Low-Temperature Steels for Cryogenic Service: 9Ni, Invar, and Austenitic Grades

The engineering of structures for cryogenic service — liquid natural gas (LNG) storage at −165°C, liquid nitrogen at −196°C, liquid hydrogen at −253°C, and liquid helium at −269°C — demands a fundamental understanding of low-temperature fracture mechanics and crystal structure effects on ductility that is entirely absent from room-temperature structural engineering. The ductile-to-brittle transition in body-centred cubic (BCC) steels kills toughness long before these temperatures are reached; the engineering solution is to select materials that either suppress the transition (9% nickel steel), eliminate it by crystal structure (austenitic grades), exploit it for dimensional stability (Invar), or bypass it by using different crystal structures entirely (aluminium, copper, FCC alloys). This article provides rigorous treatment of the metallurgical mechanisms and practical engineering criteria for all major cryogenic steel families.

1. The Fundamental Problem: BCC vs. FCC at Low Temperature

The central metallurgical challenge of cryogenic engineering is the ductile-to-brittle transition (DBT) in body-centred cubic (BCC) metals. Standard carbon steels and low-alloy ferritic steels — the workhorses of ambient-temperature structural engineering — transform from tough, ductile materials at room temperature to brittle, cleavage-fracturing materials as temperature falls, typically transitioning between −20°C and −60°C depending on composition and microstructure. This transition is catastrophic: Charpy energy can drop from 200 J to less than 5 J over a range of only 20–40°C, and fracture toughness K_IC falls correspondingly.

1.1 The Peierls-Nabarro Mechanism

The physical origin of the BCC DBT lies in the temperature dependence of dislocation mobility. In BCC metals, dislocation glide occurs on {110}⟨111⟩ and {112}⟨111⟩ slip systems. The Peierls-Nabarro stress — the minimum stress required to move a dislocation through the crystal lattice — is strongly temperature-dependent in BCC metals due to the relatively open (non-close-packed) nature of the BCC crystal and the deep Peierls valley. As temperature decreases, thermal activation becomes insufficient to assist dislocation kink-pair nucleation and migration across the Peierls barrier, and the required applied stress rises sharply.

Peierls-Nabarro stress (simplified) and DBT condition:

  τᵖ ≅ (2G / (1−ν)) × exp(−2πa/b)

  At low T: τᵖ rises steeply until  σᵅᵖᵖₛᵖᵈᵈ > σᴾᴸᵀᵅᵂᵅᵂᵀᵃᵂᵀ → cleavage fracture
  At high T: τᵖ is low; dislocations move freely → ductile fracture

FCC metals (austenite, Cu, Al):
  - Close-packed {111} slip planes; Peierls stress low and temperature-independent
  - Cleavage stress much higher than Peierls stress at ALL temperatures
  - ∴ No DBTT; ductility retained to absolute zero

Impact: A Charpy specimen of C-Mn steel absorbs ~200 J at +20°C
        and <5 J at −60°C — the same material, only temperature differs

1.2 Effect of Composition on DBTT

The DBTT of low-alloy steels is strongly composition-dependent. Carbon raises the DBTT most severely — each 0.01 wt% increase in C raises the DBTT by approximately 2.5°C. Manganese lowers the DBTT (approximately 5°C per 0.1 wt% Mn). Nickel is the most potent DBTT-lowering alloying element in steels — each 1 wt% Ni lowers the DBTT by approximately 10°C — the scientific basis for 9% Ni steel achieving −196°C toughness. Phosphorus, sulphur, nitrogen, and oxygen at grain boundaries raise DBTT by grain boundary embrittlement. The transition to fine, clean microstructures (fine-grained, low-sulphur, continuous cast, vacuum degassed) has progressively lowered the DBTT of structural steels since the mid-20th century.

2. 9% Nickel Steel — Metallurgy, Heat Treatment, and Properties

Nine percent nickel steel was developed by C.W. Simcoe at the International Nickel Company (INCO) in 1944, with first ASTM standardisation as A353 in 1955 and subsequent revision to the current A553 (Type I and Type II). It remains the dominant material for LNG flat-bottomed storage tanks worldwide, specified under EN 14620, BS 7777, JIS G3127, and equivalent national standards.

