Retained Austenite in Steel — Causes, Measurement and Sub-Zero Treatment

When a steel component is quenched from the austenitising temperature, the austenite does not always fully transform to martensite. The fraction that survives to ambient temperature — retained austenite — is controlled by the thermodynamics of the martensite transformation, the chemical composition of the steel, and the severity of the quench. Understanding, quantifying, and eliminating (or deliberately exploiting) retained austenite is critical for tool steels, bearing steels, gear steels, and advanced high-strength automotive grades.

What Is Retained Austenite?

Austenite is the face-centred cubic (FCC) allotrope of iron, stable above the A₁ temperature (~727°C for eutectoid steel). On cooling below Ms, austenite transforms to martensite by a diffusionless shear mechanism — no atomic diffusion, no composition change, only a coordinated displacement of iron atoms that converts the FCC lattice to a body-centred tetragonal (BCT) structure. This transformation is athermal: the fraction transformed depends on how far the temperature has been reduced below Ms, not on the time held at any temperature.

The Koistinen-Marburger equation quantifies this relationship:

f_M = 1 − exp[−α(Ms − T)]

where:
  f_M  = volume fraction of martensite formed
  Ms   = martensite start temperature (°C)
  T    = current temperature (°C), T < Ms
  α    = 0.011 °C⁻¹ (empirical constant, approximately valid for most steels)

Therefore:
  f_RA = 1 − f_M = exp[−α(Ms − T)]

At T = 20°C (ambient quench end):
  f_RA = exp[−0.011(Ms − 20)]

If Ms is 300°C and the quench terminates at 20°C, the equation predicts f_RA = exp(−0.011 × 280) ≈ 4.6% — consistent with measured values in medium-carbon alloy steels. For bearing steels with Ms near 180°C, predicted retained austenite reaches 15–20%, again well-corroborated experimentally.

Thermodynamic Origin: Why Austenite Persists

The martensite transformation is driven by the difference in Gibbs free energy between austenite and martensite. Below Ms, martensite is the lower-energy phase — yet some austenite remains. This apparent paradox is resolved by considering the energy balance during transformation:

ΔG_total = ΔG_chem + ΔG_strain + ΔG_surface

For martensite formation to proceed:
  |ΔG_chem| > ΔG_strain + ΔG_surface

ΔG_chem = chemical free energy change (negative, driving force)
ΔG_strain = elastic strain energy from lattice shear (≈1.8–2.5 kJ/mol)
ΔG_surface = martensite-austenite interface energy contribution

As the martensite fraction increases, surrounding austenite is placed under compressive stress by the transformed regions (which expand ~2–4% by volume). This autocatalytic back-stress raises ΔG_strain in the untransformed regions, effectively raising the local Ms of each remaining austenite volume. When the back-stress is sufficient, further transformation is arrested even though the bulk temperature is well below the nominal Ms. This mechanical stabilisation is one of several reasons that martensite transformation rarely goes to 100% completion in a single quench.

Austenite Stabilisation Mechanisms

Stabilisation refers to any process that raises the thermodynamic resistance of austenite to martensite transformation, lowering the effective Ms. Three distinct mechanisms operate:

• Chemical stabilisation: Solute elements depress Ms by reducing the chemical driving force for transformation. Carbon is dominant; alloying elements (Mn, Cr, Ni, Mo) provide additional but smaller contributions.
• Mechanical stabilisation: Plastic deformation of austenite below Ms introduces dislocations that impede the coordinated atomic displacements required for the martensitic shear. This requires temperatures close to Ms; deformation well below Ms can instead trigger martensite (strain-induced transformation).
• Thermal stabilisation: Holding austenite at temperatures within the Ms-to-Mf range (isothermal holding) allows carbon redistribution from martensite to surrounding austenite. The enriched austenite develops a lower local Ms, resisting further transformation even on continued cooling. This is the mechanism by which tempering prior to sub-zero treatment permanently stabilises retained austenite.

