Twinning-Induced Plasticity (TWIP) Steels

Acta Materialia 142 (2018) 283e362
Contents lists available at ScienceDirect
Acta Materialia
journal homepage: www.elsevier.com/locate/actamat
By invitation only: overview article
Twinning-induced plasticity (TWIP) steels

Bruno C. De Cooman a, *, Yuri Estrin b, c, Sung Kyu Kim d
a
Graduate Institute of Ferrous Technology, Pohang University of Science and Technology, Pohang, Republic of Korea
b
Department of Materials Science and Engineering, Monash University, Clayton, Australia
c
Laboratory of Hybrid Nanostructured Materials, NUST MISiS, Moscow, Russia
d
POSCO Technical Research Laboratories, Gwangyang, Republic of Korea
a r t i c l e i n f o a b s t r a c t
Article history: This article reviews original work and important new developments in the field of deformation behavior
Received 14 February 2017 of high manganese face-centered cubic g-Fe alloys. Owing to their exceptional mechanical properties,
Received in revised form these alloys, referred to as twinning-induced plasticity, or TWIP, steels, have come to the fore as prime
19 June 2017
candidate materials for light-weight applications, notably in automotive, shipbuilding, and oil and gas
Accepted 21 June 2017
Available online 24 June 2017
industries. It is established that a superior combination of strength and ductility exhibited by TWIP steels
is associated with a specific character of the variation of the dislocation density. The defining feature of
TWIP steels is the small magnitude of the intrinsic stacking fault energy. In addition to limiting the
Keywords:
Mechanical twinning
dynamic recovery rate, the low stacking fault energy of TWIP steels results in the formation of isolated
Dislocation glide stacking faults and deformation twins, which reduces the dislocation mean free path. Both effects lead to
High manganese steel an increased strain hardening rate. Despite the progress made, there are still considerable differences
Stacking fault energy between the models proposed for the microstructural evolution during the deformation of TWIP steels
Strain hardening and the concomitant strain hardening behavior. The review surveys the experimental literature, sum-
marizes the current modeling concepts, and identifies the outstanding issues with TWIP steels that
require the attention of the materials science community. Suggestions for the directions of future
research on twinning-induced plasticity steels are offered.
© 2017 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
2. Strain hardening of TWIP steels e phenomenology and early models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
2.1. The taxonomy of strain hardening stages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
2.2. Modeling approaches involving deformation twinning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
2.2.1. Role of deformation twinning in the isotropic strain hardening of TWIP steel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
2.2.2. Kinematic hardening models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
2.2.3. Stress due to size effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294
2.2.4. Controversial issues relating to the role of deformation twinning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294
2.2.5. Short range ordering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
2.2.6. Dynamic strain aging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
2.2.7. Stacking faults . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
3. The energy of planar faults in a TWIP steel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
3.1. Intrinsic vs. extrinsic stacking faults . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
3.2. Stacking fault energy computations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
3.2.1. Thermodynamic approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
3.2.2. Ab initio approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
3.2.3. First principles computations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
* Corresponding author.
E-mail address: [email protected] (B.C. De Cooman).
http://dx.doi.org/10.1016/j.actamat.2017.06.046
1359-6454/© 2017 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
284 B.C. De Cooman et al. / Acta Materialia 142 (2018) 283e362
3.3. Effect of alloying elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312

3.3.1. Carbon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
3.3.2. Manganese . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
3.3.3. Aluminium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
3.3.4. Silicon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
3.3.5. Nitrogen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
3.3.6. Copper, nickel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
3.3.7. Chromium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
4. Mechanical properties of TWIP steel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
4.1. Elastic properties of TWIP steel: Young's modulus anomaly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
4.2. The elasto-plastic transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
4.3. Solute solution strengthening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
4.4. Thermally activated plastic flow in TWIP steels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
4.5. Grain size strengthening in TWIP steels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
5. Twinning mechanisms in TWIP steels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323
5.1. Mechanisms of deformation twinning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323
5.2. Critical stress for twinning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
5.3. Twinning kinetics and twinning saturation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
5.4. Influence of dislocation slip and twinning on texture development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
5.5. Rolling texture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
5.6. Texture evolution modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
5.7. Recrystallization texture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
6. Dynamic strain aging and strain rate sensitivity in carbon-alloyed TWIP steel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340
7. Recovery annealing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
8. TWIP effect during cyclic deformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
9. High strain rate behavior of TWIP steels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
10. Fracture of TWIP steels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
11. Fracture in hole expansion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
12. Hydrogen-delayed fracture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
13. Liquid metal-induced embrittlement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
14. Hot ductility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
15. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
On-line information on TWIP steel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
1. Introduction principle calculations. The experimental analysis of the properties

of TWIP steels has also profited from the use of advanced tech-
Some iron alloys and steels have impressive plasticity- niques for microstructural characterization of materials, such as
enhancing potential, which is not yet fully exploited in engineer- synchrotron X-ray diffraction, electron backscattering diffraction,
ing applications, as it requires a thorough understanding of the 3D atom probe tomography, and micromechancial testing methods
underlying mechanisms and their activation during straining. This (nano-hardness, micro-pillar testing). Finally, a more sophisticated
necessitates an approach to steel design incorporating a selection of analysis of the results of standard macroscopic mechanical tests
composition, microstructure, and processing parameters based on involving the strain rate and temperature dependence of the me-
sound theoretical principles. Most formable ferritic steels exhibit chanical properties has contributed to a better understanding of
uniform engineering elongation less than 25% and relatively low the mechanisms underlying strength and plasticity of TWIP steels.
ultimate tensile strength (≪1 GPa). The formability of these steels is In this article, a historical overview and an assessment of the cur-
based on the control of their crystallographic texture, rather than rent state of our understanding of the mechanical properties of
the strain hardening. As a consequence, higher strength is usually TWIP steels are given.
achieved at the cost of ductility. There is a way to resolve this The ground breaking contributions of Gra €ssel et al. [1e4] and
conflict of properties, though: by designing fully austenitic steels or Frommeyer et al. [5] to the science of Fe-Mn-Si-Al TWIP steels with
austenite-containing multi-phase steels with an enhanced strain a very low carbon content (<100 ppm) prompted a global research
hardening rate, both high strength and good formability can be effort aimed at developing a better understanding of the plasticity-
achieved. Twinning-induced plasticity, or TWIP, steels belong to enhancing mechanisms which are activated during the plastic
this category of ferrous alloys. They are characterized by a high deformation of these steels and similar C-alloyed high Mn face-
strain hardening, large uniform elongation and high ultimate ten- centered cubic (fcc) g-Fe. Gra €ssel and Frommeyer observed that
sile strength levels. These properties make them candidate light- large scale deformation twinning occurred in a TWIP steel when its
weighting materials for large scale use in the automotive industry, Mn content was larger than 25%, the Al content was in excess of 3%,
LNG-shipbuilding, oil-and-gas exploration and non-magnetic and the Si content was in the range of 2e3% (by weight). A very
structural applications. The rapid progress made in the funda- favorable strength-ductility balance was attained. An important
mental understanding of the mechanical behavior of TWIP steels is quantity that characterizes this balance is the product of the ulti-
a result of a tremendous global research effort. It is a convincing mate tensile strength and the total elongation (UTS TE value, or
illustration of what materials research can achieve through Rm A value, in European publications). A very large value of this
consequent use of computational thermodynamics and first quantity in excess of 50,000 MPa %, resulting mainly from an
B.C. De Cooman et al. / Acta Materialia 142 (2018) 283e362 285
extremely large total tensile elongation of 80%, was reported. dynamic strain aging, this region of the stress-strain curve is
Contemporary TWIP steels exhibit high ultimate tensile stresses designated as DSA. The superior forming properties of specific
paired with exceptionally large tensile elongations (60%) over a TWIP steels, such as the Fe-18%Mn-0.5%C-1.5%Al TWIP steel, have
wide range of strain rates (103-103 s1). The importance of TWIP already been reviewed in great detail [9]. In fact, this class of highly
steels as breakthrough structural materials cannot be under- ductile, high or ultra-high strength steels has been the subject of a
estimated as the TWIP steel concept has also opened up the pos- substantial effort aimed at their commercial development. In the
sibility of creating new steel grades with a very wide range of case of automotive materials, the main motivation for current
properties. These developments have already been addressed in research into the development of high-performance Fe-Mn-Al-C
the literature [6e8]. The present review aims at critically evaluating TWIP steel is the expectation that it can offer ultra-high strength for
our current understanding of TWIP steels and accounting for the structural reinforcement, a superior ductility for the ease of press
most recent developments in this area. The review focusses on the forming, and a great energy absorption capability for improved
strain hardening behavior of TWIP steels, as a better understanding crashworthiness, which is crucial for a vehicle. In addition, their
of many aspects of this fundamental property has evolved in the large scale use will result in a reduction of a vehicle's mass, a
recent years, which makes it necessary to review its defining role in lowering of greenhouse gas emissions, a drastic increase of gas
the strength and plasticity enhancement in TWIP steels. Other as- mileage, and an improvement of passenger safety.
pects of the mechanical performance of TWIP steels are also dis- The alloy content of TWIP steels is typically 15e30% Mn, 0e1% C,
cussed in relation to their strain hardening behavior. 0e3% Al, and 0e3% Si. Secondary alloying additions include Cr, Cu,
N, Nb, Ti and/or V. The composition critically influences the
2. Strain hardening of TWIP steels e phenomenology and magnitude of the stacking-fault energy of TWIP steels e a quantity
early models that will be shown to be decisive for their special mechanical
properties. The intrinsic stacking fault energy of TWIP steels, gisf, is
2.1. The taxonomy of strain hardening stages commonly in the range of 15e45 mJm-2 at room temperature. In
this gisf-range, deformation twinning is activated during deforma-
Figs. 1 and 2 illustrate an exceptional strength-ductility combi- tion. A subtle interaction between glide dislocations, grain
nation of TWIP steels, as exemplified by 18%Mn-0.6%C-1.5%Al TWIP boundaries, wide stacking faults, and deformation twins results in a
steel, by comparing its properties with those of Ti-stabilized sustained high strain hardening during deformation. Strain hard-
interstitial-free (Ti-IF) ferritic steel. The latter is considered as one ening rates as high as G/20, where G is the shear modulus, large
of the most formable industrial steel grades. The formability of uniform elongations, and high ultimate tensile strength levels have
these steels is controlled by their crystallographic texture, their been reported. In this section, we present experimental facts and
grain size and the presence of precipitates. The development of a theoretical considerations that contribute to a current picture of the
pronounced ND//〈111〉 fiber texture results in an improved form- deformation processes in TWIP steels, in which deformation
ability and some limited texture hardening. Higher levels of twinning plays an important role.
strength attained through grain size reduction and precipitation Assessing the contribution of the twinning mode to strain, Qin
hardening are always achieved at the cost of ductility. The stress- [10] showed that the contribution of twinning to the overall strain
strain curves of Ti-IF and TWIP steel are presented in Fig. 2 in in TWIP steels depends on the twin volume fraction and the crys-
terms of engineering (s,e) and true (s,ε) stresses and strains, tallographic texture. Assuming that twinning was the only defor-
mation mode, and considering the twinning shear to be given
respectively. It is obvious that compared to the Ti-IF steel, the TWIP qffiffiffi
steel exhibits a higher strain hardening rate (defined throughout bybd< 112 > ¼ 36 ¼ 0:707, where b < 112 > and df111g denote the atomic
f111g
this paper as the slope of a stress-strain curve, ds/dε), larger uni- spacing in the crystallographic direction < 112 > and the distance
form elongation εu, and a greater ultimate tensile strength. It also between the crystallographic planes f111g, respectively, Qin esti-
has a relatively low yield strength and shows virtually zero post- mated that the true strain resulting from twinning was less than
uniform elongation. In addition, some serrations are visible on 0.15. This figure is much lower than the uniform true strain for
the stress-strain curve at large strains. Due to its relation to
(a) (b)
Fig. 1. Comparison of a bulge test carried out on (a) Ti-stabilized interstitial-free (IF) ferritic steel and (b) austentic Fe-18%Mn-0.6%C-1.5%Al TWIP steel. The TWIP steel is still
undamaged at a dome height that is 31% larger than the IF steel dome height at failure.
1800
1600 ( )
1400
TWIP steel
G
1200 ~
Stress, MPa
27 s (e)
1000
800 DSA
600
G 78GPa G
~ ( )
400 180
s (e) IF steel
200
G 82GPa
u u
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Strain
Fig. 2. Comparison of the uniaxial tensile stress-strain curves for a Ti-stabilized interstitial-free (IF) ferritic steel (bcc crystal structure) and an austentic Fe-18%Mn-0.6%C-1.5%Al
TWIP steel (fcc crystal structure), illustrating a considerable difference in mechanical properties resulting from the more than six times larger strain hardening rate of the TWIP steel
as compared to the IF steel.
TWIP steel, which is typically larger than 0.5. As the volume frac- in the strain hardening curve of Fe-18%Mn-0.6%C-1.5%Al TWIP
tion of mechanical twins is typically in the range of 0.1e0.2, the steel. Note that the strain hardening rate reaches a maximum
contribution of mechanical twinning to plastic strain for TWIP steel magnitude of approximately G/30, a value which is close to G/
does not exceed 3%. This implies that the plastic flow of a TWIP 20~4 GPa given by Kocks and Mecking as the highest strain hard-
steel is dominated by dislocation glide, whilst the deformation ening rate for steel [14].
twins have a secondary, albeit very important, effect on the me- While the mentioned general trend of a continual drop of the
chanical properties. As will be shown below, the positive influence strain hardening coefficient characteristic of stage III strain hard-
of deformation twinning on the mechanical performance of TWIP ening of fcc metals is largely observed for TWIP steels as well, the
steels is chiefly through its effect on the evolution of the dislocation latter exhibit a greater complexity of strain hardening. Accordingly,
density. the taxonomy of the strain hardening stages in TWIP steels pro-
The deformation curve of fcc metals is dominated by stage III posed by Kalidindi et al. is different from the classical one that
hardening, and polycrystalline TWIP steels are no exception to that. applies for materials whose plastic deformation is governed by
In the Kocks-Mecking-Estrin one internal variable model [11] the dislocation glide alone [14]. According to this taxonomy, stage A is
description of stage III strain hardening is based on the evolution of the initial strain hardening stage, which is controlled by the
the average dislocation density regarded as a single internal vari- dislocation density evolution in the absence of twin formation. It is
able. In their work on the strain hardening of fcc metals, Kocks and equivalent to the standard stage III of strain hardening of high gisf
Mecking, as well as the followers of their modeling approach fcc metals and alloys, in which the strain hardening rate decreases
[12e14], highlighted the central role of the normalized stacking continuously. This behavior is associated with the evolution of the
g
isf dislocation density characterized by a continuously decreasing rate
fault energy, Gb , as a key scaling parameter for the strain hardening
of metals as different as Al, which has a high gisf and deforms by of dislocation storage. This results from a relatively constant
dislocation glide only, and Ag, in which deformation twinning is dislocation generation rate and an increasing rate of dislocation
common due to its low magnitude of gisf. Here b denotes the annihilation (dynamic recovery), chiefly by cross slip and annihi-
magnitude of the dislocation Burgers vector. lation of screw dislocations of opposite sign. Stage A is usually
Changes in the composition of single phase fcc Fe-alloys, such as observed during a relatively short strain interval. According to
TWIP steels, result in variations in stacking fault energy, and, as a Asgari et al. [15], no twinning takes place in stage A. At this
consequence, in strain hardening and strength. In TWIP steels the deformation stage, overlapping stacking faults are usually
combination of strengthening by solute interstitial atoms (C and N) observed, and it can be considered as an incubation stage in which
and the control of gisf by suitable alloying additions (Mn, Al, Si, Cr, C dislocation interactions generate twin nuclei.
and/or N) promotes twinning as an additional deformation mode Stage B is the primary twinning stage, which is characterized by
and leads to an increase of the strain hardening rate. an increasing strain hardening. It starts when primary deformation
In their report on the strain hardening response of low gisf fcc twinning is initiated in favorably oriented grains. Over this stage
alloys, Kalidindi and co-workers [15,16] suggested that the strain the strain hardening rate initially increases and then remains more
hardening of alloys for which the plastic deformation includes both or less constant during straining. Stage B typically starts in the true
dislocation glide and deformation twinning goes through four strain range of 0.03e0.04, which indicates that dislocation slip
distinct stages. Fig. 3 identifies these four stages of strain hardening activity and dislocation-dislocation interactions occur prior to
Fe-17%Mn-0.4%C-1.3Al
4000 3800
3500 3600
Stage A
3400
Strain hardening, MPa
Strain hardening, MPa

