Materials Science and Technology

ISSN: 0267-0836 (Print) 1743-2847 (Online) Journal homepage: https://www.tandfonline.com/loi/ymst20

Austenite reversion in low-carbon martensitic

stainless steels – a CALPHAD-assisted review

Frank Niessen

To cite this article: Frank Niessen (2018) Austenite reversion in low-carbon martensitic stainless
steels – a CALPHAD-assisted review, Materials Science and Technology, 34:12, 1401-1414, DOI:
10.1080/02670836.2018.1449179

To link to this article: https://doi.org/10.1080/02670836.2018.1449179

Published online: 20 Mar 2018.

Submit your article to this journal

Article views: 360

View Crossmark data

Citing articles: 2 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=ymst20
MATERIALS SCIENCE AND TECHNOLOGY
2018, VOL. 34, NO. 12, 1401–1414
https://doi.org/10.1080/02670836.2018.1449179

REVIEW

Austenite reversion in low-carbon martensitic stainless steels

– a CALPHAD-assisted review
Frank Niessen a,b
a The Danish Hydrocarbon Research and Technology Centre, Technical University of Denmark (DTU), Kongens Lyngby, Denmark;
b Department of Mechanical Engineering, Technical University of Denmark (DTU), Kongens Lyngby, Denmark

ABSTRACT ARTICLE HISTORY

Low-carbon martensitic stainless steels with 11.5–16 wt-% Cr and < 0.07 wt-% C are charac- Received 12 January 2018
terised by high corrosion resistance, strength, ductility and impact toughness, obtained by Revised 27 February 2018
formation of fine-grained reverted austenite from lath martensite upon inter-critical anneal- Accepted 1 March 2018
ing. The review treats the mechanisms governing the formation and stabilisation of reverted KEYWORDS
austenite and is assisted by the computation of phase equilibria. Literature data on Cr and Ni con- Austenite reversion;
centrations of the reverted austenite/martensite dual-phase microstructure are assessed with microstructure evolution;
respect to predicted concentrations. Reasonable agreement was found for concentrations in solute partitioning;
martensite. Systematic excess of Cr in austenite of approx. 2 wt-% relative to calculations was thermodynamics; CALPHAD;
suspected to originate from the growth of M23 C6 with a coherent interface to austenite. Within diffusion; martensite
large scatter, measured values of Ni in austenite were on average 2 wt-% below predictions. Com- formation; residual stress
plex equilibration of the microstructure and experimental error are discussed as possible origins
of the discrepancies.

This review was chosen as a runner up of the 2018 Materials Literature Review Prize of the Institute of Materials, Minerals and Mining, run by the Editorial
Board of MST. Sponsorship of the prize by TWI Ltd is gratefully acknowledged.

Introduction
experimental and modelling methods enabled the pro-
Low-carbon martensitic stainless steels comprise the gressive transition from a processing-property-based
group of supermartensitic stainless steels, soft marten- approach to a microstructure-property-based under-
sitic stainless steels and precipitation hardening versions. standing of these materials. It therefore appears timely
While the individual alloy groups are optimised to review the present understanding of the reverse
towards different application fields such as weldabil- martensite-to-austenite phase transformation. As the
ity, corrosion resistance or hardenability, all contain a title suggests, this review includes phase equilibrium
nano-lamellar dual-phase microstructure of reverted calculations (so-called CALPHAD approach) to sup-
austenite and tempered martensite through inter- port the discussion.
critical annealing, i.e. annealing in the temperature The microstructure-property-based characterisa-
region in which both ferrite and austenite are ther- tion of the alloys in recent years has led to a wealth
modynamically stable. The obtained ‘reverted austen- of quantitative literature data on the partitioning of
ite’ is distinguished from ‘retained austenite’, i.e. that Ni and Cr after partial reversion of austenite during
is untransformed during cooling to room temperature inter-critical annealing of martensite. As reported data
[1]. Stabilisation of reverted austenite against marten- concern different alloy systems and annealing param-
site formation occurs primarily by the partitioning of eters, and as discussion of the data is generally limited
austenite stabilising elements during diffusional rever- to the scope of the specific work, a collective represen-
sion. The resulting fine-grained dual-phase microstruc- tation of the data is established here to reveal under-
ture lowers the yield strength, ultimate tensile strength lying trends. As austenite reversion during isothermal
and hardness, while ductility and impact toughness are annealing is accompanied by diffusion, it is of interest
significantly enhanced [2–4]. to verify whether the experimentally determined con-
The mechanisms leading to the reversion of austen- centrations can be predicted by phase equilibria from
ite and the analysis of the stability of reverted austenite thermodynamics modelling. Finally, the early stage of
against martensite formation to thermal or mechanical austenite reversion close to A1 and the findings from
treatments are the subject of numerous research arti- the analysis of compositional data from the literature
cles published within the last decades. Newly emerging are critically discussed.

CONTACT Frank Niessen [email protected] The Danish Hydrocarbon Research and Technology Centre, Technical University of Denmark
(DTU), Produktionstorvet, Building 425 Room 111, 2800 Kongens Lyngby, Denmark; Department of Mechanical Engineering, Technical University of Denmark
(DTU), Produktionstorvet, Building 425 Room 111, 2800 Kongens Lyngby, Denmark

© 2018 Institute of Materials, Minerals and Mining.

1402 F. NIESSEN

Scope isothermal tempering to avoid reaustinitisation during

tempering, and kept M s above 200°C to ensure com-
The present review is limited to low-carbon marten-
plete transformation to martensite during cooling [9].
sitic stainless steels with 11.5–16 wt-% Cr, 4–8 wt-%
An Fe–0.1C–12Cr–2Ni–1.5Mo–0.3V (wt-%) alloy was
Ni and low interstitial content, C < 0.07 wt-% and
found most promising with respect to strength and
N < 0.06 wt-%, for which a fine-grained dual-phase
impact toughness [9] (Table 1).
structure of reverted austenite and tempered marten-
Additional stabilisation of austenite was necessary to
site forms by a diffusional mechanism during inter-
maintain a transformable steel when increasing the Cr
critical annealing. In some steels, austenite stability is
content for enhanced corrosion performance. As fur-
sufficiently high for austenite reversion to occur by a
ther addition of C, one of the most effective austenite
displacive mechanism at temperatures where a diffu-
stabilising elements, would lead to M23 C6 precipitation
sion controlled transformation is kinetically suppressed
and thus deplete Cr, other substitutional solutes needed
[5,6], which is not treated in this review. Austenite
to be considered instead [12]. Ni proved to be the most
reversion and stabilisation by a diffusional mechanism
effective element, and substitution of C with Ni fur-
is not limited to the alloys discussed here and can
ther led to a useful softening of virgin martensite, which
also be found in other alloy groups, as for example in
manifests as increased impact toughness [12,16,17].
medium manganese steels [7,8]. The mechanisms dis-
Following this paradigm, the Swedish steel manu-
cussed in the present work can readily be transferred
facturers Bofors and Avesta reduced the C content to
to other alloy systems, but that is beyond the scope of
0.06 wt-% to develop soft martensitic stainless steels
this review. The focus of this review lies on phase trans-
with good corrosion resistance, high strength and high
formations rather than alloy properties. Only a short
impact toughness for pressure vessel applications in the
section is dedicated to the effect of reverted austenite
mid-60s of the last century [3,18–20] (Table 1). These
on microstructure properties, including references for
steels contained small amounts of retained austen-
further reading.
ite and formed reverted austenite upon inter-critical
annealing. Precipitation hardening steels are designed
Current view on austenite reversion with a similar base alloy composition, but solutes such
as Cu, Mo and Nb enable precipitation hardening
Historical aspects and alloy design of low-carbon [4,21,22].
martensitic stainless steels Further advances in steelmaking enabled the devel-
These alloys are built on the Fe–Cr–Ni ternary sys- opment of modern supermartensitic stainless steels in
tem with additions of Mn, Mo, Si and particularly the 90s of the last century, which generally consist of
low contents of the interstitial elements C and N. 10.5–13.5 wt-% of Cr to enable passivation, very low
In 1960, Irvine et al. [9] comprehensively described C content (preferably < 0.01 wt-%) to enhance weld-
aspects of designing transformable 12% Cr steels to ability and impact toughness, around 4–6 wt-% Ni to
obtain a strong and corrosion resistant material. A enable martensite formation, and increased Mo content
major challenge consisted of finding alloy composi- (0.5–2.5 wt-%) to enhance resistance to localised corro-
tions that could be solution treated without form- sion and sulphide stress cracking [12,16,17,23–25] (cf.
ing δ-ferrite, maintained A1 above 700°C during 5 h overview in Table 1).