Effect of Composition on Retained Austenite Content

Carbon: The Dominant Variable

The empirical Andrews (1965) equation for Ms in low-alloy steels is the standard reference:

Ms (°C) = 539 − 423[C] − 30.4[Mn] − 17.7[Ni] − 12.1[Cr] − 7.5[Mo]

Composition in wt%; valid for: C < 0.6%, Mn < 1.7%, Ni < 5%, Cr < 3.5%, Mo < 0.5%

Approximate Mf (°C) ≈ Ms − 215°C (plain carbon)
                      Ms − 180°C (alloy steels, wider scatter)

Each 0.1 wt% increase in carbon depresses Ms by approximately 33°C. The practical consequences are severe: at 0.6 wt% C, Ms is approximately 340°C and Mf around 125°C — the transformation completes well above ambient, leaving negligible retained austenite. At 1.0 wt% C (as in 100Cr6/52100 bearing steel), Ms drops to approximately 220°C and Mf to approximately 5°C — a conventional ambient quench leaves 10–25% retained austenite. At 1.5 wt% C (high-speed steel M2 after dissolution of carbides during austenitising), Mf can fall to −100°C, giving 20–40% retained austenite after oil quenching.

Alloying Element Effects

Austenitising Temperature Effects

The austenitising temperature directly governs how much carbon and alloy content dissolves into solution in the austenite prior to quenching. In tool steels and high-alloy grades that contain significant undissolved carbide fractions, increasing austenitising temperature dissolves more carbides, enriching the austenite matrix. This enrichment depresses Ms further and increases retained austenite content after quenching.

Morphology and Microstructural Identification

Where Retained Austenite Resides

In hardened low-alloy and medium-carbon steels (Ms > 200°C, low retained austenite), the small fraction of retained austenite occupies thin films between martensite laths, typically 10–50 nm thick. These films are essentially unresolvable by optical microscopy. In high-carbon bearing steels (e.g., 52100/100Cr6 with 15–25% retained austenite), larger austenite islands of 0.1–1 μm are present between coarser plate martensite variants and are resolvable by SEM/EBSD. In plate martensite steels (high carbon, >0.8 wt% C, Ms < 250°C), austenite occupies the midrib region between plate variants.

Optical and Electron Microscopy

Optical identification of retained austenite is unreliable in isolation. Freshly polished, nital-etched steel reveals retained austenite as white, featureless (unetched) regions between martensite laths or plates — but so does untempered martensite, primary ferrite, and undissolved carbides depending on the etchant and magnification. Klemm’s reagent (sodium thiosulphate + potassium metabisulphite) offers better discrimination at high magnification, staining bainite brown while leaving martensite and retained austenite white.

EBSD (Electron Backscatter Diffraction) in the SEM is the definitive microstructural technique: phase maps distinguish FCC (austenite) from BCC/BCT (martensite) with 50–100 nm spatial resolution, delivering both fraction and morphological information simultaneously. EBSD also reveals crystallographic orientation relationships (Kurdjumov-Sachs, Nishiyama-Wassermann) between retained austenite and the parent martensite, confirming the transformation history.

Quantitative Measurement of Retained Austenite

X-Ray Diffraction (XRD) — ASTM E975 / SAE J1685

XRD is the industry-standard quantitative method. The technique exploits the intensity difference between austenite FCC diffraction peaks and martensite BCC/BCT peaks. ASTM E975 specifies use of integrated peak areas from the following reflections:

Austenite (FCC) peaks:   (200)γ, (220)γ, (311)γ
Ferrite/Martensite peaks: (200)α, (211)α

Volume fraction of austenite (Vγ):

Vγ = (1/n) Σ [ Iγ / (Rγ × Cγ) ] / { [Iγ/(Rγ × Cγ)] + [Iα/(Rα × Cα)] }

where:
  Iγ, Iα = measured integrated peak intensities
  Rγ, Rα = theoretical intensity factors (tabulated in ASTM E975)
  Cγ, Cα = absorption correction factors
  n       = number of peak pairs used (minimum 2 pairs recommended)

Measurement uncertainty: ±1–2 vol% for well-prepared flat specimens
Technique: Cu Kα or Co Kα radiation; 2θ range 35°–120°

Sample preparation is critical: electropolishing to remove surface deformation layer (mechanical polishing can transform austenite to martensite by the TRIP mechanism, introducing negative bias in measured RA content). For near-surface measurements, synchrotron XRD or grazing-incidence geometry may be required.