3000
3200
2500
3000
2000 Point of highest strain hardening
2800
1500
Stage C Stage D DSA 2600
1000
2400
500 2200
Stage B Stage C
0 2000
200 400 600 800 1000 1200 1400 300 320 340 360 380 400
True stress, MPa True stress, MPa
(a) (b) T
Fig. 3. (a) Strain hardening of Fe-17%Mn-0.4%C-1.3%Al TWIP steel showing four strain hardening stages. (b) Enlargement of the initial strain hardening range. The transition from
stage A to stage B corresponds to the onset of deformation twinning, and the minimum in strain hardening determines the twinning stress sT. The point of maximum strain
hardening is reached at the transition from stage B to stage C. The fourth strain hardening stage, stage D, as well as the section of the curve in (a) associated with dynamic strain
aging (DSA) are discussed later in the text.
twinning. This also suggests that dislocations play a role in the of the grains by the deformation twins, higher stresses are required
creation of twin initiation sites. The experimental twinning stress to generate more twins. The first three stages of strain hardening
reported in the present article corresponds to the stress at the onset have been correlated experimentally with the strain dependence of
of stage B. Fig. 4 shows that the twinning stress, defined in this way, the twinning kinetics in Fe-20%Mn-1.2%C TWIP steel by Renard and
i.e. at the minimum of the strain hardening rate in Fig. 3b, is only Jacques [23]. According to the work of Kalidindi's group [15,16], the
slightly higher than the yield stress. The difference between the slowly decreasing strain hardening observed in stage C is termi-
two quantities for the Fe-18%Mn-0.6%C-1.5%Al TWIP steel is nated when extensive twinning is initiated on secondary twin
approximately 46 MPa. The critical strain for the onset of defor- systems, but in certain cases diffuse necking has been reported to
mation twinning is in the range of 0.006e0.012. Bracke et al. report set in at the end of stage C.
the observation of deformation twins by transmission electron In a scenario proposed by Kalidindi and co-authors, secondary
microscopy in Fe-22%Mn-0.5%C TWIP steel deformed to a strain of twin systems are activated during stage D. Multiple twin-twin in-
0.02 [17]. According to Kim et al. [18], the critical strain for defor- tersections are formed which considerably limit dislocation glide
mation twinning is slightly higher, 0.045 and 0.055, for Fe-18%Mn- distances. This lead to a reduction of the rate at which the strain
0.6%C and Fe-18%Mn-0.6%C-1.5%Al TWIP steel, respectively. hardening decreases with strain, and a more or less constant strain
Mahato et al. [19] also report that in Fe-27%Mn-2.5%Si-3.5%Al TWIP hardening rate is observed over stage D. Note that stage D is not
steel, twinning is initiated at the start of stage B of the strain observed in the strain hardening curve of some TWIP steels.
hardening curve. It should be noted, however, that a sharp increase Mahato et al. [19] consider the absence of stage D in the strain
in the strain hardening upon the onset of stage B depicted in Fig. 3 is hardening curve for Fe-27%Mn-2.5%Si-3.5%Al TWIP steel to be an
not always observed. There is also no general consensus in litera- indication of the absence of secondary twinning.
ture that stage B can be unequivocally linked to the initiation of Some reports consider more than four strain hardening stages.
deformation twinning. Thus, in their report on the strain hardening Barbier et al. [21,24] introduce a fifth stage, stage E, during which a
of Fe-15%Mn-0.7%C-2%Al-2%Si TWIP steel, Rahman et al. [20] sug- continuous decrease of the strain hardening rate occurs. They
gest that in stage B, which they refer to as stage 2, the rise of the report that the twin volume fraction grows significantly in this
hardening rate is due to an increase in the twinning rate, rather stage, with twin bundles becoming denser and thicker. This
than the initiation of deformation twinning. Barbier et al. [21] and observation appears to be at odds with the well documented
Lebedkina et al. [22] attribute the increase in the strain hardening saturation of deformation twinning. The same authors report the
rate, or the “bump in hardening”, in Fe-22%Mn-0.6%C TWIP steel to development of large intra- and inter-granular regions with high
the activation of a secondary twin system. dislocation density, which result in a pronounced strain localization
The highest strain hardening rate is observed at the transition prior to fracture.
from stage B to stage C. In stage C the strain hardening rate di- In multi-component alloys, such as TWIP steels, relatively small
minishes gradually with strain. This decreasing strain hardening is compositional differences are expected to influence the value of gisf,
attributed to a reduction in the rate of primary twin formation. As and hence the strain hardening behavior. Some of the differences in
the original grain size is reduced by the progressive segmentation strain hardening behavior discussed in this section may therefore
700
0.015
Yield stress, Twinning stress, MPa
600
Strain at onset of twinning

0.012
500
400 0.009
46MPa
300
0.006
200
Fe-18%Mn-0.6%C-1.5%Al 0.003
100 Yield stress
Twinning stress T
Twinning shear stress T
0 0.000
-200 0 200 400 -200 0 200 400
Temperature, ºC Temperature, ºC
(a) (b)
Fig. 4. (a) Temperature dependence of the yield stress, the twinning stress and the twinning shear stress of Fe-18%-0.6%C-1.5%Al TWIP steel. (b) Temperature dependence of the
strain at the onset of deformation twinning in this steel.
be due to minor differences in alloy composition. this resulted in a number of modeling assumptions, such as unre-
alistically low twinning stresses and excessively large twin volume
fractions, which were not compatible with the experimental
2.2. Modeling approaches involving deformation twinning
observations.
Using electron channeling contrast imaging (ECCI) Gutierrez-
Guttierez-Urrutia and Raabe [25] and Saeed-Akbari et al. [26],
Urrutia and Raabe [25] and Zaefferer and Elhami [29] were able
who studied Fe-22%Mn-0.6%C TWIP steel and Fe-(19e27)%Mn-
to observe the evolution of both the deformation twinning and the
(0.3e1.2)%-(0.0e3.5)%Al TWIP steels, respectively, provided in-
dislocation substructure in great detail. They argue that the dislo-
depth analysis of strain hardening data and correlated these
cation mean free path is determined by both the twin spacing and
with microstructural observations. They proposed a slightly
the characteristic length scale defined by the dislocations sub-
different interpretation of the first three stages of strain hard-
structure. In their investigation of samples deformed under uni-
ening. While they agree with the view of Kalidindi and co-workers
axial tensile loading, they observed three types of grains in the
that stage A is dominated by dislocation dynamic recovery, their
deformed microstructure. Type I grains, with a 〈001〉-type direction
conclusion is that deformation twinning initiates already in this
aligned with the tensile axis, have a low density of deformation
early deformation stage. In fact, according to Guttierrez-Urrutia
twins. Type I grains also develop a clear equiaxed dislocation cell
et al. [27] yielding and twinning occur simultaneously, i.e. the
substructure. Wavy slip is promoted in type I grains. The orienta-
critical resolved shear stress for glide and twinning are equal, i.e.
tion of type I grains also favors multiple slip. Type II grains have a
tcrss ¼ tT. Guttierez-Urrutia et al. argue that the strain hardening in higher density of twins belonging to the primary deformation twin
stage B is related to the evolution of the dislocation ensemble to a
system. Finally, type III grains, whose 〈111〉-type direction is parallel
dislocation cell/dislocation wall structure. These dislocation ar-
to the tensile axis, have a high density of deformation twins of both
rangements act as very effective obstacles to dislocation glide.
the primary and the secondary twin systems. Type III grains also
According to these authors, the influence of the deformation twins
contain a developed dislocation cell structure. High dislocation
on the strain hardening comes to bearing in stage C with the
density walls are a typical feature of planar dislocation glide. The
formation of dense deformation twin substructures, which are
twins appear to encounter little or no resistance from this dislo-
formed by deformation twins cutting through the already present
cation substructure as they are often seen to cross the entire grains.
dislocation substructure.
Based on their observations, Gutierrez-Urrutia and Raabe [25]
Rahman et al. [20] report that, in cyclically tested Fe-15%Mn-
proposed a physical model of strain hardening that accounts for
0.7%C-2%Al-2%Si TWIP steel, the nucleation of twins could even
dislocation glide and deformation twinning. They argued that the
occur at stresses below the yield stress. Their results have however
strain hardening is determined by both the deformation twin
been critically re-evaluated by Saleh and Gazder [28], who argue
spacing and the characteristic length scale of the dislocation sub-
that the conclusions of Rahman et al. were based on the failure to
structure. In particular, the dislocation substructure provides a
recognize a number of significant differences between plasticity in
large strain hardening rate in stage A. Using the Kuhlmann-
cyclic tests and monotonic tensile tests. Saleh and Gazder claim that
Wilsdorf similitude principle [30], they calculate the contribution consistent with the experimental observation that twinning does
of the dislocation cells to stress in type I grains as follows: not influence the development of the crystallographic texture. It
also implies that the contribution of the twinning strain to the total
K$G$b plastic deformation is very small, e.g. 3% at 35% strain. Therefore,
s ¼ M$fI $ (1)
D the twinning strain as such does not contribute significantly to the
Here M is the Taylor factor (M ¼ 2.44 for type I grains), fI is the uniform elongation prior to necking. However, the occurrence of
fraction of type I grains, D is the dislocation cell size, which varies deformation twinning can influence the evolution of the disloca-
from 750 nm to 180 nm when the strain increases from 0.05 to 0.40, tion density indirectly, by controlling the dislocation mean free
and the similitude constant K is a numerical parameter. A low K path (MFP). In the initial stages of deformation this quantity is
value is indicative of a high rate of dislocation storage. Gutierrez- mainly determined by the grain and/or dislocation cell size.
Urrutia and Raabe [25] note that for the TWIP steel they investi- Accordingly, the dislocation MFP decreases due to an increase of the
gated the refinement of the cell structure with strain corresponds dislocation density. The onset of deformation twinning at about 5%
to a smaller similitude constant, K ¼ 3.7, than in the case of high strain results in an additional reduction of the MFP.
stacking fault alloys (7 < K < 8), and ascribed this to the influence of According to Shiekhelsouk et al. [31] the onset of twinning is not
the stacking fault energy on the cell formation. discernible in the calculated strain hardening curves as a stop in the
The strain hardening behavior of TWIP steels has also been decrease of the ds/dε vs. ε curves at small plastic strains. They argue
considered by means of micro-mechanical models, such as the that the effect of the dislocation MFP reduction by the deformation
crystal plasticity finite element (CPFE) and visco-plastic self- twins occurs only at higher strains, when enough grains have two
consistent (VPSC) models. The important feature of these ap- active slip systems, because the twins belonging to the primary
proaches is that they offer a physically based picture of the behavior twin system are coplanar with it and therefore do not contribute to
of the material at the grain scale. Shiekhelsouk et al. [31] consider a a decrease in the MFP. The results the mentioned authors obtained
polycrystal of a low stacking fault energy material such as TWIP on single crystals illustrate two important aspects of deformation
steel to be equivalent to a representative volume element (RVE) twinning at the scale of the individual grains, which appear to have
comprised by a limited number of grains with specific orientations, been overlooked in the original Kalidindi model for strain hard-
which are subjected to homogeneous stress and strain rate ening in low gisf alloys. These are as follows: (a) the primary twins
boundary conditions. The constitutive equations for the individual do not contribute to a MFP reduction for the primary slip disloca-
grains take into account three grain orientation-dependent strain tions, and (b) the effect of the mechanical twins in reducing the
mechanisms, viz. instantaneous elasticity, viscous or rate- dislocation mean free path, and thereby increasing the strain
dependent thermally activated crystallographic slip, and rate- hardening, is not instantaneous as the first effect of the formation of
independent mechanical twinning, in well-defined imposed con- twins is to supply an additional strain which relaxes intergranular
ditions of stress, strain and temperature. In polycrystalline TWIP stresses. These conclusions are illustrated in the representative
steel intergranular interactions due to non iso-strain conditions at single grain response taken from the work of Shiekhelsouk et al.
the grain boundaries lead to internal stresses. In the self-consistent [31] shown in Fig. 5. All grains have three to five active slip systems
approach, each grain is considered to be embedded in a homoge- from the very start of plastic deformation, and no additional slip
neous matrix, which represents a homogenized effective contin- systems become significant with strain. The primary twin system
uum interacting with that grain regarded as an inclusion. (a=2½011ð111Þ, in the case of Fig. 5) is usually coplanar with the
The model proposed by Shiekhelsouk et al. [31] shows that at a primary slip system, and non-coplanar with at least one of the
strain of 15%, some 90% of the grains in a TWIP steel contain twins. other slip systems. As soon as twinning sets in, there occurs a steep
Predominantly, two twin systems operate within a grain. The twin drop in the strain hardening due to a contribution of twins to
volume fraction remains low, e.g. only 6% at 30% strain. This is plastic strain. This happens until a reduction of the MFP by the
(111)[011]
Grain
primary slip system One Two
0.3
twin twin size
1200 0.2 system systems 14
Local equivalent stress, MPa
0.1 4500
Slip
0.0 (111)[101] 12
Mean free path, m
1000
-0.1
4000
-0.2 10 Twin
d /d , MPa
800 -0.3 (111)[011] spacing