Table 1. Overview of typical alloy compositions and average mechanical properties of soft martensitic and supermartensitic
stainless steels for specified annealing treatments with reference to conventional martensitic stainless steel.
Annealing
treatment T
Alloy designation Reference Composition (wt-%) (°C)/t (h) YS (MPa) UTS (MPa) A (%) KVRT (J/cm2 )
Conventional martensitic stainless steel
Generic Irvine et al. [9] Fe–0.1C–12Cr–2Ni–1.5Mo–0.3V 650/1 670 860 19 ≥ 68
Soft martensitic stainless steels
Bofors 2RMO Grounes and Rao Fe–0.06C–13Cr–6Ni–1.5Mo–0.6Mn–0.4Si 590 ≥ 620 ≥ 830 ≥ 15 105
[3]
Avesta 248SV Grounes and Rao Fe–0.035C–16Cr–5Ni–1Mo–0.8Mn–0.5Si 580 ≥ 620 ≥ 830 ≥ 15 132
[3]
EN 1.4405 (cast alloy) Niederau [2] Fe– < 0.07C–16Cr–5Ni–1.5Mo– < 1Mn– < 1Si 580 650 900 ≥ 17 105
EN 1.4418 Dawood et al. Fe–0.05C–15.5Cr–5.6Ni–0.76Mo–0.4Mn–0.4Si 625/4 690 880 10 260
[10,11]
Supermartensitic stainless steels
13CrS (UNS S41525) Kondo et al. Fe–0.01C–12Cr–6Ni–2.5Mo–Ti 850/0.5 + 630/1 ≥ 550 ≥ 750 / 290
[12,13]
Vítkovice, generic Tvrdy et al. [14] Fe–0.017C–13Cr–6.2Ni–2.4Mo–0.6Mn–0.3Si 600/6 540 870 19 185
Industeel, generic Toussaint and Fe–0.008C–11.8Cr–4.8Ni–1.5Mo–1.9Mn– 625 600–650 880–900 25 195
Dufrane [15] 0.2Si–0.5Cu
Note: YS, Yield strength; UTS, ultimate tensile strength; A, elongation until rupture and KVRT , Charpy V impact toughness at room temperature.
 MATERIALS SCIENCE AND TECHNOLOGY 1403

Solution treatment, martensite formation and magnitude of secondary hardening in specific alloys is
tempering of martensite sensitive to the amount of available interstitial elements
and the content of precipitate forming substitutional
Before austenite reversion from martensite is obtained
alloying elements. During further heating, concur-
by inter-critical annealing, low-carbon martensitic
rent recovery of the martensite matrix and austenite
stainless steels are solution treated in the austenite
reversion at grain boundaries commence at approx.
single-phase region to obtain homogeneous austenite
500–550°C [45,51], leading to a reduction in disloca-
as a parent phase for lath martensite. Some alloys con-
tion density by approx. an order of magnitude [45].
tain micro-alloying elements, such as Ti, Nb or V, to
form carbides and nitrides that pin the austenite grain
Nucleation and growth close to A1
boundaries during solution treatment to limit grain
growth [23,26]. Austenite transforms to martensite dur- A1 , i.e. the ferrite-to-austenite (α-to-γ ) phase tran-
ing cooling over a narrow transformation range, with sition temperature in thermodynamic equilibrium,
martensite formation starting at a relatively low tem- may conventionally be obtained from thermodynam-
perature (Ms ≈ 260 − 130◦ C) and finishing just above ics modelling. In the present system, the prediction of
[27–32], or in exceptional cases even below [2,33], extensive partitioning of solute in austenite and ferrite
room temperature. Martensite forms with less than ∼ at low temperature leads to a small fraction of sta-
2◦ deviation [34] from a Kurdjumow–Sachs orientation ble austenite in thermodynamic equilibrium. In reality,
relationship with austenite, (111)γ  (011)α , [101̄]γ  an alloy undergoes martensite formation during cool-
[11̄1]α [35]. Even though the exact nature of the inter- ing, such that austenite forms from an approximately
face associated with the orientation relationship is still homogeneous distribution of solute during reheating,
subject of current research [36,37], it has to be semi- rendering the prediction of A1 from thermodynamic
coherent, consisting of periodic steps with coherent equilibrium unfeasible. Ac1 , the experimentally deter-
patches [38,39]. The low-carbon martensitic stainless mined α-to-γ transition temperature during heating,
steels have a high hardenability and are insensitive is on the other hand strongly dependent on the heating
to the applied quenching rate [4,15,16]. Recent work rate [9,27,52,53] and thus not an explicit quantity. In the
showed that, as a consequence of the low MS temper- present case, the phase transition temperatures A1 and
ature and interstitial content, no interstitial segregation A3 therefore refer to the phase transition temperatures
by auto-tempering or six months of room tempera- obtained during very slow heating (≤ 0.017 K s−1 ). Suf-
ture ageing was observed [40]. In contrast, cluster- ficiently fast heating (approx. >10 K s−1 [54,55]) does
ing of carbon at room temperature was reported in not allow enough time for long-range diffusion and
Ref. [41], without specifying the ageing time. Marten- thus leads to transformation by a displacive mechanism
sitic transformation leads to transformation strains, instead.
which may be partially accommodated by retained Close to A1 (composition dependent at ∼ 500–
austenite. During the transformation, martensite ini- 550°C) allotriomorphic reverted austenite with film
tially experiences high tensile stress, which then grad- morphology forms at lath boundaries [55–60] with
ually decreases towards the end of the transformation little or no deviation from the Kurdjumow–Sachs
(down to ∼ 40 MPa), while retained austenite experi- orientation relationship [35,61,62]. All reported micro-
ences significant compressive stress towards the end of graphs of annealed microstructures in the tempera-
the transformation (up to ∼ −900 MPa) [42]. ture range 500–575°C (cf. references list in Table 2)
Lath martensite shows very high dislocation densi- reveal that austenite grows from the lath boundary
ties, similar to heavily cold-worked alloys [43]. Quan- into only one of the laths (Figure 1(a)) [1,55,60,66].
titative studies on dislocation densities revealed ∼ 4 × Above this temperature, austenite films begin to grow
1015 m−2 in an Fe–0.03C–15.5Cr–5Ni (wt-%) precip- into both laths adjacent to a lath boundary (Figure
itation hardening steel [44] and ∼ 7 × 1014 m−2 in an 1(b)) [55,58,60]. In the case of pre-existing inter-lath
Fe–0.04C–15.5Cr–5Ni (wt-%) soft martensitic stainless retained austenite, immediate growth from retained
steel [45] in as-quenched conditions. austenite occurs, as no nucleation is required [59].
When the dislocation-rich microstructure is rehea- For the film morphology, austenite memory, i.e. the
ted to the inter-critical region, C and N partition tendency of reverted austenite to form in the ori-
from solid solution and diffuse to lattice defects in entation of the prior austenite grain, is commonly
order to minimise local strain fields [40]. Because of observed [27,59,71]. Further, reverted austenite has
the low interstitial content, no transition carbides are been reported to form cooperatively with M23 C6
formed during the tempering of martensite [21,46]. carbides with a cube-cube orientation relationship,
Tempering of Mo-containing steels at 400°C [21] and {100}γ {100}M23 C6 , 001γ 001M23 C6 [33,41,58,60,67].
Cu-containing steels at 450–480°C [29,47–50] leads to According to thermodynamic equilibrium, growth of
secondary hardening by precipitation of Mo2 C and Cu austenite requires an inward-flux of Ni and outward-
precipitates, respectively. Generally, the kinetics and flux of Cr [27] (Figure 2), while growth of M23 C6
1404 F. NIESSEN