Comparison of Measurement Methods

Effects of Retained Austenite on Mechanical Properties

Hardness

Retained austenite is substantially softer than martensite. At high carbon levels, martensite achieves HRC 62–67; austenite at the same composition has hardness of HRC 20–30 (or approximately 200–300 HV). The presence of 20% retained austenite in a hardened high-carbon steel reduces the apparent macrohardness by 3–5 HRC points. This is one reason that maximum expected hardness is not achieved in over-austenitised tool steels or inadequately quenched bearing races.

Dimensional Stability

Retained austenite is metastable and gradually transforms to martensite during service at ambient temperature (self-tempering or ageing), or more rapidly under applied stress (TRIP mechanism) or elevated temperature. The martensite transformation is accompanied by a volume expansion of approximately 2–4% (greater for higher carbon content). In precision components — gauge blocks, precision spindle bearings, injection mould cores, thread gauges — this transformation-induced dilatation causes dimensional change over time, shifting component geometry outside tolerance.

This is the primary reason that precision tool steel and bearing steel components are specified with retained austenite limits (typically <5–8 vol%) and are subjected to mandatory sub-zero treatment before final grinding.

Fatigue Life in Rolling Contact

In rolling contact fatigue (bearings, gears), retained austenite plays a complex dual role:

• Detrimental effect (high RA): Soft austenite volumes act as preferential sites for subsurface shear band initiation; transformation to martensite during cycling produces local volume expansion, generating compressive residual stress that can initiate delamination cracks. Above ~15 vol%, rolling contact fatigue life is reduced.
• Beneficial effect (controlled RA, 5–12 vol%): Stress-assisted transformation at the subsurface shear stress maximum introduces compressive residual stresses that retard crack initiation — the TRIP mechanism operating beneficially under cyclic loading. Some carburised gear steels are engineered to retain 10–15% austenite in the case for this reason.

Wear Resistance and Toughness

For abrasive and sliding wear applications (tool steels, dies), high retained austenite is generally detrimental because it reduces hardness and provides soft regions for ploughing. However, in impact wear scenarios, austenite’s ability to strain-harden rapidly via TRIP provides better resistance to chip formation than fully brittle martensite. The appropriate retained austenite target is therefore application-dependent: <5% for gauge and precision tools; 5–15% for cold-work dies; up to 25% in some impact-resistant grades where transformation toughening is desired.

Sub-Zero and Cryogenic Treatment

Principles of Sub-Zero Treatment

Since the martensite transformation is athermal, reducing temperature below the quench-end point progressively converts additional austenite to martensite. Sub-zero treatment exploits this by cooling hardened components to −60 to −100°C immediately after quenching, before any tempering. The treatment must be carried out promptly: even brief holding at ambient allows the thermal stabilisation mechanism to begin, partially protecting the retained austenite from further transformation.

After conventional quench to 20°C (Ms = 200°C bearing steel):
  f_RA(20°C)  = exp[−0.011 × (200 − 20)] = exp[−1.98] ≈ 13.8%

After sub-zero to −80°C:
  f_RA(−80°C) = exp[−0.011 × (200 − (−80))] = exp[−3.08] ≈ 4.6%

Reduction: 13.8% → 4.6% (≈ 67% decrease in retained austenite volume fraction)

Practical Sub-Zero Methods

Cryogenic Treatment: Carbide Precipitation Effects

Deep cryogenic treatment (DCT) at −196°C differs from conventional sub-zero treatment not only in the degree of austenite conversion but also in a secondary effect: precipitation of fine transition carbides (η-carbide, Fe₂C) from the supersaturated martensite during the extended soak. This is thermally activated even at −196°C over multi-hour soaks, because carbon diffusion coefficients in BCT martensite, though orders of magnitude below ambient-temperature values, are non-zero.