2.1 Composition and Standard Grades

9% Nickel Steel Composition (ASTM A553 Type I and Type II):

  C    ≤ 0.13 wt%   (low carbon to reduce DBTT and improve weldability)
  Mn   0.30–0.90%
  Si   ≤ 0.45%
  P    ≤ 0.015%     (controlled for grain boundary cleanliness)
  S    ≤ 0.005%     (controlled; sulphide inclusions nucleate cleavage)
  Ni   8.50–9.50%   (nominal 9%, lowers DBTT ~90°C below 0Ni baseline)
  Al   ≤ 0.060%     (grain refinement, deoxidation)

Type I:   Double normalised + temper heat treatment
Type II:  Quench and temper heat treatment (preferred for heavier sections)

Mechanical requirements (ASTM A553, plate):
  Yield strength (0.2% proof):    585–690 MPa
  Tensile strength:                690–825 MPa
  Elongation (50 mm gauge):        ≥20%
  Charpy V-notch at −196°C:       ≥34 J average; ≥27 J individual

2.2 Heat Treatment — Double Normalising and Quench-and-Temper

Two heat treatment routes are specified for 9Ni plate:

Type I — Double Normalising and Temper: The plate is austenitised at 900–910°C (first normalise), air cooled to room temperature, re-austenitised at 790–800°C (second normalise, within the two-phase γ+α intercritical range), air cooled, then tempered at 565–580°C. The second normalise at the lower temperature refines the austenite grain and, on air cooling, produces a finer lath martensite. The tempering stage is critical: at 565–580°C, nickel redistributes from the tempered martensite matrix to the lath boundaries, stabilising a thin film of austenite (10–15 vol%) by raising the local M_s temperature below room temperature. This temper austenite is the key microstructural feature responsible for cryogenic toughness.

Type II — Quench and Temper: Austenitise at 800–830°C, water or oil quench to produce fine lath martensite, then temper at 560–580°C. The quench produces a finer lath martensite morphology and higher initial dislocation density than air cooling, which, after tempering, gives slightly higher strength and equivalent or better toughness versus Type I for heavy sections where normalising cooling rates are slow. Type II is preferred for plate thicknesses above approximately 50 mm.

2.3 The Temper-Austenite Toughening Mechanism

The cryogenic toughness of properly heat-treated 9Ni steel far exceeds what nickel alloying alone can achieve in a fully martensitic microstructure. The critical additional contribution comes from the thin films of stable, nickel-enriched retained austenite at lath martensite boundaries formed during intercritical tempering. This temper austenite has a composition of approximately 30–35% Ni at the lath boundaries (enriched from the nominal 9% by diffusional partitioning during tempering), which lowers the M_s of this austenite well below −200°C, ensuring it remains FCC at liquid nitrogen and LNG service temperatures.

When a cleavage crack propagating through the BCC martensite matrix encounters a thin austenite film, three toughening mechanisms operate simultaneously: (1) the austenite film is ductile (FCC) and blunts the sharp crack tip by plastic deformation, absorbing energy; (2) under the triaxial stress field at the crack tip, the metastable austenite can undergo martensitic transformation, absorbing additional energy (transformation-induced plasticity effect, TRIP); (3) the austenite-martensite interface deflects the crack path, increasing the fracture surface area and the total energy absorbed. The combined effect is a dramatic increase in Charpy energy from the approximately 10–15 J a fully martensitic 9Ni microstructure would show at −196°C to the specified ≥34 J.

3. Invar (Fe-36Ni) — The Magnetovolume Effect and LNG Membrane Technology

Invar, the Fe-36Ni alloy discovered by Charles-Édouard Guillaume in 1897 (Nobel Prize, Physics, 1920), exhibits the lowest coefficient of thermal expansion of any metallic alloy over the range −100°C to +100°C — typically 1–2 × 10⁻⁶ /°C versus 12–16 × 10⁻⁶ /°C for carbon steel. This “Invar effect” is not a simple compositional effect but arises from a fundamental coupling between ferromagnetic spin ordering and lattice volume.

3.1 The Invar Effect — Physical Mechanism

In the Fe-Ni system near 36% Ni, the ferromagnetic spontaneous magnetisation M_s(T) decreases with increasing temperature over the same temperature range where thermal expansion would normally dominate. The ferromagnetic state in this composition produces a magnetostrictive volume expansion (the lattice expands when magnetised — positive magnetostriction). As temperature rises, M_s decreases, the magnetostrictive contribution decreases, and the resulting reduction in lattice volume partially cancels the positive phonon-driven thermal expansion. The net coefficient of thermal expansion is thus nearly zero.