Table 2. Overview of literature data, x meas , on measured Ni and Cr concentrations after partitioning in austenite (fcc) and martensite
(bcc) compared with data from phase equilibria, x equ .
fcc,Ni fcc,Cr bcc,Ni bcc.Cr x fcc.Ni x fcc.Cr x bcc.Ni x bcc.Cr
xequ xequ xequ xequ meas meas meas meas T t Methods
Label Ref (wt-%) (°C) (h) kequ Exp. Prep.
Yuan Gong Sun et al. 2016 A [60] 24.5 11.0 3.8 15.8 9.6 / 3.8 / 500 4 0.00 TEM-EDS EP
Lee Shin Leem et al. 2003 A [55] 16.5 11.1 5.0 13.7 11.0 14.0 / / 575 2 0.02 TEM-EDS EP
Schnitzer Radis Nöhrer et al. 2010 A [1] 16.3 11.5 4.5 13.1 16.8 13.1 4.0 12.5 575 10 0.09 APT EP
Schnitzer Radis Nöhrer et al. 2010 B [1] 16.3 11.5 4.5 13.1 14.0 12.7 3.4 12.3 575 100 0.87 APT EP
Schnitzer Radis Nöhrer et al. 2010 C [1] 16.3 11.5 4.5 13.1 14.4 13.9 / / 575 10 0.09 TEM-EDS EP
Schnitzer Radis Nöhrer et al. 2010 D [1] 16.3 11.5 4.5 13.1 14.3 13.1 / / 575 100 0.87 TEM-EDS EP
Song Li Rong et al. 2010 [58] 12.9 11.4 3.3 12.1 9.0 13.0 3.5 11.3 580 2 0.02 TEM-EDS EP
Song Li Rong et al. 2011 [63] 12.9 11.4 3.3 12.1 13.0 13.0 3.0 6.3 590 0.5 0.01 APT ?
Bilmes Solari Llorente 2001 [64] 12.1 11.2 3.5 12.1 8.3 13.7 / / 600 2 0.09 TEM-EDS EP
Song Ping Yin et al. 2010 [41] 11.9 10.7 3.3 11.3 13.3 11.8 3.0 7.5 600 4 0.17 APT EP
Ye Li Jiang et al. 2012 [65] 13.0 12.9 3.7 16.3 11.9 14.7 4.7 17.0 600 2 0.09 TEM-EDS EP
Yuan Gong Sun et al. 2016 B [60] 12.6 13.7 3.5 15.9 11.0 / 4.1 / 600 4 0.17 TEM-EDS EP
Song Li Rong et al. 2014 [66] 10.8 11.7 3.2 12.1 6.1 15.7 / / 610 2 0.16 STEM-EDS EP
Zhang Wang Li et al. 2015 [67] 9.7 12.5 3.1 13.0 10.0 13.3 2.7 11.3 620 6 0.92 TEM-EDS EP
Wang Lu Li et al. 2008 [68] 11.1 10.9 3.4 11.8 6.9 / 4.2 / 620 0.25 0.01 TEM-EDS ?
Lee Shin Leem et al. 2003 B [55] 11.7 12.1 4.3 14.0 8.5 15.5 / / 625 2 0.42 TEM-EDS EP
Liu Yang Zhang 2013 [59] 10.5 11.7 3.2 12.8 8.9 / / / 625 2 0.42 TEM-EDS EP
Escobar Poplawsky Faria et al. 2017 A [69] 10.8 11.3 3.3 12.3 11.6 10.3 3.6 10.9 625 2.5 0.53 STEM-EDS FIB
Escobar Poplawsky Faria et al. 2017 B [69] 10.8 11.3 3.3 12.3 9.7 11.5 3.5 11.6 625 2.5 0.53 APT FIB
Wang Xiao Lu et al. 2013 [70] 8.6 12.3 2.8 12.7 11.6 / 2.5 / 630 2 0.58 TEM-EDS EP
Nakada Tsuchiyama Takaki et al. 2007 [71] 10.6 12.2 0.4 13.4 9.2 / / / 640 0.5 0.27 TEM-EDS ?
Yuan Gong Sun et al. 2016 C [60] 8.7 13.7 3.1 16.4 7.5 / 4.2 / 650 4 4.16 TEM-EDS EP
Jiang Zhao Ye et al. 2013 [72] 9.0 13.5 3.0 16.9 7.1 15.3 2.1 17.1 650 2 2.08 TEM-EDS EP
Jiang Ye Li et al. 2014 [73] 9.1 13.6 3.0 17.2 11.7 15.0 3.0 20.1 650 2 2.08 TEM-EDS ?
Niessen Grumsen Hald et al. 2017 [62] 8.2 13.7 2.9 16.2 8.0 16.0 3.0 16.0 650 2 2.08 STEM-EDS EP
Lee Shin Leem et al. 2003 C [55] 8.5 12.7 3.7 14.6 7.5 14.0 / / 675 2 10.26 TEM-EDS EP
De Sanctis Lovicu Valentini et al. 2015 [33] 6.6 13.7 2.6 16.0 6.7 16.9 4.2 16.8 680 1 7.06 TEM-EDS EP
Notes: T is the annealing temperature, t is the annealing time and kequ is the equilibration factor. The experimental method and the sample preparation method
are given, where EP is electro-Polishing and FIB is focused ion beam.

requires the opposite fluxes. Cooperative growth should required partitioning of Ni (Figure 2), accelerated dif-
thus be facilitated by a gradient in chemical potential fusion kinetics and increased prominence of bulk dif-
and a resulting net flux of Ni and Cr across the interface. fusion. Further, the increased driving force for austen-
It was shown experimentally that M23 C6 can bind suffi- ite formation in this temperature range renders also
cient carbon to leave austenite and martensite virtually incoherent interfaces [56] or the formation of new
carbon free [41,63,69]. interfaces energetically favourable [55,60,76], mani-
Figure 2 shows the equilibrium austenite fraction fested as more nucleation sites. It was found that in
and Cr and Ni concentrations in austenite and ferrite an Fe–0.05C–12Cr–4Ni–0.5Mo (wt-%) steel austenite
as a function of temperature for a simple Fe–Cr–Ni reversion occurred without diffusion during isother-
system with average Ni and Cr concentrations of the mal annealing above 680°C [63]. Upon further heat-
analysed literature data in Table 2. At low tempera- ing grain growth and dissolution of M23 C6 carbides
ture, enhanced partitioning of Ni is required to form continue towards a fully austenitic microstructure,
austenite which, together with slow substitutional diffu- which was reported to recrystallise spontaneously
sion kinetics, significantly limits the kinetics of austen- at 900°C when heating with 0.17 K s−1 , 70°C above
ite reversion [27,62,75]. Nevertheless, the kinetics of Ac3 [57].
austenite reversion at low temperature were measured
to be significantly faster than predicted by modelling
of bulk diffusion, suggesting that grain boundary diffu- Stability of reverted austenite against martensite
sion and diffusion along dislocations may be important formation
mechanisms that significantly increase the transforma- Reverted austenite that is formed close to A1 is generally
tion kinetics at these temperatures [27,33,75]. more stable against martensite formation upon cool-
ing (or deformation) as compared to reverted austenite
formed at higher temperature. There is a consensus that
Nucleation and growth towards A3
Ni-enrichment in austenite decreases with increasing
At elevated temperature, typically 600–700°C, reverted annealing temperature [27,29,31,41,55,59,62–64,66,67,
austenite tends towards a globular morphology, first 69,70,73,75,77], which is in qualitative agreement with
at prior austenite grain boundaries (Figure 1(c)) the concentrations from thermodynamic equilibrium
[56,57,59] and at higher temperature within marten- in Figure 2. The Ni concentration determines mainly
site laths (Figure 1(d)) [55,60,76]. The reversion kinet- the stability of reverted austenite against marten-
ics are significantly faster, mainly because of lower site formation, as more Ni reduces M s . However,
 MATERIALS SCIENCE AND TECHNOLOGY 1405

Lath boundaries Lath boundaries

(d)

Globular Globular
austenite austenite

200 nm

(c) Prior austenite

grainboundaries

Globular Globular
austenite austenite

Increasing temperature
200 nm
Lath boundaries Lath boundaries
(b)

Austenite Austenite

200 nm

Lath boundaries Lath boundaries

(a)

Austenite

Austenite

200 nm

Increasing time
Figure 1. Bright-field micrographs and schematics of the evolution of the reverted austenite morphology with temperature and
time in low C martensitic stainless steels: (a) low temperature film morphology (Ref. [66]); (b) elevated temperature film morphology
(Ref.[62]); (c) globular morphology at prior austenite grain boundaries (Ref. [1]); (d) globular morphology inside martensite laths (Ref.
[58]); reverted austenite is marked with white arrows in the micrographs. Permissions for the use of figures from the literature are
automatically granted according to the STM signatory guidelines.

the compositional effect is considered insufficient to of austenite approached equilibrium content, compo-
explain the stability of reverted austenite alone [41,64]. sitional equilibration (partitioning), carbide growth,
The increase in grain size [66,67,78,79], the transition recovery of martensite and spheroidisation of austen-
to a more globular grain morphology [66,67] and soft- ite continue [67,69], which may affect the stability of
ening of the surrounding martensitic matrix [11,67] reverted austenite.
with increasing annealing temperature are anticipated Bilmes et al. [64] claimed that also a high dislo-
to reduce the contribution of strain energy to the critical cation density in reverted austenite could contribute
driving force for martensite nucleation [80], i.e. the sta- to the stability of reverted austenite. This finding is
bility of austenite. Findings by Zhang et al. [67] are par- doubtful, as the high dislocation density was identi-
ticularly supportive of this mechanism, as inter-critical fied by the dark appearance of austenite in a bright-
annealing at 620°C was found to lead to higher fractions field micrograph, which is generally indicative of an
of reverted austenite with annealing time up to a max- orientation contrast when using an objective aperture
imum value, after which the fraction decreased again rather than the presence of dislocations [81]. Further,
on further annealing. Even though the phase fraction reverted austenite developed under continued diffusion
1406 F. NIESSEN

xbcc
Cr
xfcc
Cr
xNibcc xfcc
Ni
the scope of this review. A more detailed overview on
20 1.0
the mechanical and corrosion properties of the alloys
0.8 can be obtained from Refs. [2,4,10,16] and [16,91,92],
15
respectively. Nevertheless, a short summary of the
x [wt.%]
0.6
mechanical properties appears necessary to realise the

f [1]
10 f
0.4 significant property changes induced by formation of
5
reverted austenite.
0.2 As an example, Figure 3 shows the mechanical prop-
0 0.0 erties of an Fe–16Cr–5Ni–1Mo (wt-%) soft marten-
600 650 700 A3 sitic stainless steel (EN 1.4418) at room temperature
T [°C] after 4 h soaking at various temperatures, adapted from
Refs. [10,11]. At first the ultimate tensile strength, the
Figure 2. Ni and Cr concentration (x) in austenite (fcc) and
ferrite (bcc) as well as the molar fraction of austenite (fγ , 0.2% yield strength and the hardness increase by heat-
grey area) from an equilibrium calculation of a represen- treating up to 475°C, which is an effect of secondary
tative Fe–13.3Cr–5.4Ni (wt-%) ternary alloy. Partitioning of hardening from precipitation of Mo2 C. Then softening
Ni increases with lower austenite fraction and temperature occurs up to 625°C, mainly due to austenite reversion
(allowed phases: liquid, fcc, bcc; software and thermodynamics and recovery of martensite. A new increase in ulti-
database: Thermo-Calc 2017a – TCFE6 [74]).
mate tensile strength, 0.2% yield strength and hardness
is observed upon heat treatment at 700°C and above,
is known to have low dislocation density [33,68], as originating from transformation of reverted austenite
opposed to reverted austenite formed by a displacive to fresh martensite and precipitation of M23 C6 car-
mechanism [6]. bides. The elongation and impact toughness develop in
Reverted austenite either remains stable upon cool- inverse relation to the ultimate tensile strength, 0.2%
ing, or transforms partially or completely to marten- yield strength and hardness with soaking temperature,
site [30,45,62]. Reverted austenite was reported to even i.e. are enhanced by reverted austenite formation and
have remained stable after sub-zero treatment at boil- recovery of martensite.
ing N2 [2,64,77] and boiling He temperatures [41]. It In the presence of reverted austenite, soft martensitic
is critically remarked that martensite formation in lath and supermartensitic stainless steels show remarkable
martensite, especially at sub-zero Celsius temperatures,
is time-dependent, i.e. thermally activated, and kineti-
UTS 0.2% YS Elongation f
cally suppressed at very low temperature. Transforma- (a) 1200 30
tion generally occurs in the temperature range −150°C
UTS and 0.2% YS [MPa]