The resulting carbide dispersion is finer and more uniform than that produced by conventional tempering, contributing to:

• Improved wear resistance (fine carbide distribution increases matrix hardness homogeneity)
• Marginally higher hardness after subsequent tempering (typically +1–2 HRC in tool steels)
• Better dimensional stability (less carbon in solution means smaller subsequent martensite decomposition strains)

Sequence Requirements: Why Tempering Must Follow Sub-Zero Treatment

Untempered martensite is in a high-stress, metastable state with high dislocation density. Sub-zero treatment without subsequent tempering would leave the component in this vulnerable condition. The mandatory sequence is:

1. Quench (oil, polymer, or salt bath as required by steel grade)
2. Sub-zero or cryogenic treatment — as soon as practical, typically within 1 hour of quench completion
3. Warm to ambient — allow slow, uniform warm-up to avoid thermal shock in complex geometries
4. Temper immediately — single or double temper per steel manufacturer specification

Double tempering is specified for most high-speed and high-alloy tool steels: the first temper cycle relieves quench stresses and partially decomposes martensite; the second temper cycle tempers any fresh martensite formed from secondary austenite transformation during the first temper cool-down, and further precipitates carbides. Sub-zero treatment between the first and second temper cycles is occasionally specified for extreme-precision applications.

Beneficial Retained Austenite: TRIP and Q&P Steels

TRIP Steels (Transformation-Induced Plasticity)

In deliberate contrast to the tool steel and bearing steel philosophy, advanced high-strength steel (AHSS) engineers design metastable retained austenite into the microstructure of first- and second-generation TRIP steels. These steels (typical composition: 0.15–0.25 wt% C, 1.5–2.0 wt% Mn, 1.2–1.8 wt% Si) are intercritically annealed in the α+γ two-phase field, then isothermally held in the lower bainite region (300–450°C) long enough to stabilise austenite with carbon enrichment (from bainite decomposition) without precipitating carbides — which silicon suppresses effectively.

The resulting microstructure contains 10–20 vol% metastable austenite with carbon content elevated to ~1.2–1.8 wt%, surrounded by a matrix of ferrite and bainite. Under tensile deformation:

Stress-assisted transformation (T > Md):
  Applied stress supplements ΔG_mech to reach G_martensite
  Austenite → martensite at stresses below macroscopic yield

Strain-induced transformation (T < Md, near Md):
  Plastic deformation of austenite creates shear band nucleation sites
  Martensite nucleates at shear band intersections
  Rate depends on strain, strain rate, and austenite stability parameter Md30

Md30 temperature (°C) = 551 − 462[C+N] − 9.2[Si] − 8.1[Mn] − 13.7[Cr] − 29[Ni+Cu] − 18.5[Mo]
(For 304/316 austenitic stainless; TRIP steel equivalents require modified coefficients)

The TRIP effect provides two mechanical benefits: local compressive residual stress from the ~4% volume expansion stabilises the propagating crack tip; and the increased dislocation density from martensitic shear increases local work hardening rate, maintaining high strength at large strains. TRIP steels achieve tensile strengths of 800–1200 MPa with total elongations of 25–40%, enabling crash energy absorption superior to conventional high-strength steels of equivalent weight.

Quench and Partition (Q&P) Steels

Q&P processing, developed by Speer et al. (2003), is a second-generation approach that engineers higher retained austenite fractions and greater thermal stability than conventional TRIP processing allows. The steel is quenched to a temperature between Ms and Mf (the quench stop temperature, QT), producing a predetermined martensite fraction; it is then heated to a partition temperature (typically 250–450°C) where carbon partitions from supersaturated martensite into austenite along the martensite-austenite interface. The carbon-enriched austenite is thermally stable on further cooling, giving retained austenite fractions of 15–40% with local carbon content of 1.0–1.5 wt%.

Q&P steels are increasingly specified in automotive body-in-white structures where tensile strength targets of 1180–1470 MPa are combined with elongation requirements of 15–25% that cannot be achieved by dual-phase or martensitic grades alone.

Retained Austenite in Specific Steel Families

Bearing Steels (100Cr6 / 52100)

Through-hardened 100Cr6 bearing steel is the canonical case. Austenitised at 830–850°C (dissolving a fraction of the chromium carbide population into austenite), oil-quenched to give Ms near 200°C, and quenching to ambient produces 15–25% retained austenite depending on austenitising temperature and residual carbide dissolution. Sub-zero treatment at −75 to −80°C for 1–4 hours reduces this to 3–8%, followed by tempering at 150–180°C for 1–2 hours. Final microstructure: tempered martensite matrix, undissolved spheroidal carbides (Cr₇C₃ / M₂₃C₆), retained austenite <8 vol%.