Twin volume fraction
0.06 3500 8
One Two
600 (111) (011) (111) 0.05 twin twin
system systems 6
0.04 3000
400 P Forest
(101) (101) 0.03 (111) 4
(001) dislocations
0.02 [121] 2500
200 2
(111)
0.01
(111) (011) (111)
[211]
0 0 2000 0
0 0.05 0.10 0.15 0.20 0.25 0 0.05 0.10 0.15 0.20 0.25 0 0.05 0.10 0.15 0.20 0.25 0 0.05 0.10 0.15 0.20 0.25
Local viscoplastic Local viscoplastic Local viscoplastic Local viscoplastic

(a) equivalent strain (b) equivalent strain (c) equivalent strain (d) equivalent strain
Fig. 5. (a) Calculated stress-strain curve for a grain with the orientation shown in the inset. Three slip systems and two twinning systems are activated during straining. (b)
Evolution of the slip for the three activated slip systems and the volume fraction of the activated twin systems in the grain. (c) Strain hardening vs. strain curve. (d) Evolution of the
dislocation mean free path with strain: effect of grain size, forest dislocation density, and deformation twins. The figures are based on data published by Shiekhelsouk et al. [31].
twins comes to bearing to give rise to an increase in strain hard- dislocation mean free path, especially for dislocations on slip planes
ening. Thus, twinning contributes to the deformation in two intersecting the twinning plane. A further factor to consider is the
competing ways. As seen from Fig. 5c, the initial steep decrease of high dislocation storage capacity due to the strong inhibition of
the strain hardening occurs up to a strain of 6%. At that strain, a dislocation cross-slip resulting from the low value of gisf. Finally,
decrease of the dislocation MFP and the concomitant rise of the additional strengthening may stem from the very small thickness of
strain hardening rate sets in due to the activation of three non co- the deformation twins (the size effect).
planar slip systems. The primary twin system is activated at about Three types of models put forward to explain the mechanism of
the same strain of 6%. It is co-planar with one of the three slip enhancement of strain hardening and tensile ductility due to
systems. A secondary twin system is activated at about 15% strain. deformation twinning are central to our subject: (a) the “dynamic
There is also a small dip in the ds/dε vs. ε curve at the onset of the Hall-Petch effect” [33,34], (b) the back stress effect [35], and (c) the
activation of this second twin system. As it is non coplanar with the “composite effect” proposed by Guttierez-Urrutia and Raabe [25].
primary slip systems and the primary twin system, twin-twin in- The key aspects of the three types of models are illustrated in Fig. 6.
teractions inhibit the twin activity and the model predicts that no The first one accounts for isotropic strain hardening, which is related
further twin systems are activated at higher strains. to the growth of the dislocation density e a scalar quantity con-
Another group of authors, Dancette et al., also used crystal trolling the overall level of stress. The other two describe kinematic
plasticity based finite element method to predict the behavior of hardening and are associated with the directionality of the strain
TWIP steel at the grain level [32]. They showed that in the course of hardening behavior. We begin with the isotropic strain hardening
uniaxial tensile deformation the grains get re-oriented towards the model due to Allain and Bouaziz, which generated a number of
stable 〈100〉//td and 〈111〉//td orientations (td: sample tensile di- follow-up modeling exercises, and then move on to the kinematic
rection). The model predicts a lower, yet non-zero, twin volume hardening approaches in Section 2.2.2.
fraction in grains ending up close to the 〈100〉 orientation. Grains
with a high density of twins originate from grains with orientations 2.2.1. Role of deformation twinning in the isotropic strain hardening
between 〈110〉//td and 〈111〉//td. of TWIP steel
While the different attempts at modeling the mechanical Schematics of the models based on the Allain-Bouaziz dynamic
properties of TWIP steels give generally a good agreement with Hall-Petch concept are shown in Fig. 7. The results illustrate the
experimental results, it is important to realize that more work is influence of the deformation twinning kinetics on the mechanical
needed in this area, as the values of some of the key model pa- properties of a TWIP steel in terms of the two internal variable
rameters vary strongly from report to report. For example, the value model proposed by Kim et al. in which the twin boundaries act as
of the critical resolved shear stress tcrss for slip in Fe-22%Mn-0.6%C impenetrable obstacles for dislocation glide [18,36]. The calcula-
TWIP steel obtained through fitting by Shiekhelsouk et al. [31] is tions are based on a constitutive model developed by Estrin and
95 MPa; the critical resolved shear stress for twinning is much Kubin [37], which accounts for the mobile and forest dislocation
lower, tT ¼ 47.5 MPa, i.e. tTztcrss/2. In contrast, Dancette et al. [32] density evolution during plastic deformation. The model also al-
came up with the values of tcrss ¼ 125 MPa and tT ¼ 230 MPa for Fe- lows the computation of the contribution of dynamic strain aging
20%Mn-1.2%C TWIP steel, i.e. tT z 2 tcrss. to the strain hardening of TWIP steel. In addition, the model con-
The main conclusions that transpire from the studies mentioned siders the fact that only a certain fraction of the grain population
above can be condensed to the following statement. The key reason develop deformation twins during straining.
for the combination of high strength and large ductility of TWIP Despite considerable advances in the state of the knowledge of
steels owing to deformation twinning is a decrease of the TWIP steels and the wide acceptance of the model for the strain
A A A
T~2GPa ys~0.5GPa
A i,T i,T
<d
BS
A A A
(a) (b) (c)
Fig. 6. Mechanisms of plasticity enhancement due to deformation twinning according to (a) the Bouaziz-Allain “dynamic Hall-Petch effect” [33,34], (b) the Bouaziz back stress effect
[35], and (c) the “composite effect” introduced by Guttierez-Urrutia and Raabe [25]. Here sA and tA denote the applied stress and the shear stress, respectively, and tBS denotes the
back stress; sys and sT represent the yield strength and the twinning onset stress, respectively; d stands for the average grain size and L for the average twin spacing. Finally, si,M
and si,T are the reaction stresses in the composite model that develop to compensate plastic strain incompatibility between the twins and the matrix: si,M denotes the tensile stress
in the matrix, while si,T is the associated compressive stress in the twins.
( ) 2
0.12
F 0.12 (1 e init
) 1800 5000
=8 =8
=6 1600 =6
0.10 =4 4000
Strain hardening (MPa)

1400
=4
True stress (MPa)