1100 25

f and elongation [%]

to room temperature, which means that in the refer-
1000 20
enced cases [2,41,64,77] martensite formation may be
fully suppressed if cooling to, or heating from, boiling 900 15
N2 temperature is performed sufficiently fast to sup- 800 10
press the thermally activated phenomena that control
700 5
the rate of the austenite-to-martensite transformation
in steel [82]. In a single case, it was claimed that the frac- 600 0
HC 400 500 600 700 800
tion of austenite increases by applying sub-zero treat- (b) 400 300
ment [57], which is in obvious conflict with all above
375
Impact toughness [J]

referenced observations. 250

Hardness [HV10]

Reverted austenite may also transform to marten- 350

site by strain-induced transformation and in this way 325 200
enhances the plastic regime [45,64,70,83]. This mecha- 300
nism known as transformation-induced plasticity, has 150
275 Hardness
not been treated in depth in low-carbon martensitic Impact Toughness
stainless steels, but advanced in situ experiments and 250 100
HC 400 500 600 700 800
modelling of strain-induced martensite formation have
T [°C]
been performed in other variations of steels [84–90].
Figure 3. Fraction austenite and mechanical properties of a
16Cr–5Ni–1Mo stainless steel at room temperature in the hard-
Effect of reverted austenite on microstructure ened condition (HC) and after 4 h soaking at various temper-
properties atures: (a) ultimate tensile strength (UTS), 0.2% yield strength
(YS), elongation and fraction austenite (fγ ); (b) hardness and
The effect of reverted austenite on the microstructure Charpy V impact toughness; the lines are spline functions of
properties of low-carbon martensitic stainless steels is measured data (symbols) and do not represent physical values.
a broad topic and a proper treatment would exceed Source: adapted from Dawood et al. [10,11].
 MATERIALS SCIENCE AND TECHNOLOGY 1407

impact toughness also at sub-zero Celsius temperature (b and d) compared with the respective concentra-
( > 100 J at −80°C [16]). Solheim et al. [93] showed tions from computed phase equilibria. Short proxim-
that reverted austenite increased the solubility of hydro- ity of data points to the diagonal line indicates good
gen in supermartensitic stainless steel samples dramat- agreement of the measured concentrations with val-
ically, and that the ductility of such samples was greatly ues reflecting thermodynamic equilibrium for the alloy
reduced, suggesting that reverted austenite plays an under consideration. Even though different levels of
important role in hydrogen embrittlement. agreement are obtained for different components and
phases, experimental data generally seem to agree with
calculations from thermodynamics modelling. This
Critical assessment of compositional data from
supports that austenite reversion occurs mainly by a
the literature
diffusional mechanism, as potential large strain ener-
All considered references on experimentally deter- gies from a displacive transformation are not reflected
mined Cr and Ni contents of partitioning in reverted in the applied thermodynamics model. In tempered
austenite and tempered martensite are sorted in order martensite, apart from few outliers, both the reported
of ascending annealing temperature in Table 2. More- Ni and Cr concentrations are in reasonable agree-
over, the sample preparation methods and measure- ment with predictions from thermodynamics mod-
ment methods, energy-dispersive X-ray spectroscopy elling (Figure 4(a,c), respectively). In austenite, apart
(EDS) in (scanning) transmission electron microscopy from a single measurement, all reported Cr concen-
([S]TEM) and atom probe tomography (APT), are indi- trations exceed the concentrations for thermodynamic
cated together with the respective measurement and equilibrium (Figure 4(d)). The reported Ni concentra-
sample preparation methods. The reported annealing tions of reverted austenite show poor agreement with
parameters and alloy compositions were used to deter- calculated thermodynamic equilibrium values (Figure
mine the equilibrium concentrations of Cr and Ni in 4(b)). In the latter case, data from APT appeared to
austenite and ferrite by computing phase equilibria agree more convincingly with data from thermody-
(Thermo-Calc 2017a [94] with TCFE6 database [74]). namic equilibria compared to data from (S)TEM-EDS.
For this purpose, the alloying elements Cr, Ni, Mo, Mn, Within significant scatter a trend is discernible: Pre-
Si, Cu, C and N and the phases fcc, bcc, cementite, M6 C, dicted concentrations of Ni in austenite by thermo-
M7 C3 and M23 C6 were taken into account, representing dynamics modelling are approx. 2 wt-% higher than
the most common elements and phases reported in the measured concentrations.
literature [95]. As the prediction of the Ni concentration in austen-
Figure 4 gives reported Ni and Cr concentrations in ite is most relevant for the stability of austenite,
tempered martensite (a and c) and reverted austenite the origin of the discrepancy between experimental

APT (a) 25 (b) 25

Schnitzer Radis Nöhrer et al. 2010 A
Schnitzer Radis Nöhrer et al. 2010 B
Song Li Rong et al. 2011
Song Ping Yin et al. 2010
20 20
Escobar Poplawsky Faria et al. 2017 B
TEM-EDS
15 15
xNi,bcc

Yuan Gong Sun et al. 2016 A

xNi,fcc
meas
meas

Lee Shin Leem et al. 2003 A

Schnitzer Radis Nöhrer et al. 2010 C
Schnitzer Radis Nöhrer et al. 2010 D 10 10
Song Li Rong et al. 2010
Bilmes Solari Llorente 2001
Ye Li Jiang et al. 2012 5 5
Yuan Gong Sun et al. 2016 B
Song Li Rong et al. 2014
Zhang Wang Li et al. 2015 0 0
Wang Lu Li et al. 2008 0 5 10 15 20 25 0 5 10 15 20 25
Lee Shin Leem et al. 2003 B
Ni,bcc Ni,fcc
Liu Yang Zhang 2013 x equ
x equ
Escobar Poplawsky Faria et al. 2017 A
Wang Xiao Lu et al. 2013 (c) 25 (d) 25
Nakada Tsuchiyama Takaki et al. 2007
Yuan Gong Sun et al. 2016 C
Jiang Zhao Ye et al. 2013 20 20
Jiang Ye Li et al. 2014
Niessen Grumsen Hald et al. 2017
Lee Shin Leem et al. 2003 C 15 15
xCr,bcc

xCr,fcc
meas
meas

De Sanctis Lovicu Valentini et al. 2015

10 10

5 5

0 0
0 5 10 15 20 25 0 5 10 15 20 25
Cr,bcc Cr,fcc
xequ xequ

Figure 4. Comparison of the Ni and Cr concentrations in tempered martensite (a and c), and reverted austenite (b and d) from
EDS and APT analysis, xmeas , with data from thermodynamics modelling, xequ , under input of the respective alloy compositions and
annealing temperatures; the legend is sorted in order of ascending annealing temperature.
1408 F. NIESSEN

(a) 80 (b) 140

700 °C 120
60 100

f [vol.%]
80

t 0.95 [h]
675 °C
40 60

650 °C 40
20
20
625 °C
600 °C 0
0
-3 -2 -1 0 1 2 600 620 640 660 680 700
10 10 10 10 10 10
t [h] T [°C]

Figure 5. (a) Kinetics of austenite reversion from kinetics modelling of diffusion of an Fe–13.3Cr–5.4Ni (wt-%) steel (cf. Figure 2).
The broken lines indicate the austenite fractions in equilibrium at the respective temperatures and the open symbols indicate t0.95 ,
the time at which 95% of the equilibrium phase fraction is formed; (b) fit of t0.95 as a function of temperature by an exponen-
tial decay function, which is the foundation to assess equilibration of the evaluated heat treatments by the equilibration factor
kequ (T, t).