Invar coefficient of thermal expansion (CTE) vs. Ni content:

  Fe-30Ni:    α ≈ 8×10⁻⁶ /°C   (partial Invar effect)
  Fe-36Ni:    α ≈ 1–2×10⁻⁶ /°C  (minimum CTE; maximum Invar effect)
  Fe-42Ni:    α ≈ 6×10⁻⁶ /°C   (effect diminishes above ~40% Ni)
  Fe-50Ni:    α ≈ 10×10⁻⁶ /°C  (near normal)

  Curie point of Fe-36Ni Invar: ~230–280°C
  Above Curie point: magnetostriction contribution disappears;
                     α rises abruptly to ~12×10⁻⁶ /°C

  Super Invar (Fe-31Ni-5Co):    α ≈ 0.5×10⁻⁶ /°C (lower still)
  Kovar (Fe-29Ni-17Co):         matched to borosilicate glass α = 4.6×10⁻⁶/°C

3.2 Invar in LNG Carrier Membrane Technology

The ultra-low CTE of Invar is exploited in the membrane-type LNG carrier containment system (Gaztransport & Technigaz No. 96 membrane system). The primary containment barrier consists of 0.7 mm thick Invar corrugated sheet, which contracts only 2.8 mm per metre of length when cooled from ambient to −165°C (LNG boiling point at atmospheric pressure), compared to 2.4 mm/m for stainless steel or 3.1 mm/m for aluminium. The corrugations accommodate even this small movement without stress buildup. The secondary containment barrier is also Invar sheet. The system achieves thermal insulation through glass wool and plywood insulation boxes. Approximately 40% of all LNG carriers worldwide use this membrane system; the alternative is aluminium alloy (5083-H111 or 5083-H321) Moss-type spherical tanks.

3.3 Mechanical Properties and Weldability of Invar

Invar (Fe-36Ni) has a fully austenitic (FCC) microstructure at all temperatures from below −200°C to above its 1400°C melting point, because 36% Ni is well above the level needed to stabilise austenite to absolute zero. This means Invar has no DBTT and maintains ductility at cryogenic temperatures. Its yield strength is relatively modest (~250 MPa, annealed) — significantly below 9Ni steel — but adequate for the thin-sheet membrane application where the primary stresses are thermal and pressure-induced are transmitted to the ship’s structure. Invar is welded by GTAW (TIG) using matching Invar filler (AWS A5.14 ERNiFe-Cl) with stringent control of heat input to avoid hot cracking in the fully austenitic weld pool, a concern analogous to that discussed in our article on welding austenitic stainless steel.

4. Austenitic Stainless Steels in Cryogenic Service

The austenitic stainless steels are the simplest engineering solution to cryogenic toughness — their FCC crystal structure is intrinsically ductile at all temperatures. The selection among grades for cryogenic service involves three considerations: resistance to strain-induced martensitic transformation (SIMT), avoidance of sensitisation during welding, and weld deposit ferrite content.

4.1 Material Properties and Grade Comparison

4.2 Strain-Induced Martensitic Transformation (SIMT)

Not all austenitic stainless steels are equally stable against transformation to martensite under strain at cryogenic temperatures. Grades with lower nickel content (particularly 200-series austenitic grades with Mn replacing Ni, or 301 SS with only 6–8% Ni) can transform partially to martensite (BCC α’ or HCP ε) under the high strains of forming operations or under the operating stresses in cryogenic service. This strain-induced martensite is harder and less ductile than austenite and can introduce a local BCC phase with a DBTT into an otherwise FCC component.

The martensite start temperature under deformation (M_d30) is the temperature at which 50% martensite forms under 30% true strain — a practical austenite stability index. The Angel equation gives:

Mₜ⁵₀ (°C) = 413 − 462(C+N) − 9.2Si − 8.1Mn − 13.7Cr − 29(Ni+Cu) − 18.5Mo − 68Nb

For cryogenic service selection:
  Mₜ⁵₀ < −100°C  → Acceptable stability for LN₂ service (−196°C)
  Mₜ⁵₀ < −196°C  → Stable against SIMT in LN₂ service under normal operating strains
  Mₜ⁵₀ < −253°C  → Required for LH₂ service; requires high Ni (316L min) or N additions

Worked example — typical 304L (18Cr, 10Ni, 0.03C, 0.10N, 2Mn, 0.5Si):
  Mₜ⁵₀ = 413 − 462(0.03+0.10) − 9.2(0.5) − 8.1(2) − 13.7(18) − 29(10) − 0 − 0
         = 413 − 60.1 − 4.6 − 16.2 − 246.6 − 290
         = −204°C  → stable for LNG service; borderline for LN− at strain levels

Note: Individual heat chemistry must be verified; 304L Mₜ⁵₀ can range −130°C to −220°C
      depending on exact composition within the 304L specification range.