Twinning fraction
0.08 1200 =8
=2
3000
1000
0.06 =6
800 =4
=2 2000
0.04 600
400 =2
1000
0.02
200
init 0.03
0.00 0 0
0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5
True strain True strain True strain
(a) (b) (c)
Fig. 7. Effect of a reduction of the twinning kinetics parameter b entering the exponential in the strain dependence of twin volume fraction, F, on (a) the strain dependence of the
twin volume fraction, (b) the stress-strain curve, and (c) the strain hardening rate. The figures are based on the model developed by Kim et al. [36].
hardening of TWIP steels based on the dynamic Hall-Petch effect, fact that the deformation twins in TWIP steel are initially very
there is still no consensus on the fundamental mechanisms of thin, and that they do not thicken much with strain. The shape of
strain hardening in TWIP steels. In its simplest form, the basic the twins is usually considered to be plate-like, but close inspec-
model for strain hardening assumes that the dominant deformation by means of TEM shows that they tend to have a highly
tion mode in TWIP steels is dislocation glide, i.e. that strain hard- variable thickness (Fig. 8). Growth of the volume fraction of twins
ening is controlled by the kinetics of the material's dislocation is mostly furnished by the formation of new twins in the course of
storage during plastic deformation. The deformation-induced deformation. The steady nucleation of new deformation twins
twinning leads to a decrease of the effective dislocation glide dis- results in an increasing fragmentation of the grains by twin
tance, or mean free path. This results in a “dynamic Hall-Petch ef- lamellae with increasing strain. This is the origin of the dynamic
fect”, which leads to a pronounced increase of the strain hardening Hall-Petch effect. The Kocks-Mecking model originally used by
rate, a high tensile strength, and a substantial uniform elongation. Bouaziz and Allain [33,34] operates with a single internal variable
This early model has evolved quite considerably, and its more e the total dislocation density. It was refined by Kim et al. [18] to
recent versions will be discussed below. include two dislocation densities: the mobile and the forest ones.
In the original one internal variable model for strain hardening In this approach the original Kocks-Mecking equation for the
of TWIP steel put forward by Bouaziz and Guelton [33], the key evolution of the dislocation density was replaced by a set of two
factor enhancing strain hardening is considered to be a decrease of coupled differential equations based on the work of Estrin and
the dislocation mean free path due to progressive segmentation of Kubin, modified to take into account the contribution of the grain
the grains into smaller microstructural entities by deformation boundaries and the twin boundaries to the dislocation density
twins. The model is based on the Taylor relation [38] between the evolution [37]:
shear flow stress, t, and the dislocation density, rd(g), whose evo-
lution with the shear strain g can be described by an appropriate drm C1 ri C pffiffiffiffi 1 1 F
¼ 2 C2 rm 3 ri
version of the Kocks-Mecking equation [14]: dg b rm b bd btT 1 F
(5)
pffiffiffiffiffiffiffiffiffiffiffiffi dri C pffiffiffiffi 1 1 F
t ¼ t0 þ aGb rd ðgÞ ¼ C2 ri þ 3 ri þ þ C4 ri
dg b bd btT 1 F
drd ðgÞ 1 1 1 1 pffiffiffiffiffiffiffiffiffiffi (2)
¼ f rd ðgÞ ¼ ð þ þ k rd ðgÞ f rd ðgÞ Here rm and ri are the mobile and immobile (forest) dislocation
dg bl b d tðgÞ
densities. The terms containing the parameters C1, C2, C3 and C4 are
Here l is the effective dislocation mean free path, whose inverse related to the following physical processes. C1 specifies the
is defined as the sum of the inverse distances between various magnitude of the dislocation pinning term, assuming that immo-
impenetrable obstacles to dislocation motion and t(g) is the bile forest dislocations act as pinning points of fixed dislocation
average distance between the deformation twins. The latter sources. The decrease of the mobile dislocation density by in-
quantity is expressed in terms of the twin volume fraction F(g) and teractions with other mobile dislocations is taken into account
the twin thickness tT by the Fullman relation [39]: through the term C2rm. C3 describes the immobilization of mobile
dislocations with a mean free path proportional to r1/2 i , i.e.
1 FðgÞ assuming a spatially organized dislocation forest. C4 is associated
tðgÞ ¼ tT $ (3)
FðgÞ with dynamic recovery, i.e. the annihilation of forest dislocations by
cross slip. The approach makes it possible to compute the contri-
A shear strain increment is comprised by the weighted contri-
bution of dynamic strain aging to the flow stress. This contribution
butions from dislocation glide, dgd, and twinning shear, dgT:
was found to be minor in term of stress, viz. approximately 20 MPa.
However, through a negative contribution to the strain rate sensi-
dg ¼ ð1 FÞ$dgd þ F$dgT (4)
tivity of the flow stress, it is very important in the context of strain
This isotropic hardening model takes into consideration the localization and ductility, see below. The two internal variable
100 nm
( )
(a)
matrix
primary
~1nm
~1nm twins
double
diffraction
planar
faults
(b) (c)
Fig. 8. (a) TEM micrograph of Fe-18%Mn-0.6%C TWIP steel strained to fracture. The twin thickness, typically less than 50 nm, is highly variable and appears to be the results of a
chaotic succession of stacking faults. The small white arrows give an indication of the twin thickness variability. The two open white arrows indicate stacking faults in the matrix
region between twins. Note that the formation of a 30 nm thick twin requires 144 partials trailing intrinsic stacking fault on successive {111} glide planes. Although the twins are
very narrow one must consider the fact that most models for twin nucleation envisage two to three layers thick twin nuclei. A considerable amount of twin thickening must
therefore occur via a specific mechanism. (b) High resolution image of a stacking fault inside the matrix region (open white arrow). (c) Diffraction pattern showing evidence for
primary twin formation and stacking faults on secondary slip planes.
model due to Kim et al. [18] also took into account that deformation (w), respectively. Distinctly different values of the mean free path
twinning does not occur in some grains as a result of their crys- were considered for the cell interior (lc) and the cell walls (lw):
tallographic orientation. This aspect of the model will be discussed pffiffiffiffiffi
below. 1 1 1 rc
¼þ þ
A somewhat different constitutive model for dislocation-cell l c d iT t ic
pffiffiffiffiffiffi (7)
forming fcc metals and alloys was proposed by Roters et al. [40]. 1 1 1 rw
They considered three distinct dislocation populations: mobile ¼ þ þ
lw d iT t iw
dislocations in the cell interior, mobile dislocations in the cell
walls, and immobile dislocation (dipoles) in the cell walls. Their Here d is the grain size and t is the twin spacing. The quantities ic
three internal variable model was used by Steinmetz et al. [41] to and iw denote, respectively, the average number of dislocation
model the flow stress of Fe-22%Mn-0.6%C TWIP steel. The dislo- spacings in the cell interior and the cell walls a dislocation travels
cation density evolution equations of the model by Roters et al. do before being immobilized. Similarly, iT is the average number of
not contain a mean free path term l in the sense of the equations twin spacings a dislocation travels before being immobilized. A rule
of the Kocks-Mecking-Estrin model, cf. Eq. (2). Instead, the mean of mixtures was used by Steinmetz et al. [41] for the flow stress:
free path is introduced via the Orowan equation for the strain rate
controlled by the thermally activated passing of an obstacle by a t ¼ fc teff ;c þ fw teff ;w (8)
dislocation [42]:
where fc and fw are the volume fractions of the cell interiors and the
p q ! cell walls, respectively.
Qs teff ;i
g_ ¼ vd brd ¼ li n0 bri $ exp $ 1 pffiffiffiffi
kB T t0 þ aGb ri
2.2.2. Kinematic hardening models
(6)
In the models reviewed so far, the strain hardening was essen-
Here nd is the average dislocation velocity, rd is the mobile tially controlled by the dynamic Hall-Petch mechanism, the for-
dislocation density, n0 is the Debye frequency, Qs is the activation mation of twins gradually reducing the dislocation mean free path,
energy for slip, kB is the Boltzmann constant, and T is the absolute thereby contributing to enhanced dislocation storage and ensuing
temperature. The heuristic parameters p and q are related to the isotropic hardening. Bouaziz et al. [35] have identified another
force-distance profile for dislocation-obstacle interaction. The important mechanism of strain hardening. By using forward-
subscript i represents c or w, for the cell interior (c) and the cell wall reverse shear tests, they measured the back stress and
determined that at a strain of 0.2 it contributed about 50% of the 0.03. The situation is different when a TWIP steel is first strained in
flow stress. These results suggest that, in addition to isotropic tension and then in compression. In this case a considerable back
hardening considered above, substantial kinematic hardening (giv- stress develops, as shown in Fig. 9 (b).
ing rise to the Bauschinger effect) was involved. According to Bouaziz et al. [35] proposed a refinement of their original strain
Bouaziz et al. [35], the back stress is generated by pile-ups of dis- hardening model, which includes a term for the kinematic hard-
locations arrested at twin boundaries. Fig. 9 shows TEM micro- ening effect caused by long range internal stresses originating from
graphs of a pile-up of identical partial dislocations at a twin. The dislocation pile-ups containing n dislocations in a grain with a
progressively increasing interdislocation distance is a clear indi- dislocation mean free path L. The key equations in the model
cation of a repulsive interaction between the dislocations and the describe the evolution of the dislocation density and the dislocation
twin boundary. The micrograph also suggests that stress amplifi- mean free path:
cation at the tip of the pile-up has initiated the formation of a
dislocation within the twin and also shows that slip is not trans-
n
1
dr n0 k pffiffiffi
mitted across the twin. From their in situ neutron diffraction ex- ¼ þ r fr
periments, Saleh et al. [43] concluded that the back stress stems dg bL b (9)
from a combination of (a) intergranular residual stresses (Type II 1 1 1 FðgÞ
stresses), arising from the inhomogeneous deformation of grains ¼ þ
L d 2t 1 FðgÞ
due to the orientation dependence of the elastic modulus and the
yield strength, and (b) dislocation pile-ups at the intersections of Here n0 is the largest number of dislocations stopped at a twin
stacking faults. They report a back stress of 210 MPa, which boundary. Note that L is independent of d at small strains, and that
amounts to 65% of the stress in forward tension. the number of dislocations n in the pile-ups rapidly reaches its
It should be noted that precise measurement of the back stress is saturation value n0.
difficult as it is critically dependent on the offset strain chosen to The back stress can also be considered to be a result of the
measure the backward yield stress. This is the most likely expla- localized shear associated with a twin, the back shear stress
nation for the very different back stress values found in literature. stemming from the constraint of the adjacent grain. In other words,
The back stress was reported to be 68%, 65%, 50%, and 20% of the an internal forward stress acts on the twins and an internal back
flow stress according to Gil Sevillano [44], Saleh et al. [43], Bouaziz stress acts on the matrix. The back stress can be estimated by
et al. [35], and Guttierrez-Urrutia et al. [45], respectively. Collet [46] means of the Brown-Clarke equation [48] for a twin that is
determined the back stress by X-ray diffraction studies and re- considered as a hard Eshelby inclusion in the TWIP steel:
ported values similar to those published by Bouaziz et al. The
occurrence of a pronounced Bauschinger-type back stress in fcc tBS ¼ 2$c$FT $g*$DG $G (10)
alloys susceptible to deformation twinning was also observed in Fe-
24Mn-3Al-2Si-1Ni-0.06C TWIP steel by Saleh et al. [43] and in Fe- Here c is the Eshelby accommodation factor, which depends on
12%Mn-1.2%C Hadfield steel (with gisf ¼ 35 mJ/m2) by Karaman the shape and orientation of the twin formed in the softer TWIP
et al. [47]. The data of Chung et al. [9] shown in Fig. 9 (a) shows a steel matrix during forward loading, g* is the unrelaxed shear strain
small asymmetry for sheet samples of TWIP steel tested in tension built up on a specific slip system during forward loading, and DG is a
and compression along the rolling direction. The tension and correction factor for the shear modulus. Approximating the twins
compression curves converge rapidly, i.e. the same strain hardening as isotropic spherical inclusions with GT ¼ GM, i.e. assuming the
behavior in tension and compression is observed after a strain of elastic constants of the TWIP matrix and the twins to be equal, the
Fe-18%Mn-0.5%C-1.5%Al 800 Fe-22%Mn-0.6%C

ys: 457MPa
700 Strain rate: ~10-3s-1
Thickness: 1.47mm
600 Compression
True stress, MPa
Backstress, MPa
400
500
Tension
400
300
200
200
E=180GPa XRD data

100 Mechanical tests
0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.05 0.10 0.15 0.20 0.25
(a) True strain (b) True strain
Fig. 9. (a) Tension-compression asymmetry observed for a Fe-18%Mn-0.6%C-1.5%Al TWIP steel. (b) Strain dependence of the back stress as measured by tension-compression tests
and X-ray diffraction [35].
parameters c and D are expressed by size effect). This contribution to stress can be evaluated by noting
that the critical flow stress for the activation of a Frank-Read type
7 5n 1 dislocation source within a twin of thickness t is given by
c¼ z
15ð1 nÞ 2
(11) G$b110
D¼
GM
¼1
sT z3$ (18)
t
GM cðGT GM Þ
Combining the previous equations, one obtains the following where b110 is the atomic spacing in the direction 〈110〉. Assuming
equation for the back stress: t ¼ 30 nm, G ¼ 78 GPa, and b110 ¼ 0.255 nm, one obtains
sT z 2.0 GPa. High long-range internal stresses will therefore be
tBS zGM $FT $g* (12) generated in the grains of TWIP steels which contain very narrow
deformation twins (t z 20e30 nm). The very hard twins separate
For tensile tests, one has s ¼ Mt and g ¼ Mε, where M is the regions of softer non-twinned matrix. According to Gil Sevillano
Taylor factor. Taking М ¼ 3.06, FT ~0.1 and GM ¼ 78 GPa, one there are therefore two contributions adding up to the overall flow
obtains: stress through a rule of mixtures: (1- fT) sM, the contribution of
the matrix region, which is gradually refined by an increasing
sBS ðGPaÞz73$εp (13)
volume fraction of deformation twins (fT), and fT sT, the contri-
For a plastic strain εp of 0.01, the back stress is about 730 MPa. bution from the very thin, high strength deformation twins:
This estimate suggests that Bauschinger back stresses can also be 0 1
very substantial in fcc alloys susceptible to deformation twinning if
B K C
accommodation stresses, rather than dislocation pileups, are s ¼ ð1 fT Þ$B
@s0 þ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

HP C
A þ fT $sT (19)
considered. Gil Sevillano and de las Cuevas also noted that strain t 1
compatibility between twins and the matrix requires the genera- b1 $ fT 1
tion of backward and forward internal stresses [49] in the spirit of
the two-phase models of Mughrabi [50,51]. Due to the equilibrium where fT ¼ b2 $bg3 $M$ε. In the above equations, KHP is the coefficient in
T
condition, the backward stress that develops in the softer austenite the Hall-Petch relation (257 MPa mm1/2), b1 is a numerical factor
matrix in which the twins are formed must be balanced by forward related to the shape of the twins (b1 z 1/2, for a high aspect ratio
stresses in the hard twins. This leads to a composite strain hard- twin), b2 is a numerical factor associated with the number of
ening effect, with the flow stress given by: deformation systems activated in the individual grains (b2 z 1/5,
assuming five deformation systems, consisting of four slip systems
sf ðεÞ ¼ ð1 FT ðεÞÞsf ;M ðεÞ þ FT ðεÞsf ;T ðεÞ and one twin system), b3 is a numerical factor related to the degree
pffiffiffiffiffiffiffiffiffi
¼ ð1 FT ðεÞÞaGb rðεÞ þ FT ðεÞsf ;T ðεÞ (14) to which the uniform deformation conditions are relaxed (b3 ¼ 1,
for full-constraints conditions), M is the Taylor orientation factor
Here sf,M and sf,T are the flow stresses in the matrix and in the (M ¼ 3.06), gT is the twinning shear (gT ¼ √2/2), and ε is the applied
twins, respectively. As sf ;M ðεÞ < sf ðεÞ < sf ;T ðεÞ, the internal stress in true plastic strain.
the matrix, si,M, and in the twins, si,T, are given by: Idrissi et al. [52] studied the structure of deformation twins in
two TWIP steels and reported that the deformation twins formed in
si;M ðεÞ ¼ sf ðεÞ sf ;M ðεÞ > 0 Fe-20%Mn-1.2%C steel were thinner (20 nm < tT < 70 nm) and
(15)
si;T ðεÞ ¼ sf ðεÞ sf ;T ðεÞ < 0 contained a much higher density of sessile dislocations as
That is to say, the back stress in the matrix is positive and acts compared to twins in carbon-free Fe-28%Mn-3.5%Si-2.8%Al steel
against the external stress, while the ‘back stress’ in the twins is (100 nm < tT < 700 nm). They argued that this observation, together
negative, i.e. it acts in the forward direction e the same direction as with the strain hardening rate being considerably higher (close to
the external stress. G/20) for the Fe-20%Mn-1.2%C TWIP steel, supported the strain
The internal stresses must satisfy the equilibrium condition: hardening model proposed by Gil Sevillano and de las Cuevas.
ð1 FT ðεÞÞsi;M þ FT ðεÞsi;T ¼ 0 (16)