and predicted Ni content should be investigated. The Comparing the time t0.95 at reported temperatures
martensite to austenite phase transformation can be with the respective reported annealing times from the
(partly) interface-controlled or local equilibrium may literature, a factor kequ = t/t0.95 is obtained to qualita-
be affected by factors as residual stresses, interfacial seg- tively estimate the amount of equilibration for each data
regation and precipitation of carbides. Then the Cr and point (Table 2).
Ni concentrations can vary with time before the con- Figure 6 shows the difference of the measured con-
dition of (local) equilibrium is obtained. In order to centrations with respect to concentrations from ther-
investigate, whether the discrepancy between Ni con- modynamic equilibrium, ε, in austenite as a function
centrations from phase equilibria and measurements of the equilibration factor kequ . Clearly, the difference
decreases with approaching equilibration, a criterion between the measured and calculated Ni concentration
was established to filter the literature data accordingly decreases with increasing equilibration (higher kequ ),
(Figure 5). For this purpose, the kinetics of form- while a similar trend, if present at all, is less obvious
ing reverted austenite from a 200 nm wide marten- for Cr. Figure 7 shows the data from Figure 4 after fil-
site lath of an Fe–13.3Cr–5.4Ni alloy (cf. Figure 2) tering with the equilibration criterion kequ > 0.1. As
by bulk diffusion at temperatures between 600°C and expected from Figure 6(b), the scatter of the Ni con-
700°C were assessed in steps of 25°C by kinetics mod- centration in reverted austenite is reduced, such that
elling of diffusion with DICTRA [96] (see Ref. [27] for the experimental data and data from thermodynam-
further details on the kinetics model). Kinetics mod- ics modelling show reasonable agreement (Figure 7(b)).
elling assumes purely diffusion controlled martensite- Filtering of the data on the Cr concentration in reverted
to-austenite transformation and local equilibrium at austenite reduces the scatter to a small extent and fur-
the martensite/austenite interface [97]. While these ther confirms that the Cr concentration in reverted
assumptions, as discussed above, may not be entirely austenite is systematically measured to be 1.0–3.5 wt-%
justified in the present case, kinetics modelling should
reliably reflect the general kinetics of austenite rever-
sion in low-carbon martensitic stainless steels at differ-
ent temperatures to establish a qualitative criterion. (a) (b)
The transformation kinetics in Figure 5(a) reveals (xCr,fcc)=|(xCr,fcc
sim
-xCr,fcc
meas
)/xCr,fcc
sim
| (xNi,fcc)=|(xNi,fcc
sim
-xNi,fcc
meas
)/xNi,fcc
sim
|
60 60
that the time to reach 95% of the equilibrium fraction
(xCr,fcc) [%]

(xNi,fcc) [%]

(t0.95 ) increases from 0.2 h at 700°C to > 100 h at 600°C 40 40

(see open symbols in Figure 5(a) and the temperature
dependence in Figure 5(b)). The dependence of t0.95 on 20 20
the annealing temperature follows an exponential decay
function (Figure 5(b)): 0 0
0.0 0.5 1.0 0.0 0.5 1.0

  kequ [1] kequ [1]

T0 − T
t0.95 = A exp + c, (1) Figure 6. Difference of measured concentrations with respect
b
to concentrations from thermodynamic equilibrium, ε, as a
function of the equilibration factor kequ (Figure 5(b)) of the (a)
with A = 126 h, T 0 = 600°C, b = 19.1°C and c = Cr and (b) Ni concentration in reverted austenite. The symbols
−0.5 h. correspond to the legend in Figure 4.
 MATERIALS SCIENCE AND TECHNOLOGY 1409

Figure 7. Comparison of the Ni and Cr concentrations in tempered martensite (a and c), and reverted austenite (b and d) with data
from thermodynamics modelling from Figure 4 for data with an equilibration factor of kequ > 0.1.

higher than predicted by equilibrium thermodynamics Stress plays a fundamental role in martensite for-
(Figure 7(d)). mation. The strict hierarchy of lath martensite fol-
lows the minimisation of the total shape strain [98].
It therefore appears unavoidable to consider the role
Discussion of stress in the reverted transformation to austenite.
The previously mentioned austenite memory effect,
Early stage of reverted austenite formation close the strong tendency of reverted austenite to form in
to A1 an identical crystallographic orientation to the prior
Microstructure characterisation of the initial stage of austenite grain, gives important insight into the role
reverted austenite formation is difficult to conduct and of stress in the early stage of reverted austenite for-
of more fundamental interest. So far, only limited stud- mation. Nakada et al. [71] suggested that the reverted
ies have been conducted on the early stage of austen- austenite variants are theoretically limited to two vari-
ite formation, while the mechanisms during growth ants within a martensite packet: the prior austenite and
in different temperature regimes are fairly well estab- a twin-related orientation variant. The variant restric-
lished. Nevertheless, by studying related alloy systems tion originates from crystal symmetry and consider-
and carefully interpreting related mechanisms, such as ations of interfacial energies and was validated with
martensite formation and austenite memory, an under- experiments on a supermartensitic stainless steel with
standing of the early stage of austenite formation can be no retained austenite. Indeed, only the two predicted
acquired from the existing literature. variants formed. Interestingly, the twin-related variant
Raabe et al. [8] observed segregation of Mn up to 24 made up less than 5% area fraction of reverted austen-
at.-% at lath boundaries before formation of reverted ite and could be increased to approx. 40% by applying
austenite on tempering an Fe–9Mn (at.-%) marag- a uniaxial tensile stress of 100 MPa during inter-critical
ing alloy at 450°C. The experimentally observed grain annealing. As a consequence, Nakada et al. suggested
boundary enrichment factors, which relate the bulk that also residual stress affects austenite memory in that
concentration of an element to the concentration at the it favours the prior austenite orientation [71].
grain boundary, were found similar for Mn and Ni [8]. Miyamoto et al. [99] provided experimental evi-
Thus, Ni is likely to segregate to lath boundaries in low- dence of the accommodation of transformation strains
carbon martensitic stainless steel during heating which from lath martensite formation in austenite by mea-
would locally increase the driving force for austenite surement of lattice strains (rotations). Sandvik et al. [34]
formation and thereby aid austenite nucleation. So far, suggested plastic accommodation of transformation
no experimental evidence of such mechanisms has been strains in martensite laths by characterising the dislo-
reported that can corroborate this hypothesis. cation substructure. Both independently reported that
1410 F. NIESSEN

Figure 8. Footage from the literature indicating thickening of lath martensite to primarily one side; (a) misorientation map of austen-
ite from electron backscatter diffraction (EBSD) after growth of lath martensite (black) in an Fe–20Ni–5Mn (wt-%) alloy indicating
strain accommodation of the transformation strain in austenite in primarily one direction of the lath (Ref. [99], permission was auto-
matically granted according to the STM signatory guidelines.); (b) example of a martensite lath in an Fe–20Ni–6Mn–0.01 (wt-%) alloy,
showing a relatively straight (left side) and irregular (right side) martensite/austenite interface (Ref. [34], RightsLink license number
4294080199168).

the martensite/austenite interface is relatively straight At first, it is noted that the calculated thermody-
on one side and irregular on the other side of a lath namic equilibria originate from a modelling method,
(Figure 8), which suggests thickening of martensite which inherently suffers from uncertainties originating
laths mainly towards the irregular side of the interface. from databases [100] and relies on extrapolation from
Consequently, transformation strains should accumu- constituent subsystems to multi-component alloy sys-
late with movement of the martensite/austenite inter- tems [101,102]. In the present case, further uncertain-
face and manifest as build-up of residual stress between ties are expected from treating all alloys with the same
adjacent laths. list of phases and components, where in some cases not
When annealing lath martensite, austenite forma- all components were reported or included. Considering
tion in a crystallographically reverse sense with respect these uncertainties, the observed amount of system-
to martensite formation, i.e. an inverse interface move- atic deviation (approx. 2 wt-% for Cr in martensite)
ment and orientation relationship, should lead to relax- appears too excessive to be introduced by the CAL-
ation of these stresses. Residual stress between adjacent PHAD method or the database. The analysed Fe–Cr–Ni
laths may thus act as a mechanical driving force that, in alloy system is amongst the best-described systems in
addition to the chemical driving force, promotes forma- thermodynamics modelling and has proven to pro-
tion and diffusional growth of austenite in a distinct ori- vide excellent predictions in previous cases [103–105].
entation and growth direction. The observed formation Moreover, it appears difficult to explain the more ran-
of austenite in only one lath of at lath boundary below dom deviation of Ni in austenite with potential uncer-
575°C is consistent with this hypothesis (Figure 1(a)) tainties in the CALPHAD method. Therefore, other
[1,55,60,66]. At sufficiently high temperature, recovery reasons for the observed deviation need consideration.
of martensite partially relaxes the residual stress and the According to thermodynamic equilibrium, Cr is
chemical driving force for austenite formation is higher. in all analysed cases supposed to be rejected from
Consequently, the contribution of the mechanical driv- austenite and dissolved in carbides and martensite dur-
ing force is greatly reduced, which gives a consistent ing inter-critical annealing. The difference in concen-
explanation for austenite formation in both laths of a tration between austenite and martensite should in
lath boundary above 575°C (Figure 1(b)) [55,58,60]. average amount to 2 wt-% (Table 2). From 13 experi-
Further evidence from experiments and modelling is mental datasets on partitioning of Cr in austenite and
required to corroborate this hypothesis. martensite, only 3 datasets qualitatively confirm such
partitioning [65,72,73], while 6 datasets show no parti-
tioning within 0.5 wt-% of accuracy [1,33,62,69] and 4
Discrepancy of data on solute partitioning from
datasets (all originating from the same research group)
equilibrium calculations and literature
even show inverse partitioning, i.e. Cr-enrichment in
While reasonable agreement of concentrations from lit- austenite [41,58,63,67]. Even though these inconsistent
erature data with concentrations from thermodynamic observations are difficult to interpret, it appears likely
equilibrium was obtained for Cr and Ni in martensite that growth of M23 C6 carbides affects the local Cr con-
(ferrite), in austenite systematic and more random dis- tents of austenite and martensite. M23 C6 carbides share
crepancies were found for Cr and Ni, respectively. In the a coherent interface with reverted austenite, which is
following, possible reasons will be discussed. anticipated to facilitate the diffusion flux of Cr and Ni
 MATERIALS SCIENCE AND TECHNOLOGY 1411