High-Speed Steels (M2, M42)

High-speed steels are the most complex retained austenite management challenge. Austenitising temperatures of 1200–1280°C (M2: 1220–1240°C) dissolve large quantities of W, Mo, V, and Cr carbides, enriching the austenite matrix. This highly alloyed austenite has Ms near 150–200°C and Mf near −80 to −120°C. After quenching to ambient, retained austenite content is 25–40%. Triple tempering at 540–560°C is standard: each temper cycle precipitates secondary carbides (Mo₂C, VC, W₂C) that deplete the austenite of solute, raising Ms of remaining austenite and enabling it to transform during cooling between temper cycles. Sub-zero treatment between the first and second temper further reduces RA before the temper-induced carbide precipitation cycle begins.

Cold-Work Tool Steels (D2, D3)

D2 (1.5 wt% C, 12 wt% Cr) austenitised at 1010–1040°C retains 20–35% austenite after oil or air quenching. The high chromium content forms large primary M₇C₃ carbides that do not dissolve at practical austenitising temperatures, leaving the matrix austenite at ~1.0 wt% C and ~5 wt% Cr dissolved — still sufficient for a depressed Ms near 200°C and Mf near 0°C. Sub-zero or cryogenic treatment followed by double tempering at 150–200°C is standard, targeting final RA <10 vol%.

Detection in Failure Analysis

Retained austenite is a recurring finding in steel component failure investigations. Common scenarios include:

• Dimensional drift in precision components: A hardened gauge block or precision spindle showing gradual growth over months of service; XRD of the returned part reveals 12–18% retained austenite, indicating sub-zero treatment was omitted from the heat treatment route.
• Softness below specification: A tool steel punch measuring HRC 58 when HRC 62–64 is specified; EBSD reveals 22% retained austenite; root cause traced to over-austenitising temperature raising carbon dissolution beyond design intent.
• Early spalling in bearings: Subsurface cracks at 100–200 μm depth in a bearing race with 25% retained austenite; the transformed zone around spall shows fresh untempered martensite, confirming stress-assisted transformation during service was the triggering mechanism.

For techniques supporting these investigations, see our guides on hardness testing methods and martensite formation in steel. Understanding the full iron-carbon phase diagram is also essential background for interpreting retained austenite levels in context of the alloy’s equilibrium phase relationships.

The quenching and tempering heat treatment process must be understood in conjunction with sub-zero treatment to fully control retained austenite. Similarly, annealing and normalising before hardening can influence the carbide population and hence the subsequent austenite composition during the hardening cycle.

Retained austenite also has implications in heat-affected zone microstructure in welding of hardenable steels, where the HAZ thermal cycle may partially re-austenitise the base metal and produce untransformed austenite depending on cooling rate and carbon equivalent. In carburised grades, it is closely related to the bainite microstructure in the transition zone and to pearlite colony growth considerations in hypoeutectoid subcases.

Grain boundary condition during austenitising is relevant because austenite grain boundaries are preferential nucleation sites for martensite; coarser austenite grains (from over-austenitising) produce coarser martensite packets, which independently reduces toughness beyond the retained austenite effect.

References and Further Reading

• Bhadeshia, H.K.D.H. and Honeycombe, R., Steels: Microstructure and Properties, 4th ed. Butterworth-Heinemann, 2017.
• Krauss, G., Steels: Processing, Structure, and Performance, 2nd ed. ASM International, 2015.
• ASM Handbook Vol. 9: Metallography and Microstructures. ASM International, 2004.
• ASM Handbook Vol. 4: Heat Treating. ASM International, 1991.
• ASTM E975-22: Standard Practice for X-Ray Determination of Retained Austenite in Steel with Near Random Crystallographic Orientation.
• Andrews, K.W., Empirical formulae for the calculation of some transformation temperatures. J. Iron Steel Inst., 203, 721–727, 1965.
• Koistinen, D.P. and Marburger, R.E., A general equation prescribing the extent of the austenite-martensite transformation in pure iron-carbon alloys. Acta Metall., 7, 59–60, 1959.
• Speer, J., et al., Carbon partitioning into austenite after martensite transformation. Acta Mater., 51, 2611–2622, 2003.

Recommended References

Further Reading