2.2.4. Controversial issues relating to the role of deformation
The back stress results in an increase of the flow stress, thus twinning
contributing to strain hardening or to softening when the direction There are many observations which suggest that the relation
of the stress is reversed. In a Bauschinger experiment the back between deformation twinning and strain hardening may be more
stress is given by complex than the models outlined in the foregoing sections, which
emphasize the relation between the evolution of the density of
sf ;forward þ sys;backward deformation twins and the strain hardening. Most TWIP steels
sBS ¼ (17)
2 retain a large ductility when deformation twinning is restricted, i.e.
by a change in composition, grain refinement or an increase in
sf,forward (>0) and sys,backward (<0) are the forward flow stress and temperature. For example, two TWIP steels with the same gisf, viz.
the backward yield stress measured during the test, respectively.
Fe-30%Mn and Fe-22%Mn-0.6%C, develop a very different disloca-
While sBS can be measured experimentally, si,M and si,T cannot be
tion microstructure, with extensive twinning in Fe-30%Mn and no
obtained directly from back stress measurements.
twinning in Fe-22%Mn-0.6%C [53]. Both steels have a large ductility.
Nakano [54] suggested that this was related to the sign of the
2.2.3. Stress due to size effect parameterDkGTm, the dimensionless driving force for the g/ε
B
Gil Sevillano and de las Cuevas proposed an additional mecha- transformation, where DGm denotes the Gibbs free energy of the
nism for strengthening by deformation twins [44,49]. Namely, the transformation. Using an empirical correlation with the mechanical
deformation twins are considered to contribute to the strength of properties of high Mn steels, Nakano argued that this parameter
the twin-austenite aggregate due to their very small thickness (the must be positive, as in the case of Fe-22%Mn-0.6%C, to promote
deformation twinning. dislocation. Planar glide, which is known to increase strain hard-
Another case suggesting that deformation twinning may not ening, is therefore promoted in the systems where SRO occurs.
necessarily play a key role in the strain hardening of TWIP steels is It should be mentioned that a competition between the
shown in Fig. 10. It compares the flow curves of Fe-18%Mn-0.6%C- destruction of SRO by moving dislocations and its thermally acti-
1.5%Al TWIP steel at temperatures in the range from 70 C to vated restoration may lead to a negative strain rate sensitivity of the
300 C. Although in this temperature range gisf varies significantly, flow stress e an effect akin to dynamic strain aging effect discussed
from 17 mJ/m2 to 82 mJ/m2, the ductility of the material remains in the following section. Referred to as the pseudo-Portevin-Le
high, even when dynamic strain aging is pronounced, as can be Chatelier effect [59], it is associated with discontinuous plastic
concluded from the serrated deformation curves measured in the flow manifesting itself in serrations on the deformation curve.
high temperature part (100 C-300 C) of this temperature range. The distribution of the substitutional solutes Mn, Al, etc. in a
Deformation twinning is suppressed when gisf ¼ 45 mJ/m2, i.e. at TWIP steel is usually assumed to be fully random. The C distribution
108 C. This suggests that the deformation twins may not be the key is however influenced by the strong attractive C-Mn interaction
factor responsible for the high strain hardening rate observed for which leads to the formation of C-Mn octahedral complexes, and
TWIP steels. Other processes leading to a favorable dislocation short range order. The repulsive C-C interaction and the attractive
density evolution and large strain hardening have therefore been C-Al interaction also promotes SRO in Fe-Mn-C-Al TWIP steel by
proposed. These include short range ordering, dynamic strain aging formation of solute complexes with a structure similar to that of k-
due to the interaction between mobile dislocations and carbon carbide. Choi et al. [60] reported electron diffraction evidence for
atoms [55], pseudo-twinning, strengthening associated with the SRO in a Fe-28%Mn-9%Al-0.8%C steel. Park et al. [61], studied Fe-
lattice distortion due to a change of the position of an interstitial C 22%Mn-0.6%C (gisf ¼ 21.5mJ/m2), Fe-22%Mn-3%Al-0.6%C
atom from an octahedral to a tetrahedral site under twinning shear (gisf ¼ 36.5mJ/m2) and Fe-22%Mn-6%Al-0.6%C (gisf ¼ 50.7mJ/m2),
when the carbon atom resides in the glide plane [56], and stacking and Jin and Lee [62], who studied Fe-18%Mn-0.6%C (gisf ¼ 17mJ/m2),
fault formation. All of these effects will be considered in the Fe-19%Mn-1%Al-0.6%C (gisf ¼ 29mJ/m2) and Fe-19%Mn-2%Al-0.6%C
following sections. (gisf ¼ 36.3mJ/m2), analyzed the effect of Al-additions on the strain
hardening of TWIP steel. Park et al. [61] reported that planar
dislocation glide always occurred before mechanical twinning,
2.2.5. Short range ordering independently of the value of gisf. An increase of gisf by Al-additions
Gerold and Karnthaler [57] have argued that short range resulted in a more pronounced planar glide and less mechanical
ordering (SRO) and short range clustering, rather than the value of twinning. The authors argued that the pronounced planar glide was
gisf, control the occurrence of planar dislocation glide. According to due to short range ordering.
Fisher [58] short range order, i.e. a local deviation from randomness
in the arrangement of atoms on the crystal lattice, results in a strain
hardening effect. The mechanism is as follows. The first dislocation 2.2.6. Dynamic strain aging
on a glide plane experiences an increased resistance against its Dynamic Strain Aging (DSA) is observed in C-added TWIP steels
motion due to the SRO. As the SRO is destroyed by the lead dislo- at room temperature, i.e. at a temperature at which the diffusivities
cation, the resistance to the motion of subsequent dislocations on of solutes, including that of interstitial carbon, is extremely low and
the same glide plane is reduced, which is tantamount to softening. diffusion of C to temporarily arrested dislocations is virtually
This softening favors glide on the glide plane of the initial impossible, as the activation energy for C diffusion in g-Fe is
75
1200 1200 YS TEL
UTS 70 UEL
1100 c 1100 65
1000 1000 60
900 900 55
Engineering stress, MPa
Engineering strain, %
50
800 800
45
700 c 700
40
600 600 35
500 500 30
SFE, mJ/m2 25
400 -70ºC 35 400
-50ºC 37 20
300 -30ºC 40 300
+25ºC 49 15
200 +100ºC 65 200
10
+200ºC 86
100 +300ºC 108 100 5
0 0 0
0 10 20 30 40 50 60 70 -100 0 100 200 300 -100 0 100 200 300
Engineering strain, % Temperature, ºC Temperature, ºC

(a) (b) (c)
Fig. 10. (a) Engineering stress-strain curves of Fe-18%Mn-0.6%C-1.5%Al TWIP steel in the temperature range of 70 C to 300 C as measured at a strain rate of 103s1. (b)
Temperature dependence of the yield stress (YS) and the ultimate tensile stress (UTS). (c) Temperature dependence of the uniform elongation (UEL) and total elongation (TEL). Note
in (a) that the stress-strain curves in the range of 25 C to 300 C are serrated. The TWIP steel exhibits an inverse behavior for the occurrence of serrations at high temperature: a
critical strain (εc) of about 40% needs to be reached for the initiation of the serrations at 25 C, whereas lower onset strains are required at lower temperatures. At 300 C a critical
strain for the termination of serrated yielding is about 35%. It should be noted in (b) that the uniform elongation peaks at a temperature of 200 C. At this temperature gisf is
approximately 62 mJ/m2, and deformation twinning is definitely suppressed.
approximately 1.6 eV. Solute C in fcc Fe-alloys is known to have very variety of TWIP steels by Bouaziz et al. [35] the strain hardening
low diffusivity at room temperature and interaction of solute C with appeared to be independent of the stacking fault energy, an ob-
dissociated dislocations during room temperature deformation will servations very aptly referred to as the “carbon paradox” by Allain
therefore necessarily be of very short range nature. Even if local [66]. Bouaziz et al. [35] see this as an evidence for a mechanism
heating, e.g. due to increased strain rate within a localized defor- related to solute complexes playing a role in the strain hardening of
mation band, is taken into consideration, the bulk diffusion of C at TWIP steels. Based on a range of TWIP steel compositions, they
room temperature is too limited to allow for a DSA mechanism have proposed the following heuristic equation for the back stress
based on large diffusion distances. Rose and Glover [63] attribute contribution to the strain hardening due to the presence of the
DSA in fcc alloys to the presence of Cevacancy complexes and their deformation twins during uniaxial tensile deformation:
rapid reorientation in the stress field of dislocations. They point to
the low activation energy for DSA (83 kJ mol1, or 0.86 eV/atom) " 2 #
%C %C
and argue that reorientation is due to the movement of a vacancy sBS ¼ 60661$ 261874$ $ε1:75
%Mn 5 %Mn 5
rather than a C atom. Reorientation of a C atom in a tetragonally
distorted CFe6-xMnx solute complex by a single diffusional hop to (20)
an energetically more favorable neighboring octahedral position
According to this formula, the back stress is a maximum when
relative to the stress field of a partial dislocation or the lattice
%C , which the authors refer to as the “equivalent C
the ratio z ¼ %Mn5
contraction perpendicular to the stacking fault plane has been
proposed as a DSA mechanism [64,65]. DSA is only observed when content”, is equal to 0.016. In this case sBS ¼ 365 MPa at a true
the C atom reorientation time is shorter than the waiting time of a strain of 0.4. The equation suggests that an increase in the C content
dislocation at the location of the solute complex. Vacancy-C atom beyond this level of m would result in an increase of the strain
complexes could also cause DSA in TWIP steels, as plastic defor- hardening due to the back stress contribution of the deformation
twins. The lines corresponding to constant values of the %Mn5 %C ratio
mation results in the formation of excess vacancies. DSA associated
with such complexes could therefore be observed after a large in Fig. 11 are nearly perpendicular to the iso-gisfe curves, suggesting
amount of strain. This is however not observed in practice, as DSA is that strain hardening is gisf-independent. The effect of C appears to
initiated at very low strains in Fe-22%Mn-0.6%C. be a result of a complex interplay of multiple effects, involving
In the semi-empirical analysis of the strain hardening of a wide dynamic strain aging and the effect of C on dislocation mobility and
z=0.02 TOI=0.1
35 35
30 z=0.05 30
Manganese, wt-%
Manganese, wt-%
25 25
20 20
TOI=0.3
z=0.10
15 z=0.116 15
z=0.15
10 TOI=0.5
10
TOI=0.9
5 5
0 0.2 0.4 0.6 0.8 1.0 1.2 0 0.2 0.4 0.6 0.8 1.0 1.2
(a) Carbon, wt-% (b) Carbon, wt-%
4000 0.025
z: 0.240, TOI:0.55
z2
3500 z: 0.116, TOI:0.33

0.020 z: 0.043, TOI:0.15
z – 261874
3000 z: 0.008, TOI:0.03
2500 0.015
2000
1500 0.010
60661
1000
0.005
500
z: 0.116
0 0
0 0.05 0.10 0.15 0.20 0.25 0 1 2 3 4 5 6
(c) z=C/(Mn-5) (d) Mn atoms in CFe6-xMnx
Fig. 11. Iso-gisf C-Mn composition diagram for Fe-Mn-C showing the direction of increasing strain hardening in C-added high Mn TWIP steels (a) according to Allain et al. [71] and
(b) according to Saeed-Akbari et al. [26]. (c) Composition dependence of the m-parameter; and (d) the TOI-parameter dependence of the volume fraction distribution of the various
CFe6-xMnx complexes.
dislocation patterning. Saeed-Akbari et al. [26] arrived at a similar standard TWIP steels. Internal friction measurements and the
conclusion regarding the effect of C, which they considered to be occurrence of dynamic strain aging only in C-added TWIP steels,
related to the formation of C-Mn complexes. They also attempted to provide experimental support for the presence of CFe6-xMnx
relate the strain hardening with the density and distribution of the complexes.
different CFe6-xMnx octahedral complexes, as these are believed to Internal friction measurements are a powerful tool for identi-
interact with the dislocations. Saeed-Akbari et al. suggested that fying the nature of point defects and more complex groups of solute
the population of CFe6-xMnx octahedra is a function of the C/Mn atoms in TWIP steels. Thus, it was shown by Lee et al. [64] and Jung
atomic ratio, which they refer to as the theoretical ordering index et al. [65] that the Finkelstein-Rosin (FR) internal friction peak
TOI ¼ XXMn
C
. They reported that the highest density of CMn6 octahedra observed in Fe-18%Mn-0.6%C-x%Al TWIP steel (Fig. 12) can be
occurred when TOI was around 0.31. associated with the stress-induced reorientation of elastic dipoles
According to von Appen and Bronkowski [67], the CMn6 octa- formed by substitutionaleinterstitial atom complexes. The pres-
hedra are the most stable of the various possible CFe6-xMnx solute ence of point defect complexes involving C which may be the cause
complexes. Lee et al. [64] have used the Grujicic and Owen cell of DSA in C-alloyed TWIP steel revealed by internal friction mea-
surements was also confirmed by Mo €ssbauer spectroscopy [64].
model [68,69] to describe the population of CFe6-xMnx octahedra in
Fe-18%Mn-0.6%C-1.5%Al. According to that study, the CFe5Mn1 and This suggests that a pronounced DSA of TWIP steels in the absence
CFe4Mn2 octahedra are the most common solute complexes in of Al additions may result from the interaction of C-Mn complexes
0.0020 Fe-18%Mn-0.6%C 0.0020 Fe-18%Mn-0.6%C-1.5%Al