during diffusion accompanied growth of both phases, from EDS generally less trustworthy. The lower exper-
as also suggested in Ref. [67]. imentally determined Ni concentrations, especially at
In contrast to the Cr concentration, the Ni concen- lower temperature, may therefore to a certain extent
tration in austenite deviated more randomly from ther- originate from a higher experimental error of (S)TEM-
modynamic equilibrium (Figure 4(b)). Within large EDS measurements.
scatter, Ni was in average predicted with 2 wt-%
higher concentration than characterised in literature.
It appears that two major effects cause the discrep- Conclusions
ancy: insufficient equilibration of the microstructure
The mechanism of austenite reversion in low-carbon
and experimental error.
martensitic stainless steels was critically reviewed by
It was found that the Ni concentration in austen-
collating literature on nucleation and growth, stabil-
ite can be predicted with reasonable accuracy by phase
ity against martensite formation and the effect on
equilibria when the microstructure approaches equili-
microstructure properties of reverted austenite. Discus-
bration (higher kequ in Figure 6(b)). Escobar et al. [69]
sion of the morphology of austenite films close to A1
reported that, even when the equilibrium phase frac-
in the light of austenite memory and the mechanism
tion was obtained after annealing for 2.5 h at 625°C,
of martensite formation led to suggest that residual
individual austenite lamellae revealed approx. ±2 wt-
stresses from the martensite microstructure aid nucle-
% difference in Ni concentration. Further, gradients of
ation of reverted austenite close to A1 . It was further
Ni concentration within austenite lamellae in the range
investigated whether literature data on the Ni and Cr
of approx. 4 wt-% from interface to bulk were mea-
concentrations in austenite and martensite after inter-
sured with APT. Apparently, the initial composition of
critical annealing comply with thermodynamic equilib-
reverted austenite is far from equilibrium.
rium from thermodynamics modelling. In martensite,
It is noted that martensite itself must be consid-
measured Cr and Ni concentrations matched predic-
ered as metastable ferrite, as it deviates from ferrite
tions from thermodynamics modelling with reasonable
in thermodynamic equilibrium. Nucleation of reverted
accuracy. Systematic excess of Cr in austenite by approx.
austenite in equilibrium with martensite (metastable
2 wt-% relative to predictions was suspected to origi-
ferrite) may thus occur with different phase compo-
nate from growth of M23 C6 with a coherent interface to
sitions as predicted for austenite and ferrite in ther-
austenite. Within large scatter, measured values of Ni
modynamic equilibrium. It is suggested that the phase
in austenite were in average 2 wt-% below predictions
fractions and compositions of austenite and marten-
from thermodynamics modelling. The scatter reduced
site during continued annealing evolve towards global
dramatically when only microstructures with advanced
equilibrium, which appears to be a slow and complex
equilibration were considered. Further, APT data
process.
matched predictions more convincingly than data from
The majority of insufficiently equilibrated microst-
(S)TEM-EDS, indicating better experimental accu-
ructures (kequ < 0.1) were obtained from annealing
racy for determining concentrations in the partitioned
below 650°C. As partitioning of Ni increases with lower
microstructure.
temperature (Figure 2), sharp compositional measure-
ments over thin austenite films (40–150 nm width)
and tempered martensite become more challenging.
Acknowledgements
Figure 4 reveals that measurements from APT matched
the predictions from thermodynamic equilibrium more The author is grateful to Professor Marcel A.J. Somers and
Professor John Hald, Department of Mechanical Engineering
convincingly than measurements from (S)TEM-EDS. at the Technical University of Denmark, for fruitful discus-
As the spatial resolution (near atomic) and composi- sions and proofreading of the manuscript.
tional resolution (few ppm) of APT are by far superior
to that of standard (S)TEM-EDS [106], this observa-
tion appears reasonable. (S)TEM-EDS faces the inher- Disclosure statement
ent issue of insufficient counting rate when samples are No potential conflict of interest was reported by the author.
too thin, leading to insufficient compositional accuracy
when applying low counting times and contamination
and beam drift when applying high counting times Funding
[107]. Increasing the sample thickness, on the other
The Danish Underground Consortium is gratefully acknowl-
hand, reduces the spatial resolution and complicates edged for financial support to the Danish Hydrocarbon
measurement of a sharp austenite/martensite interface. Research and Technology Centre.
Reliable measurements of fine-grained and partitioned
microstructures are certainly possible (good agreement
with data from APT was, for instance, obtained in Refs. ORCID
[1,69]), but several potential error sources render data Frank Niessen http://orcid.org/0000-0001-5849-710X
1412 F. NIESSEN

