-damaged P91 replica specimens. Class A = no action; B = increase inspection frequency; C = engineering assessment required and reinspect within 2 years; D = engineering assessment, removal from service or immediate repair; E = emergency removal from service.

Welding Metallurgy and Consumable Selection for P91

Welding P91 is fundamentally different from welding conventional low-alloy steels. The narrow composition window required for adequate creep strength, the mandatory preheat and PWHT requirements, and the Type IV susceptibility of any misapplied procedure make P91 welding one of the most technically demanding operations in power plant construction.

Welding Process Selection

SMAW (GTAW/TIG + SMAW fill) and GTAW (root) + FCAW (fill/cap) are the dominant processes for P91 pipe butt welds. GTAW is the preferred root process for 100% penetration control and absence of hydrogen from the flux; SMAW with low-hydrogen E9015-B91 or E9018-B91 electrodes is used for fill and cap. FCAW with metal-cored E91T1-B9 wire is gaining acceptance for large-bore header girth welds where productivity matters. SAW is used for longitudinal seam welds in large-diameter pipe manufacture.

The welding heat input for P91 must be controlled within a defined range: too low (<0.5 kJ/mm) produces a wide martensite ribbon adjacent to the fusion line that transforms but may not be fully tempered by subsequent PWHT; too high (>3.5 kJ/mm) produces excessive grain growth in the CGHAZ and may inadvertently widen the ICHAZ softening zone. A typical target range for P91 SMAW fill passes is 1.0–2.5 kJ/mm at a 200–280 °C interpass temperature. See the HAZ microstructure guide and the hydrogen-induced cracking article for the fundamentals underpinning these requirements.