Internal Friction, Q-1
0.0015 Internal Friction, Q-1 0.0015
0.0010 0.0010
0.0005 0.0005
0.0000 0.0000
200 300 400 500 200 300 400 500
(a) Temperature, ºC (b) Temperature, ºC
0.0060 Fe-18%Mn-0.6%C-1.5%Al
0% Strain
10% Strain
0.0050
30% Strain
Internal Friction, Q-1
0.0040
0.0030
0.0020
0.0010
0.0000
0 100 200 300 400 500 600

(c) Temperature, ºC
Fig. 12. Internal friction spectrum of TWIP steels [65]. (a) FR relaxation peak in Fe-18%Mn-0.6%C; (b) FR relaxation peak in Fe-18%Mn-0.6%C-1.5%Al at the measurement frequency of
about 1 kHz; (c) Enhancement of the FR peak height for Fe-18%Mn-0.6%C-1.5%Al by straining.
with stacking faults. Indeed, the formation of stable C-Mn com- Deformation twinning in the presence of interstitial solutes is
plexes due to attractive interaction of C atoms with substitutional referred to as pseudo-twinning. In this case C atoms are transferred
Mn atoms leads to a symmetry of the local coordination shells from octahedral to tetrahedral sites. This mechanism was proposed
around an interstitial carbon atom which is lower than cubic. This to cause a strain hardening effect. Ab initio calculations show that a
effect gives rise to the formation of an elastic dipole. While an random distribution of C results in an increase of gisf [74], which
isolated interstitial atom in an fcc metal would not contribute to provides a strong thermodynamic driving force for C to leave the
anelastic relaxation because it occupies an octahedral interstitial stacking fault plane. This ab initio insight has made it possible to
site having the same symmetry as the host lattice, an elastic dipole understand some apparent inconsistencies in experimental obser-
would re-orient under external stress. In the case of fcc lattice of g- vations. Fig. 13 illustrates how dissociated dislocations incorporate
Fe alloys, in which substitutional-interstitial complexes are formed, C atoms in the glide plane during their crystallographic glide. This
a FR internal friction peak can be seen as a signature of the resulting results in a considerable local increase of the stacking fault energy.
elastic dipoles. The peak position at the temperature Tmax (at the As the C atoms are transferred to a tetrahedral position during the
frequency f ¼ 1 kHz) observed in steels is within the temperature passage of a stacking fault, they arrive at an intermediate high
range of 470e570 K, depending on the composition of the steel. The energy state for interstitial C diffusion. Accordingly, subsequent
activation energy of the FR peak observed in Fe-18%Mn-0.6%C and movement to an octahedral position away from the stacking fault
Fe-18%Mn-0.6%C-1.5%Al (1.2 eV and 1.39 eV, respectively [64,65]) is plane does not require the thermally activated overcoming of a high
close to the activation energy for carbon diffusion in alloyed energy barrier associated with this intermediate stage [75]. The
austenitic steels (1.166 eV for g-Fe-0.6%C [70]). The FR peak was passage of a dissociated dislocation will therefore clear its glide
very stable and was not influenced by annealing at high tempera- plane of all C atoms initially present there, and this will therefore
tures. It was also not affected by the addition of Al, suggesting that facilitate the glide of subsequent dislocations on the glide plane of
Al atoms are not part of the solute complex giving rise to the peak. this lead dislocation, including the glide on cross slip planes. This
The height of the FR peak in the TWIP steels studied was shown to results in an increase of the strain hardening due to a limited dy-
grow with strain, which suggests that the peak stems from the namic recovery and the attendant pronounced planar glide [56]
interaction between strain-induced dislocations and solute mentioned above. This strain hardening mechanism is dependent
complexes. on the stacking fault width, the dislocation velocity, and the ki-
A carbon atom with a shell consisting of one or two manganese netics of the re-orientation of the C atoms. The mechanism is also
atoms is thermodynamically more likely than any other octahedral expected to be both stress and temperature dependent. It should be
shell configuration in Fe-18%Mn-0.6%C-1.5%Al TWIP steel. It in- noted, however, that there is experimental evidence suggesting
volves an anisotropic (tetragonal) lattice distortion, which can only that the significance of the pseudo-twinning effect may be over-
be achieved when the solute C is part of a tetragonally distorted estimated. Indeed C-free TWIP steels studied by Gra €ssel et al. [3]
point defect complex such as the CFe5Mn1 or CFe4Mn2 octahedra. exhibit a high strain hardening in the absence of C.
The occurrence of serrations on the stress-strain curves also point
to dynamic strain aging associated with C-Mn complexes, as no
serrations were detected on the stress-strain curves of C-free Fe- 2.2.7. Stacking faults
Mn-Si-Al TWIP steels [3,19,71e74]. Perfect dislocations and partial dislocations emitted from grain
Fe Mn C
O 3
O 3 3
A A
T 2
B B 1 2 1 2
1
C A Initial
Intermediate
2.00
1.75
1.50
Energy, eV
1.25
1.00
a/6<112> 0.75
0.50 Re-oriented
0.25
0.00
0.0 0.1 0.2 0.3 0.4
Distance, nm
(a) (b) (c) (d)
Fig. 13. Schematic illustrating the reorientation of a carbon atom in a tetragonally distorted CFe6-xMnx complex during the passage of a partial dislocation. (a) Starting atom
arrangement for the CFe6-xMnx complex. (b) Lattice shear by a partial dislocation causes the carbon atom to be transferred from an octahedral interstitial position to a high energy
tetrahedral position. (c) Single thermally activated hop of the carbon atom to an energetically more favorable neighboring octahedral position out of the stacking fault plane. (d) Ab
initio calculation of the energy pathway for C atom diffusion in g-Fe at zero Kelvin temperature. The diagram illustrates that when an interstitial C atom is transferred from an
octahedral site to a near-tetrahedral site when it is situated in the stacking fault plane of a leading partial dislocation, no thermal energy needs to be supplied for the C atom to jump
to a nearby octahedral site away from the stacking fault plane [75].
Grain
boundary
Grain
boundary
0.2 m 0.2 m
(a) (b)
Fig. 14. (a) Stacking faults emitted from grain boundary defect sites (black arrows) in Fe-15%Mn-0.6%C (gisfe ¼ 14 mJ/m2) TWIP steel. (b) Perfect screw dislocation pileup emitted by
a grain boundary defect sites (black arrow) in Fe-15%Mn-0.6%C-1.5%Al (gisfe ¼ 27 mJ/m2).
boundary defects and trailing stacking faults are commonly Wilsdorf [81], a is equal to 0.3 “within a factor of two”.
observed in strained TWIP steels (Fig. 14). Isolated stacking faults Kuhlmann-Wilsdorf [30] derived the following equation for a for
are known to act as obstacles to dislocations, and they have been dislocation-cell forming metals:
reported to contribute to the strain hardening in low stacking fault
energy metals and alloys [76]. Although the strain hardening effect
1 2n 3
of stacking faults per se has not yet been analyzed in depth in a¼ pffiffiffiffiffi (21)
b$ rd
$ln
6$p$ð1 nÞ
relation to TWIP steels, it is worthy of note that Meyers et al. [77]
and Tian et al. [76] recently proposed that the significant stacking Assuming v ¼ 0.312, b ¼ 0.25 nm and a value of rd of 1013m2 or
fault contribution to strain hardening in Cu-15%Al (gisfe ¼ 7 mJ/m2) 1015m2, one obtains a ¼ 0.54 or a ¼ 0.39, respectively. These value
observed from low strains on, may also apply to TWIP steels. They are close to a ¼ 0.4 used by Allain et al. [34] for Fe-22%Mn-0.6%C
report that while deformation twinning is grain size dependent, TWIP steel, and a ¼ 0.35 used by Kim et al. [18] for Fe-18%Mn-0.6%
twinning being suppressed at small grain size, the strain hardening C-1.5%Al TWIP steel.
effect of stacking faults is independent of the grain size. Fig. 15 il- While constitutive equations that capture the effect of defor-
lustrates the ubiquitous presence of stacking faults in TWIP steels, mation twinning on the mechanical properties of TWIP steels in a
even at low strains. It also shows that they appear to be emitted semi-phenomenological way are in existence and will be discussed
from grain boundaries. Thus, Fig. 15 (a), illustrates the emission of a at length below, the models developed so far at the level of
high density of stacking faults from a grain boundary in Fe-18%Mn- dislocation-twin interactions are scarce. They consider only a spe-
0.6%C-1.5%Al TWIP steel at 2% engineering strain. A similar mech- cial case when deformation twinning generates S3{111}〈110〉-type
anism is also evident in Fig. 14 (a). Clear observations of both coherent twin boundaries (CTBs) which act as impenetrable ob-
stacking faults and twins being emitted form grain boundaries have stacles to dislocations gliding on their slip planes not parallel to the
led Beladi and co-workers [78] to conduct an in-depth analysis of twin plane. The influence of the dislocation-CTB interaction itself is
the possible relation between the formation of deformation twins ignored. It is well known that dissociated dislocations which
and the boundary characteristics of the grains at which the twins encounter a coherent twin boundary are stopped by the boundary,
are formed. They report that the 60 {111} annealing twin bound- as illustrated in Fig. 16. As the applied stress increases, the two
aries were the most likely boundaries to emit deformation twins. partial dislocations recombine to a perfect dislocation and the
This is rather unexpected as 60 {111} twin boundaries are low dislocation can either be incorporated in the twin plane or be
energy boundaries. transmitted through the CTB to form a dissociated screw disloca-
To close this section of the relation between deformation tion in the twin [82e84]. Large scale molecular dynamics (MD)
twinning and strain hardening in TWIP steel, it is important to models for the interaction between screw dislocations and CTBs
address a number of issues related to the physically-based have shown that the spontaneous absorption of the dislocation into
modeling of the mechanical properties of TWIP steels. the CTB is possible when gisf is high, as in the case of Al. When gisf is
A common problem with the physically-based constitutive low, this process requires that a dislocation constriction occurs first.
equations is the identification of the values of the physical pa- This can result in the absorption of the dislocation into the CTB or
rameters used in the model. An example is the parameter a in the transmission into the twin. Sangid et al. [85] have shown by MD
Bailey-Hirsch equation [79]. This parameter enters the relation simulations and experimental observations that the type of
between the flow stress and the square root of the dislocation coherent S ¼ 3 twin boundaries created during deformation
density. An overview written by Kang et al. [80] compiles published twinning act as strong obstacles to slip, resulting in a considerable
values for a ranging from 0.1 to 0.5. According to Kuhlmann- strengthening contribution. Dislocations eventually slip into the
between dislocations and deformation twin boundaries in TWIP

0.2 m steels. The authors report that pileups of extended primary dislo-
Grain cations in a Fe-20%Mn-1.2%C TWIP steel impinging on CTBs did not
transmit slip into the twins. Neither did they observe transmission
boundary of primary screw dislocations into twinned regions by a process
which involves the constriction of a dislocation at the CTB and its
cross slip into the twin, where it re-dissociates. Extended primary
dislocations were, however, found to be incorporated into the CTB
g plane after recombination of the partial dislocations. The perfect
dislocations then nucleated a sessile Frank partial dislocation and a
Shockley partial dislocation in the CTB plane (Fig. 17). These ob-
(a) servations provide support to some of the MD simulation results
reported by Jin et al. [83] for the interaction of non-screw dislo-
0.1 m cations with CTBs. Idrissi and Schryvers argue that the large
number of Frank partial dislocations observed in their TEM analysis
of CTBs prevent twin thickening and increase the strain hardening
by impeding the interaction between the subsequent dislocations
and the CTB.
An additional limitation of the mentioned physically-based
constitutive models is that they do not take into account the sur-
prisingly high rate of dislocation multiplication and the activation
of multiple glide systems at low strains observed experimentally in
<110> ZA TWIP steels. An example is shown in Fig. 18 for a single crystal of a
TWIP steel deformed in stage I (easy glide), as documented by the
(b) slip lines observed on the sample surface. We note, however, that
TEM observations revealed a high density of dislocations in two
0.1 m non co-planar slip systems, which puts in question the identifica-
tion of strain hardening as being associated with easy glide.
3. The energy of planar faults in a TWIP steel
3.1. Intrinsic vs. extrinsic stacking faults
There are relatively few publications which have addressed the