4.3 Weld Metal Considerations — FN = 0 for LH₂

For most cryogenic applications of austenitic stainless steel, the standard weld deposit target of FN 3–8 (4–8 vol% delta ferrite — BCC) is acceptable because the ferrite content is low and distributed as discontinuous islands rather than a continuous BCC network, and the service temperature (−165°C to −196°C) does not significantly embrittle this small BCC fraction. However, for liquid hydrogen service (−253°C) and liquid helium service (−269°C), any BCC ferrite in the weld deposit can contribute to hydrogen-assisted embrittlement and toughness loss. ASME Section VIII Div. 1 and NASA design standards specify fully austenitic weld deposits (FN = 0, confirmed by magnetic measurement per AWS A4.2 or ISO 8249) for LH₂ pressure vessel welds. Fully austenitic deposits carry higher hot cracking risk — this is managed by minimising S and P content of the base metal and filler, using low heat input, and designing joints to minimise weld restraint. See our article on welding austenitic stainless steel for the full treatment of delta ferrite control and hot cracking.

5. Welding Cryogenic Steels

5.1 Welding 9% Nickel Steel

The metallurgical challenge in welding 9Ni steel is unique: the parent plate achieves its cryogenic toughness through the specific temper-austenite microstructure generated by the controlled heat treatment cycle. A weld deposit of matching 9Ni composition, if deposited and allowed to cool without the same intercritical tempering treatment, would be predominantly untempered martensite with poor cryogenic toughness (typically <15 J at −196°C, below the 34 J specification). Post-weld heat treatment of an entire LNG tank at 560–580°C after welding is generally impractical.

The engineering solution is to use high-nickel austenitic filler metals whose fully FCC weld deposit inherently maintains cryogenic toughness without any heat treatment:

Key procedural requirements for 9Ni welding (per EN 15614-1, ASME IX, and EN 14620):

• Preheat: None required. Carbon equivalent of 9Ni is low and there is no hydrogen cracking risk at ambient temperature in the as-welded condition.
• Interpass temperature: Maximum 150°C. Higher interpass temperatures increase heat accumulation, increase residual stress, and may locally sensitise the HAZ of the base plate where it passes through 425–850°C during cooling.
• Post-weld heat treatment: Generally not required and not recommended for the weld deposit itself. If local PWHT is performed on the base plate (e.g., for vessel code stress-relief requirements), it must not exceed 580°C to avoid re-dissolution of temper austenite.
• Weld procedure qualification: Mandatory Charpy testing of weld metal and HAZ at −196°C per ASTM A20/A20M and EN ISO 15614-1. Minimum 34 J average in weld metal; HAZ values are typically lower and must be evaluated case-by-case.

5.2 Welding Austenitic Stainless Steel for Cryogenic Service

The welding of 304L and 316L for cryogenic service follows standard austenitic SS practice (ER308L/ER316L filler, no preheat, interpass <150°C) with the additional requirement that weld procedure qualification includes Charpy impact testing at the minimum design temperature. For LH₂ service, FN = 0 filler is specified — typically ERNiCrFe-6 (Alloy 82 wire) or ERNiCrMo-3 (Alloy 625 wire) to achieve fully austenitic deposits. The hydrogen-induced cracking risk in austenitic welds is low due to the FCC structure’s higher hydrogen diffusivity and lower susceptibility, but post-weld degassing (low-temperature bake) is still specified for LH₂ pressure vessels per NASA-STD-5012.

6. LNG Storage Tank Design Standards and Material Specifications

The Charpy impact test is the primary acceptance criterion for all cryogenic steel plate and weld procedure qualification. Complementary fracture toughness testing by CTOD or K_IC (ASTM E1820, BS 7448) is specified in higher-consequence designs, particularly for nuclear LNG storage or hydrogen infrastructure where the consequences of failure are severe. The relationship between Charpy energy and fracture mechanics toughness K_IC — through the Barsom-Rolfe correlation — enables engineering critical assessments when full K_IC data are unavailable.

Recommended References