References [19] Grounes M. Swedish work on brittle-fracture prob-

lems in nuclear reactor pressure vessels. Sweden: AB
[1] Schnitzer R, Radis R, Nöhrer M, et al. Reverted austen-
Atomenergi; 1966. p. 1–34.
ite in PH 13-8 Mo maraging steels. Mater Chem Phys.
[20] Grounes M, Lindhagen P. Irradiation effects in pres-
2010;122:138–145.
sure vessel materials for steam-cooled fast reactors.
[2] Niederau HJ. Eigenschaften und Anwendungsmögli-
Irradiation Effects in Structural Alloys for Thermal
chkeiten der Guss- und Schmiedelegierung (G)X-5
and Fast Reactors. ASTM STP 457; 1969, p. 156–179.
CrNiMo 16 5 (Werkstoff-Nr. 1.4405). Zeitschrift des
[21] Servant C, Gherbi EH, Cizeron G. TEM investigation
Vereins Dtsch. Ingenieure für Maschinenbau und Met
of the tempering behaviour of the maraging PH 17.4
Eig. 1982;21:801–808.
Mo stainless steel. J Mater Sci. 1987;22:2297–2304.
[3] Grounes M, Rao S. New alloy steels for nuclear reac-
[22] Morley JI. Some recent developments in stainless
tor pressure vessels and vessel internals. Trans ASM.
steels. Iron Steel. 1955;28:183–188.
1969;62:902–914.
[23] Kondo K, Ogawa K, Amaya A, et al. Development
[4] Brezina P, Sonderegger B. Wärmebehandlung, Gefüge
of weldable super 13Cr martensitic stainless steel for
und Eigenschaften des korrosionsträgen martensi-
flowline. Proceedings of Twelfth International Off-
taushärtbaren Stahles X 5 CrNiMoCuNb 14 5 (14-5
shore and Polar Engineering Conference; Kyushu,
PU). Härterei-Technische Mitteilungen. 1978;33:1–12.
Japan; 2002, p. 303–309.
[5] Tomimura K, Takaki S, Tanimoto S, et al. Optimal
[24] Van der Winden H, Toussaint P, Coudreuse L. Past,
chemical composition in Fe-Cr-Ni alloys for ultra
present and future of weldable supermartensitic alloys.
grain refining by reversion from deformation induced
Supermartensitic Stainless Steels 2002, Brussels; 2002,
martensite. ISIJ Int. 1991;31:721–727.
p. 9–13.
[6] Tomimura K, Takaki S, Tokunaga Y. Reversion mecha-
[25] Kvaale PE, Stein O. Experience with supermartensitic
nism from deformation induced martensite to austen-
stainless steels in flowline applications. Stainless Steel
ite in metastable austenitic stainless steels. ISIJ Int.
World Conference; Brussels, Belgium; 1999, p. 19–26.
1991;31:1431–1437.
[26] Ma X, Wang L, Subramanian SV, et al. Studies on
[7] Kuzmina M, Ponge D, Raabe D. Grain boundary
Nb microalloying of 13Cr super martensitic stainless
segregation engineering and austenite reversion turn
steel. Metall Mater Trans A Phys Metall Mater Sci.
embrittlement into toughness: example of a 9 wt-%
2012;43:4475–4486.
medium Mn steel. Acta Mater. 2015;86:182–192.
[27] Niessen F, Villa M, Hald J, et al. Kinetics analysis of
[8] Raabe D, Sandlöbes S, Millán J, et al. Segregation
two-stage austenitization in supermartensitic stainless
engineering enables nanoscale martensite to austenite
steel. Mater Des. 2017;116:8–15.
phase transformation at grain boundaries: A pathway
[28] Villa M, Pantleon K, Reich M, et al. Kinetics of anoma-
to ductile martensite. Acta Mater. 2013;61:6132–6152.
lous multi-step formation of lath martensite in steel.
[9] Irvine KJ, Crowe DJ, Pickering FB. The physical met-
Acta Mater. 2014;80:468–477.
allurgy of 12% chromium steels. J Iron Steel Inst.
[29] Hsiao CN, Chiou CS, Yang JR. Aging reactions
1960;193:386–406.
in a 17-4 PH stainless steel. Mater Chem Phys.
[10] Al Dawood M, El Mahallawi IS, Abd El Azim ME,
2002;74:134–142.
et al. Thermal aging of 16Cr – 5Ni – 1Mo stainless steel
[30] Niessen F, Villa M, Apel D, et al. In situ techniques
part 1 – microstructural analysis. Mater Sci Technol.
for the investigation of the kinetics of austenitization
2004;20:363–369.
of supermartensitic stainless steel. Mater Sci Forum.
[11] Al Dawood M, El Mahallawi IS, Abd El Azim ME, et al.
2017;879:1381–1386.
Thermal aging of 16Cr – 5Ni – 1Mo stainless steel part
[31] Bojack A, Zhao L, Morris PF, et al. In-situ deter-
2 – mechanical property characterisation. Mater Sci
mination of austenite and martensite formation in
Technol. 2004;20:370–374.
13Cr6Ni2Mo supermartensitic stainless steel. Mater
[12] Kondo K, Ueda M, Ogawa K, et al. Alloy design of
Charact. 2012;71:77–86.
super 13Cr martensitic stainless steel. supermarten-
[32] Rodrigues CAD, Lorenzo PLD, Sokolowski A, et al.
sitic stainl. Steels. 1999;1999:11–18.
Titanium and molybdenum content in supermarten-
[13] Sagara M, Yamada K, Taniyama A, et al. Development
sitic stainless steel. Mater Sci Eng A. 2007;460-461:
for material and application technique of stainless steel
149–152.
line pipes. Nippon Steel & Sumitomometal Technical
[33] Sanctis MD, Lovicu G, Valentini R, et al. Microstruc-
Report; 2015, p. 59–64.
tural features affecting tempering behavior of 16Cr-
[14] Tvrdy M, Vodarek V, Roznovska G, et al. Production,
5Ni supermartensitic steel. Metall Mater Trans A Phys
development and industrial application of 12Cr-6Ni-
Metall Mater Sci. 2015;46:1878–1887.
2.5Mo steels. Supermartensitic Stainless Steels 2002,
[34] Sandvik BPJ, Wayman CM. Characteristics of lath
Brussels; 2002, p. 29–33.
martensite: part I. Crystallographic and substructural
[15] Toussaint P, Dufrane J-J. Advances in the making and
features. Metall Trans A. 1983;14:809–822.
base material properties of supermartensitic stainless
[35] Kurdjumow G, Sachs G. Über den Mechanismus der
steels (SMSS). Supermartensitic Stainless Steels2002,
Stahlhärtung. Zeitschrift für Phys. 1930;64:325–343.
Brussels, KCI Publishing; 2002, p. 23–27.
[36] Bos C, Sietsma J, Thijsse BJ. Molecular dynamics sim-
[16] Marshall AW, Farrar JCM. Welding of ferritic and
ulation of interface dynamics during the fcc-bcc trans-
martensitic 11–14% Cr steels. Weld World. 2001;45:
formation of a martensitic nature. Phys Rev B Condens
19–42.
Matter Mater Phys. 2006;73:1–7.
[17] Karlsson L, Rigdal S, Bruins W, et al. Development of
[37] Maresca F, Curtin WA. The austenite/lath martensite
matching composition supermartensitic stainless steel
interface in steels: structure, athermal motion, and in-
welding consumables. Svetsaren. 1999;54:3–7.
situ transformation strain revealed by simulation and
[18] Baggström G. New steel for turbine runners. Water
theory. Acta Mater. 2017;134:302–323.
Power. 1964;16:516–521.
 MATERIALS SCIENCE AND TECHNOLOGY 1413

[38] Bhadeshia HKDH. Worked examples in the geometry [55] Lee Y-K, Shin H-C, Leem D-S, et al. Reverse trans-
of crystals. 2nd ed. Brookfield: The Institute of Metals; formation mechanism of martensite to austenite and
2001. amount of retained austenite after reverse transforma-
[39] Hall MG, Aaronson HI, Kinsma KR. The structure of tion in Fe-3Si-13Cr-7Ni (wt-%) martensitic stainless
nearly coherent fcc: bcc boundaries in a CuCr alloy. steel. Mater Sci Technol. 2003;19:393–398.
Surf Sci. 1972;31:257–274. [56] Nakada N, Tsuchiyama T, Takaki S, et al. Temperature
[40] Niessen F, Danoix F, Hald J and Somers MAJ. Struc- dependence of austenite nucleation behavior from lath
tural analysis in atom probe tomography – application martensite. ISIJ Int. 2011;51:299–304.
to tempering of supermartensitic stainless steel. 7th [57] Liu L, Yang ZG, Zhang C, et al. An in situ study on
European Atom Probe Work, 2017, Gullmarsstrand, austenite memory and austenitic spontaneous recrys-
Sweden; 2017. tallization of a martensitic steel. Mater Sci Eng A.
[41] Song Y, Ping DH, Yin FX, et al. Microstructural 2010;527:7204–7209.
evolution and low temperature impact toughness of [58] Song YY, Li XY, Rong LJ, et al. Formation of the
a Fe–13%Cr–4%Ni–Mo martensitic stainless steel. reversed austenite during intercritical tempering in a
Mater Sci Eng A. 2010;527:614–618. Fe-13%Cr-4%Ni-Mo martensitic stainless steel. Mater
[42] Villa M, Niessen F, Somers MAJ. In situ investigation Lett. 2010;64:1411–1414.
of the evolution of lattice strain and stresses in austen- [59] Liu L, Yang Z-G, Zhang C. Effect of retained austenite
ite and martensite during quenching and tempering of on austenite memory of a 13% Cr–5% Ni martensitic
steel. Metall Mater Trans A. 2017;49:1–13. steel. J Alloys Compd. 2013;577:S654–S660.
[43] Maki T. Morphology and substructure of marten- [60] Yuan WH, Gong XH, Sun YQ, et al. Microstructure
site in steels. In: Pereloma E, Edmonds DV, editors. evolution and precipitation behavior of 0Cr16Ni5Mo
Phase Transformations in Steels, Vol. 2. Diffusionless martensitic stainless steel during tempering process. J
Transformations, High Strength Steels, Modelling and Iron Steel Res Int. 2016;23:401–408.
Advanced Analytical Techniques. 1st ed. Cambridge: [61] Morito S, Tanaka H, Konishi R, et al. The morphology
Woodhead Publishing; 2012. p. 34–58. and crystallography of lath martensite in Fe-C alloys.
[44] Christien F, Telling MTF, Knight KS. Neutron diffrac- Acta Mater. 2003;51:1789–1799.
tion in situ monitoring of the dislocation density dur- [62] Niessen F, Grumsen FB, Hald J, et al. Formation and
ing martensitic transformation in a stainless steel. Scr stabilization of reversed austenite in supermartensitic
Mater. 2013;68:506–509. stainless steel. Proceedings of 24th IFHTSE Congress;
[45] Wiessner M, Gamsjäger E, van der Zwaag S, et al. Effect Nice, France; 2017, p. 138–145.
of reverted austenite on tensile and impact strength in [63] Song Y, Li X, Rong L, et al. The influence of tem-
a martensitic stainless steel − an in-situ X-ray diffrac- pering temperature on the reversed austenite forma-
tion study. Mater Sci Eng A. 2017;682:117–125. tion and tensile properties in Fe-13%Cr-4%Ni-Mo low
[46] Speich GR. Tempering of Low-carbon martensite. carbon martensite stainless steels. Mater Sci Eng A.
Trans Metall Soc AIME. 1969;245:2553–2564. 2011;528:4075–4079.
[47] Slunder CJ, Hoenie AF, Hall AM. Thermal and [64] Bilmes PD, Solari M, Llorente CL. Characteristics
mechanical treatment for precipitation-hardening stain- and effects of austenite resulting from tempering
less steels. Washington (DC): NASA; 1967. of 13Cr–NiMo martensitic steel weld metals. Mater
[48] Ludwigson DC, Hall AM. The physical metallurgy Charact. 2001;46:285–296.
of precipitation-hardenable stainless steels. DMIC [65] Ye D, Li J, Jiang W, et al. Effect of Cu addi-
Report 111, Battelle Memorial Institute, Columbus, tion on microstructure and mechanical properties of
OH; 1959. 15%Cr super martensitic stainless steel. Mater Des.
[49] Yeli G, Auger MA, Smith GDW, et al. Atom probe 2012;41:16–22.
tomography (APT) characterization of the sequence [66] Song YY, Li XY, Rong LJ, et al. Reversed austenite in
of phase nucleation in a 17-4PH steel. Proceedings of 0Cr13Ni4Mo martensitic stainless steels. Mater Chem
International Conference on Solid-Solid Phase Trans- Phys. 2014;143:728–734.
formations in Inorganic MaterialsInt Conf Solid-Solid [67] Zhang S, Wang P, Li D, et al. Investigation of the
Phase Transform Inorg Mater; 2015, p. 167–168. evolution of retained austenite in Fe-13%Cr-4%Ni
[50] Yeli G, Auger MA, Wilford K, et al. Sequential nucle- martensitic stainless steel during intercritical temper-
ation of phases in a 17-4PH steel: microstructural char- ing. Mater Des. 2015;84:385–394.
acterisation and mechanical properties. Acta Mater. [68] Wang P, Lu S, Li D, et al. Investigation of phase tran-
2017;125:38–49. formation of low carbon martensitic stainless steel
[51] Shi ZM, Gong W, Tomota Y, et al. Study of temper- ZG06Cr13Ni4Mo in tempering process with low heat-
ing behavior of lath martensite using in situ neutron ing rate. Acta Metall Sin. 2008;44:681–685.
diffraction. Mater Charact. 2015;107:29–32. [69] Escobar JD, Poplawsky JD, Faria GA, et al. Composi-
[52] Bojack A, Zhao L, Morris PF, et al. Austenite for- tional analysis on the reverted austenite and tempered
mation from martensite in a 13Cr6Ni2Mo super- martensite in a Ti-stabilized supermartensitic stainless
martensitic stainless steel. Metall Mater Trans A. steel: segregation, partitioning and carbide precipita-
2016;47:1996–2009. tion. Mater Des. 2018;140:95–105.
[53] Ma XP, Wang LJ, Qin B, et al. Effect of N on microstruc- [70] Wang P, Xiao N, Lu S, et al. Investigation of
ture and mechanical properties of 16Cr5Ni1Mo marten- the mechanical stability of reversed austenite in
sitic stainless steel. Mater Des. 2012;34:74–81. 13%Cr-4%Ni martensitic stainless steel during the
[54] Leem D-S, Lee Y-D, Jun J-H, et al. Amount of uniaxial tensile test. Mater Sci Eng A. 2013;586:
retained austenite at room temperature after reverse 292–300.
transformation of martensite to austenite in an [71] Nakada N, Tsuchiyama T, Takaki S, et al. Variant selec-
Fe–13%Cr–7%Ni–3%Si martensitic stainless steel. Scr tion of reversed austenite in lath martensite. ISIJ Int.
Mater. 2001;45:767–772. 2007;47:1527–1532.
1414 F. NIESSEN