<111> ZA important experimental verification of the character of stacking
faults occurring in TWIP steels. Despite the fact that the intrinsic
(c) and extrinsic stacking faults of fcc metals and alloys are believed to
Fig. 15. Illustration of the prevalence of isolated stacking faults in the deformed
be very similar [87] extrinsic stacking faults are not often observed
microstructure of Fe-18%Mn-0.6%C-1.5%Al TWIP steel. (a) Overlapping stacking faults in practice, and a stacking fault in fcc metals and alloys is usually
emitted by a grain boundary at low strains. (b) Stacking faults which appear to be assumed to be intrinsic. Vercammen et al. [88] were the first to
associated with the end point of deformation twins at intermediate strains. (c) Regular report that the stacking faults in C-free Fe-30%Mn-3%Al-3%Si TWIP-
pattern of interacting stacking faults at high strains.
steel were extrinsic. Idrissi et al. [89] observed that in C-free Fe-
19.7%Mn-3.1%Al-2.9%Si TWIP steel the character of the stacking
twin when the imposed load is substantially increased, or when fault changed with temperature: the stacking faults were extrinsic
dislocation pileups create a significant local amplification of the at low temperature and intrinsic at high temperature. Whereas
imposed stress. Transmission of slip is always associated with the deformation at room temperature resulted in the strain-induced
formation of dislocations at the twin boundary. To the best of the transformation to ε-martensite and the occurrence of stacking
authors' knowledge, the TEM observations reported by Idrissi and faults which were predominantly extrinsic in nature, deformation
Schryvers [86] are the only ones which elucidate the interaction at high temperature (160 C) resulted in the formation of
g g g
(a) (b) (c)

Fig. 16. Transmission electron micrographs showing a pileup of sixteen identical a/6<112>{111}-type partial dislocations impinging on a deformation twin. (a) Partial dislocation
contrast. (b) Stacking fault contrast. (c) Tilted view of a deformation twin showing evidence of the incident slip dislocation pileup generating a dislocation in the twin (arrow A),
dislocations crossing the twin between the two coherent twin planes (arrow B), and an array of straight boundary dislocations in the twin/matrix interface plane (arrow C).
CTB (K1 /conjugate plane)
Primary slip (K2) plane
D D
A D
A
A
B
D DA
DA
DA A D
C
A
No slip transmission Twin Matrix

on K3 plane
Fig. 17. Schematic of the dislocation/CTB interaction observed by Idrissi and Schryvers [86]. Multiple similar interactions result in a large density of sessile Frank partial dislocations
located in the twin boundary.
deformation twins and stacking faults, which were all intrinsic. The nucleated at extrinsic stacking faults. This viewpoint is at odds with
authors suggested that at low temperatures ε-martensite is the commonly accepted view that ε-martensite is formed by a
specific sequence, i.e. the passage of one partial dislocation trailing
an intrinsic stracking fault on every other {111} slip plane. Tisone
has suggested that this may be due to a difference in the temper-
ature dependence of gesf and gisf [90]. Idrissi et al. [91] also claimed,
using the analysis of the stacking fault fringe contrast in trans-
mission electron microcopy, that the stacking faults used to
determine gisf for C-alloyed Fe-20%Mn-1.2%C TWIP steel
(gisf ¼ 15 mJ/m2) were all intrinsic. Karaman et al. [92] observed
extrinsic stacking faults in 12.3%Mn-1.1%C Hadfield steel single
(a)
crystals (gesfe ¼ 23 mJ/m2). Mahato et al. [19] analyzed dissociated
dislocations in Fe-27%Mn-2.5%Si-3.5%Al TWIP steel. These authors
frequently observed extrinsic stacking faults, which they consid-
ered to be three-layer twin embryos. Pierce et al. [93] used the
diffraction contrast method to determine the character of the
stacking faults in the Fe-22%Mn-3%Al-3%Si, 25%Mn-3%Al-3%Si, and
28%Mn-3%Al-3%Si alloys, and identified the stacking faults as being
(b)
intrinsic in all cases.
3.2. Stacking fault energy computations
Classical texts on the mechanical behavior of fcc metals and

alloys consider that the intrinsic stacking fault energy, gisf, is the
(c) (d) key parameter controlling the emergence of plasticity mechanisms
Fig. 18. (a) TWIP steel single crystal after 4% strain under compression in the 〈110〉
other than dislocation glide. gisf is considered to be the parameter
direction. The parallel slip traces at the crystal surface, indicative of planar dislocation which controls whether plasticity is carried by glide of perfect
glide, are clearly visible. (b) Transmission electron micrographs using {220}-type re- dislocations, dissociated dislocation glide, or deformation twinning
flections showing the presence of long edge-type glide dislocations (white arrows) in resulting from the motion of partial dislocations trailing stacking
the macroscopic primary slip plane of the sample shown in (a). The micrographs also
faults on parallel glide planes. A low energy of intrinsic stacking
reveal the presence of forest dislocations on an intersecting glide plane. (c) and (d) are
views of the same area as in (b) after tilting the sample from the 〈111〉 zone axis to two faults (gisf < 15 mJ/m2) results in stress- or strain-induced solid-
different 〈112〉 zone axes. The micrograph shows that the shorter dislocation segments solid phase transformations (TRIP-effect), an intermediate range of
are forest dislocations which intersect the primary glide plane. Note the absence of gisf (15 mJ/m2 < gisf < 45 mJ/m2) is associated with mechanical
dislocation pileups. As the sample was strained to just 4%, i.e. in the easy glide stage twinning, and a high level of gisf (gisf > 45 mJ/m2) leads to perfect
(Stage I hardening), the micrographs are also evidence for multiple slip in the early
stages of deformation, despite the fact that the macroscopic single crystal appears to
dislocation glide and shear banding. The temperature plays a key
deform in single slip. (Image and micrographs courtesy of Dr.-Ing. Stephanie role in the relation between gisf, deformation twinning, and plas-
Sandlo€ bes, RWTH Aachen). ticity of TWIP steels, as the magnitude of gisf increases with
gisf ¼ n$r$DGgC/ε
g/ε g/ε
dgisf d DHC T$DSC g/ε 8 1 g/ε (22)
¼ n$r$ ¼ n$r$DSC ¼ pffiffiffi$ 2 $DSC
dT dT 3 ag ,NA
temperature. This can be shown by considering Hirth's definition of linked to the magnitude of gisf [98,99]. Although these observations
an intrinsic stacking fault as a thin plate of the hcp ε-phase [94]: are largely empirical, they have contributed to the wide acceptance
where ag is the lattice parameter of the g-phase, NA is Avogadro's of Venables' opinion on the subject and it is therefore not surprising
number, and DGgC/ε, DHgC/ε, and DSgC/ε are, respectively, the Gibbs that major efforts have been made to calculate the temperature and
free energy, enthalpy, and entropy changes associated with the composition dependence of gisf for TWIP steels in order to predict
g/ε transformation. For the case when the g-phase is the ther- their mechanical properties. The empirically determined gisf-
modynamically stable phase, the inequalities DGgC/ε > 0, DSgC/ε < 0, ranges in which specific deformation modes are activated, are
and dg/dT > 0 hold. At high temperatures, perfect dislocation glide shown in Fig. 19 [34,82,97,101].
with frequent cross-slip is expected when gisf becomes larger than There are two basic computational approaches to determine gisf:
45 mJ/m2. At lower temperatures, when 15 mJ/m2 < gisf < 45 mJ/m2, the thermodynamic approach based on the Olson-Cohen model for
large dislocation dissociation widths lead to the nucleation of twins the intrinsic stacking fault in fcc metals and alloys [102] and the ab
and their growth by thickening. As the temperature is further initio approach based on the determination of key intrinsic energy
reduced, ε-martensite formation gradually replaces deformation barriers in the generalized stacking fault energy surface.
twinning. The analysis of data on a low-gisf single crystal alloy
prompted Venables [95] to assume a parabolic relationship be- 3.2.1. Thermodynamic approach
tween the twinning stress and gisf - despite a considerable scatter in In the thermodynamic approach, the intrinsic stacking fault
the available data. Remy and Pineau [96,97] also emphasized the energy is computed by means of the CALPHAD method used in
role of gisf by arguing that the occurrence of deformation twinning modeling the equilibrium thermodynamic properties of multi-
required gisf to be within a specific range: 20 mJ/m2 < gisf < 50 mJ/ component alloys. The CALPHAD method employs experimental
m2. In the case of high Mn TWIP steels, various gisf-ranges have and theoretical information to obtain a complete thermodynamic
been proposed as the ones in which the TWIP mechanism is acti- description of the phase equilibria and the thermo-chemical
vated: 12e35 mJ/m2 [71], 25e60 mJ/m2 [96], 25e80 mJ/m2 [97]. properties of alloy systems. This thermodynamic description is
The occurrence of strain-induced transformations has also been provided in terms of the Gibbs free energy of each phase, expressed
Twins+
Twinning Dislocation cells

Fe-20-32Mn-Cr-C alloys Transformation
Remy and Pineau
isf
1977 0 10 20 30 40 50 60 70 80 90 100
Twins+
Twinning Dislocation cells

Co-15-45Ni-14Cr-Mo alloys Transformation
Remy
isf
1981 0 10 20 30 40 50 60 70 80 90 100
Twins+
Twinning
Fe-Mn-C alloys Transformation
Allain et al.
isf
2004 0 10 20 30 40 50 60 70 80 90 100
Twins+
’ Twinning
Fe-18Cr-10Mn-N-C alloys Transformation
Lee et al.
isf
2010 0 10 20 30 40 50 60 70 80 90 100
Fig. 19. Ranges of the intrinsic stacking fault energy gisf for phase transformation, deformation twinning, and dislocation glide in fcc g-Fe alloys reported in the literature for Fe-(20-
32)Mn-Cr-C alloys [96], Co-15-45Ni-14Cr-Mo alloys [82], Fe-Mn-C TWIP steels [34], and Fe-18Cr-10Ni-C-N alloys [101]. The relation between the gisf axis and DGg/ε is also indicated,
assuming that 2 s g/ε ¼ 15 mJ/m2.
as a function of composition and temperature. Extrapolation or transformation is small, approximately 70 J/mol for Fe-17%Mn, and
interpolation makes it possible to compute thermodynamic prop- it is not dependent on the Mn content [103]. The addition of C
erties when no experimental data is available. This approach has strongly decreases Aε/ g and Mg/ε. C also increases the driving
s s
been used for all the thermodynamic calculations of gisf of high Mn force to approximately 300 J/mol for a 0.6 mass-% addition of C to
TWIP steels by employing the sub-regular solution model Fe-17%Mn, indicating that C is a strong g-stabilizing alloying
approximation. In the thermodynamic approach, gisf is calculated addition.
based on the equilibrium thermodynamic formalism originally The lattice parameter is influenced by the composition
proposed by Olson and Cohen [102], which separates the stacking [104e108]. Fig. 20 shows the effect of C, Mn, Al, and Si on the lattice
fault formation energy into contributions from the Gibbs free en- parameter of TWIP steels. Fig. 20 (a) includes room temperature
ergy difference between the hexagonal close-packed (hcp) and and low alloy content extrapolations for g-Fe-C and g-Fe-Mn binary
face-centered cubic (fcc) phases, based on Hirth's definition of an alloys.
intrinsic stacking fault as a thin plate of the hcp ε-phase [94], cf. Eq. Equation (23) was further developed to include magnetism,
(22), as extended by Olson and Cohen [102] to include the g/ε strain, and grain size effects:
interface energy:
4
4 g/ε
g∞
isf ¼ 2$pffiffiffi $DGg/ε þ 2$sg=ε
gisf ¼ 2$pffiffiffi $DGC þ 2$sg=ε (23) 3$a2 $NA
3$a2 $NA
4 g/ε g/ε g/ε
g∞
isf ¼ 2$pffiffiffi $ DGC þ DGMag þ ES þ DGex þ 2$sg=ε
Here, DGgC/ε is the chemical driving force for the g/ε trans- 3$a2 $NA
formation, i.e the difference between the Gibbs free energy of the
(25)
g-phase and the ε-phase. s g/ε is the g/ε interface energy. The factor
2 in the first term on the right-hand side of the equation is the Both the g and the ε phase undergo an anti-ferromagnetic-to-
/ε
number of close-packed planes of the hcp ε-phase. The equation is paramagnetic transformation. DGgMag is the Inden-Hillert-Jarl
for an ideal stacking fault, i.e. it does not take into account the effect description of the difference of the magnetic contributions to the
of partial dislocations, the change in the atomic density in the Gibbs free energy of the two phases [109,110]. EgS /ε is the strain
stacking fault plane and the possible effect of an externally imposed energy resulting from the shear deformation and the lattice
stress on the Gibbs free energy balance. The molar surface density r compression due to the difference in atomic density between the g
in the {111} plane (in moles per unit area) is taken as and the ε phase. This strain increases the energy of the stacking
fault region. This term was analyzed in detail by Olson and Cohen
4 [102], as well as by Cotes et al. [111] and Müllner et al. [112]. Various
r ¼ pffiffiffi (24)
methods, usually based on the Eshelby inclusion theory, have been
3$a2g $NA
proposed to calculate EgS /ε. These are listed in Table 1. The equa-
The lattice constant ag of TWIP steels typically equals about tions generally predict that the strain contribution is small, and
0.36 nm. The temperature and composition dependence of the independent of the Mn content. Note that the strain energy equa-
Gibbs free energy for the g-phase can easily be determined using tions do not take into account the elastic strain associated with the
the CALPHAD approach. This is not the case for the ε-phase. For the partial dislocations involved in the formation of the stacking fault
Fe-Mn system ε-martensite formation data is used, based on the during plastic deformation.
assumption that, above the Ne el temperature, the free energies of The grain size is expected to influence dislocation glide, stacking
the g-phase and the ε-phase are equal at the temperature T0, which fault formation and deformation twinning, but there are also re-
is located midway between the Aε/ g and Mg/ε transformation ports of an unexpected influence of the grain size on gisf. Volosevich
s s
temperatures. The driving force for the g/ε martensitic et al. [113] were the first to report an effect of the grain size on the
0.3620
Al 0.3680
Mn
Lattice parameter, nm
Lattice parameter, nm
0.3610 0.3660 -Fe-xMn