[72] Jiang W, Zhao K, Ye D, et al. Effect of heat treatment on [89] Latypov MI, Shin S, De Cooman BC, et al. Microme-
reversed austenite in Cr15 super martensitic stainless chanical finite element analysis of strain partitioning in
steel. J Iron Steel Res Int. 2013;20:61–65. multiphase medium manganese TWIP + TRIP steel.
[73] Jiang W, Ye D, Li J, et al. Reverse transformation mech- Acta Mater. 2016;108:219–228.
anism of martensite to austenite in 00Cr15Ni7Mo2 [90] Liu H, Du L-X, Hu J, et al. Interplay between
WCu2 super martensitic stainless steel. Steel Res Int. reversed austenite and plastic deformation in a directly
2014;85:1150–1157. quenched and intercritically annealed 0.04C-5Mn
[74] Thermo-Calc Software. TCFE6 Steels/Fe-alloys low-Al steel. J Alloys Compd. 2017;695:2072–2082.
database version 6.2, 2009. [91] Anselmo N, May JE, Mariano NA, et al. Corrosion
[75] Escobar JD, Faria G, Wu L, et al. Austenite rever- behavior of supermartensitic stainless steel in aerated
sion kinetics and stability during tempering of a Ti- and CO2 -saturated synthetic seawater. Mater Sci Eng
stabilized supermartensitic stainless steel: correlative A. 2006;428:73–79.
in situ synchrotron x-ray diffraction and dilatometry. [92] Woollin P. Understanding and avoiding intergranular
Acta Mater. 2017;138:92–99. stress corrosion cracking of welded supermartensitic
[76] Song Y, Li X, Rong L, et al. Anomalous phase trans- steel. NACE Corrosion Conference Expo (Corrosion
formation from martensite to austenite in Fe-13%Cr- 2007); 2007, p. 5379–5394.
4%Ni-Mo martensitic stainless steel. J Mater Sci Tech- [93] Solheim KG, Solberg JK, Walmsley J, et al. The
nol. 2010;26:823–826. role of retained austenite in hydrogen embrittlement
[77] Bojack A, Zhao L, Morris PF, et al. In situ thermo- of supermartensitic stainless steel. Eng Fail Anal.
magnetic investigation of the austenitic phase during 2013;34:140–149.
tempering of a 13Cr6Ni2Mo supermartensitic stain- [94] Andersson JO, Helander T, Höglund L, et al. Thermo-
less steel. Metall Mater Trans A Phys Metall Mater Sci Calc & DICTRA, computational tools for materials
2014;45:5956–5967. science. Calphad Comput Coupling Phase Diagrams
[78] García-Junceda A, Capdevila C, Caballero FG, et al. Thermochem. 2002;26:273–312.
Dependence of martensite start temperature on fine [95] Bojack A, Zhao L, Sietsma J. Thermodynamic anal-
austenite grain size. Scr Mater. 2008;58:134–137. ysis of the effect of compositional inhomogene-
[79] Yang HS, Bhadeshia HKDH. Austenite grain size and ity on phase transformations in a 13Cr6Ni2Mo
the martensite-start temperature. Scr Mater. 2009;60: supermartensitic stainless steel. Solid State Phenom.
493–495. 2011;172-174:899–904.
[80] Ghosh G, Olson GB. Kinetics of F.C.C. → B.C.C. het- [96] Borgenstam A, Engström A, Höglund L, et al. DIC-
erogeneous martensitic nucleation-I. The critical driv- TRA, a tool for simulation of diffusional transfor-
ing force for athermal nucleation. Acta Metall Mater. mations in alloys. J Phase Equilibria. 2000;21:269–
1994;42:3361–3370. 280.
[81] Williams DB, Carter CB. 9.3.B Bright-field and dark- [97] Larsson H. A model for 1D multiphase moving phase
field imaging. Transmission Electron Microscopy. Part boundary simulations under local equilibrium con-
1 Basics. 2nd ed. New York: Springer. ditions. Calphad Comput Coupling Phase Diagrams
[82] Villa M, Somers MAJ. Thermally activated marten- Thermochem. 2014;47:1–8.
site formation in ferrous alloys. Scr Mater. 2018;142: [98] Morito S, Huang X, Furuhara T, et al. The morphology
46–49. and crystallography of lath martensite in alloy steels.
[83] LeBrun T, Nakamoto T, Horikawa K, et al. Effect Acta Mater. 2006;54:5323–5331.
of retained austenite on subsequent thermal pro- [99] Miyamoto G, Shibata A, Maki T, et al. Precise
cessing and resultant mechanical properties of selec- measurement of strain accommodation in austenite
tive laser melted 17-4 PH stainless steel. Mater Des. matrix surrounding martensite in ferrous alloys by
2015;81:44–53. electron backscatter diffraction analysis. Acta Mater.
[84] Kromm A, Brauser S, Kannengiesser T, et al. High- 2009;57:1120–1131.
energy synchrotron diffraction study of a transforma- [100] Stan M, Reardon BJ. A Bayesian approach to evaluat-
tion induced plasticity steel during tensile deforma- ing the uncertainty of thermodynamic data and phase
tion. J Strain Anal Eng Des. 2011;46:581–591. diagrams. Calphad. 2003;27:319–323.
[85] Xiong XC, Chen B, Huang MX, et al. The effect [101] Kattner UR. The thermodynamic modeling of multi-
of morphology on the stability of retained austen- component phase equilibria. Jom. 1997;49:14–19.
ite in a quenched and partitioned steel. Scr Mater. [102] Hillert M. Empirical methods of predicting and repre-
2013;68:321–324. senting thermodynamic properties of ternary solution
[86] Brauser S, Kromm A, Kannengiesser T, et al. In-situ phases. Calphad. 1980;4:1–12.
synchrotron diffraction and digital image correlation [103] Kuehmann C, Tufts B, Trester P. Computational design
technique for characterizations of retained austenite for ultra high-strength alloy. Adv Mater Process.
stability in low-alloyed transformation induced plas- 2008;166:37–40.
ticity steel. Scr Mater. 2010;63:1149–1152. [104] Perrut M. Thermodynamic modeling by the calphad
[87] Yan K, Liss K-D, Timokhina IB, et al. In situ method and its applications to innovative materials.
synchrotron X-ray diffraction studies of the effect AerospaceLab; 2015, p. 1–11.
of microstructure on tensile behavior and retained [105] Nilsson J-O. Super duplex stainless steels. Mater Sci
austenite stability of thermo-mechanically processed Technol. 1992;8:685–700.
transformation induced plasticity steel. Mater Sci Eng [106] Miller MK, Kelly TF, Rajan K, et al. The future of
A. 2016;662:185–197. atom probe tomography. Mater Today. 2012;15:158–
[88] Oliver EC, Daymond MR, Withers PJ, et al. Stress 165.
induced martensitic transformation studied by neu- [107] Williams DBB, Carter CB, Barry Carter C, et al.
tron diffraction. Mater Sci Forum 2002;404-407: Transmission electron microscopy. 2nd ed. New York:
489–494. Springer; 2009.