Si
0.3600 0.3640
C -Fe-xC
0.3590 0.3620
-Fe-xMn
0.3580 0.3600
4 4
Si ( nm / at %) 1 10 Mn (nm / at %) 1 10
4
0.3570 Al (nm / at %) 3 10 0.3580
4
C ( nm / at %) 9 10
0.3560 0.3560
0 1 2 3 4 5 6 7 8 9 10 11 12 10 20 30 40 50 60 70 80 90
Atomic-% Atomic-% Mn
(a) (b)
Fig. 20. (a) Effect of solute C [104], Al [105] and Si [106] on the lattice parameter of g-Fe-Mn alloys at room temperature. The data for the binary g-Fe-C and g-Fe-Mn alloys are
extrapolations from high temperature and high Mn content, respectively [106]. (b) Effect of the Mn content on the lattice parameter of binary g-Fe-Mn alloys at room temperature
[106,108].
Table 1
Review of the proposed methods for calculating the contribution of the free energy of the g/ε transformation to the stacking fault strain energy. Here Vfcc hcp
m , Vm are the molar
volumes of the fcc and hcp phases, respectively; the Poisson's ratio n equals 0.333, and the shear modulus G equals 74 GPa. Lattice contraction normal to the stacking fault plane
was estimated as ε33~0.01 [102] and ε33~0.02 [111]. NA is the Avogadro number, and cfcc is the c-axis lattice parameter of the conventional hexagonal unit cell of the fcc crystal
structure.
g/ε g/ε
Shear: Eshear Dilatation: Edilatation
Olson-Cohen Spherical [102] g/ε 75n 2 fcc 2 !2

Eshear ¼ 15,ð1nÞ,3,G,Vm ,ε33 g/ε 2,ð1nÞ fcc Vmhcp Vmfcc
Edilatation ¼ 9,ð1þ nÞ,G,Vm , Vmfcc
pffiffiffi
Cotes et al. Spherical [111] Olson-Cohen method, with hcp 3 hcp 2 hcp
fcc hcp Vm ¼ ða Þ c ,NA
c c Olson-Cohen method, with 4
ε33 ¼
cfcc fcc
Vm ¼ ðafcc Þ3 ,NA
fcc 2,afcc
c ¼ pffiffiffi
3
Müllner-Ferreira Isotropic theory Platelet [112] g/ε G,ε2
ES ¼ 2,df111g ,
4,ð1 nÞ
afcc
df111g ¼ pffiffiffi
3
stacking fault energy. They explained their observations for the metastable austenitic Fe alloys because gisf can still be positive
grain size dependence of gisf of Fe-17.8%Mn-0.47%C steel by (i) a when DGg/ε is negative, due to the relatively large contribution of
higher solute C content obtained in small grained samples due to the interface energy term 2 sg/ε.
the thermal processing of their materials which involved a Geissler et al. [121] have pointed out that ambiguities in the
quenching step, and (ii) the presence of grain size dependent in- Olson-Cohen formalism produce contradictions. According to
ternal stresses which caused a change in the dislocation dissocia- them, the inclusion of a major interface energy term in the model is
tion width. Takaki et al. [114], and Lee and Choi [115] proposed to inconsistent with the requirement that the interface energy should
take the grain size effect into account by considering the influence be three orders of magnitude smaller than gisf, i.e. approximately
of the grain size on the g/ε transformation start temperature 0.5 mJ/m2. They conclude that the use of sg/ε leads to an over-
through a grain size-dependent excess free energy term: estimate of gisf and that it can only be considered as a correction
term without a direct physical meaning. Li et al. [122] also ques-
dðmmÞ tioned the physical meaning of sg/ε and regarded it as a correction
DGGS
ex ðJ=molÞ ¼ 170:06$e
18:55 (26)
term. According to them, the contribution of the interface energy to
Such an extra term in the Gibbs free energy was introduced sg/ε is small, in the range of 1e2 mJ/m2. In their view, sg/ε is not
originally by Jun and Choi [116] to take into account the effect of the interface energy related, but is rather a correction term of a volu-
grain size on gisf for Fe-Mn alloys, and it has been used by others, metric origin. It stems from the need to calculate the free energy
too [117,118]. The grain size effect on the transformation kinetics difference DGg/ε for the same interatomic distance in the g and ε
can be rationalized as follows. When the kinetics are controlled by phase and not for the equilibrium interatomic distance for each
the autocatalytic propagation of the transformation from one grain bulk phase at equilibrium.
to the next, small grain size will inhibit the transformation in the The ab initio calculations of sg/ε reported by Lee et al. [123] for
same way the Hall-Petch effect inhibits propagation of plastic pure g-Fe at zero Kelvin indicate a very large negative value
deformation from grain to grain. While this explanation is plausible of 241 mJ/m2. This value is however not inconsistent with the
conceptually, the grain size effect remains poorly understood from value of gisf of 380 mJ/m2 obtained by ab initio calculations for
a thermodynamic viewpoint. pure g-Fe at 0 K, if one considers the well-known approximations
The interfacial energy sg/ε is also a poorly defined parameter and gisf z 2 gtwin and sg/ε z gtwin.
it is often adjusted to reach better agreement with the experi- Talonen and Ha €nninen [124] made an interesting observation
mental data. Reported sg/ε values for Fe-Mn alloys are in the range related to the equilibrium width of overlapping stacking faults in
of 5e30 mJ/m2 [111,113,118]. Pierce et al. reported values for sg/ε in stress-free conditions, which also illustrates the importance of the
the range of 15e33 mJm-2 for Fe-Mn alloys, and 8e12 mJm2 for quantity sg/ε in low gisf situations. Neglecting the interaction be-
TWIP steels [93]. There is no experimental technique to determine tween the partial dislocations, they derive the following equation
sg/ε directly. Theoretical values from ab initio calculations are not for the dependence of the equilibrium width w on the number of
yet available either, although the interface energy term sg/ε should overlapping intrinsic stacking faults N:
be approximately similar to gtwin, the energy of a coherent twin.
This has resulted in a fundamental uncertainty in the value of gisf
N$G$b2P 3 1
based on the thermodynamic approach. A value of 10 mJ/m2 ap- w¼ $ (28)
2$p$ N$ gisf 2$sg=ε þ 2$sg=ε 4 4$ð1 nÞ
pears to be a reasonable choice in the case of Fe-Mn-C and Fe-Mn-
Al-C TWIP steels [119,120]. Pierce et al. [93] discussed the possi-
Here bP denotes the magnitude of the Burgers vector of partial
bility of determining sg/ε experimentally, through measurements of
dislocations. When the volume contribution to gisf is lower than the
gisf by X-ray diffraction or high resolution electron microscopy
surface energy contribution 2 sg/ε, the overlap of additional fault
techniques. Once the experimental value of gisf is known, sg/ε can be
planes causes w to diverge. Talonen and H€ anninen [124] consider
calculated by inverting Eq. (25):
that the formation of ε-martensite, i.e. the overlapping of intrinsic
stacking faults on every second {111}-type plane is a special case of
1
sg=ε ¼ $ gisf 2$ DGgC/ε þ DGgMag
/ε g/ε
þ ES þ DGGS
ex (27) this situation. They also consider a set of N dissociated dislocations
2 forming overlapping intrinsic stacking faults on successive planes,
Fig. 18 shows that, in the Olson-Cohen model for gisf, the i.e. a twin, and derive the following equation for the dependence of
deformation-induced g/ε transformation is possible in the twin width w on N in equilibrium:
(1.6e4.1)%Al-0.08%C TWIP steel. Nakano and Jacques [128] pro-

posed an alternative model for the stacking fault energy calcu-
N$G$b2P 3 1
w¼ (29) lations for Fe-Mn and Fe-Mn-C alloys after a thorough re-
4$p$sg=ε 4 4,ð1 nÞ
$
evaluation of the available thermodynamic parameters. Using
The equation implies that, if the interaction between the partial improved parameters for the Mn-C and Fe-Mn-C systems pro-
dislocations is ignored, the energy of the twin is not twin volume vided by Djurovic et al. [129,130], Pierce et al. evaluated the
dependent and that it has a constant value of 2 sg/ε determined stacking fault energy for Fe-Mn-Al-Si-C alloys [93]. Their model
by the surface energy. It also implies that when the volume correctly predicts the experimentally observed influence of Si on
contribution to gisf is higher than the surface energy contribution the stacking fault energy. They carried out an in-depth deter-
2 sg/ε, the overlap of additional fault planes causes w to increase mination of the value of the interfacial energy for various high
very gradually. Fig. 21 illustrates the fundamental difference in the Mn TWIP steels, which they report to be in the range of
equilibrium widening behavior in the case of overlapping stacking 8.6e11.8 mJ/m2 for Fe-18%Mn-0.6%C-(0e1.5)%Al/Si TWIP steel.
faults and twin formation. They also found a strong correlation between the interfacial
Two points are worth mentioning. First, it is important to energy sg/ε and the difference in free energy between the fcc g-
realize that the CALPHAD method relies on the thermodynamic phase and hcp ε-phase, 2 r (DGgC /ε þ DGgMag /ε
).
parameters in the databases used. We note specifically that the Table 2 provides a survey of the main aspects of these three key
expression for the excess energy difference between the fcc and studies, and Fig. 22 compares some of their results. In the figures
hcp phase in the Fe-Mn system in the TCFE7 database differs the line for alloy compositions with Mgs /ε ¼ RT (where R is the gas
from that proposed elsewhere in the literature. Second, the Olson constant and T the absolute temperature) is also indicated [131].
and Cohen thermodynamic formalism has been claimed to be This line delineates the limits for thermal ε-martensite formation.
ambiguous, especially in relation to the use of large sg/ε values Olson and Cohen have suggested that gisf is close to zero at the
[121]. Still, this approach has been used to model the composi- Mgs /ε temperature [102]. The composition dependence of Mgs /ε(K),
tion and temperature dependence of gisf with considerable suc- the temperature of the martensitic g/ε transformation, is given by
cess. A few notable examples are as follows. A model developed a heuristic relation [131]:
by Allain et al. gives the composition and temperature depen-
dence of gisf for Fe-(10e35)%Mn-(0e1.2)%C TWIP steels [71]. The g/ε
Ms ðKÞ ¼ 576 489%C 9:1% Mn þ 21:3Al 17:6% Ni
model due to Dumay et al. focusses on the effect of alloying
9:2% Cr þ 4:1% Si 19:4 % Mo 1% Co
additions of Al, Cr, Cu and Si on the stacking fault energy of the
Fe-22%Mn-0.6%C (gisf ¼ 23 mJ/m2) TWIP steel [125]. The model 41:3 %Cu 50 % Nb 86 %Ti 34 % V 13% W
developed by Saeed-Akbari [117] addresses the Fe-Mn-C and Fe-
Mn-Al-C TWIP steels. While the previous models were developed Here the concentrations are in mass-%. A comparison between
explicitly for high Mn TWIP steels, that by Curtze et al. [126] the experimental and the calculated values of gisf is shown in
applies to a wide range of complex austenitic Fe-Cr-Ni-Mn-Al- Fig. 23. The figure demonstrates a reasonable agreement between
Si-Cu-C-N alloys. Curtze and Kuokkala [127] have also applied the measured gisf and the value predicted by the thermodynamic
their model specifically to low-C, high-Mn Fe-(25e28)%Mn- model of Saeed-Akbari et al. [117], assuming an interfacial energy
Transformation
0 10 20 30 40 50 60
2 =15mJ/m2 isf
-20 -10 0 10 20 30 40 G
isf : 13 mJ/m 14 mJ/m
2 2
1000 1000
Width of overlapping
Width of overlapping
stacking faults, nm
stacking faults, nm
15 mJ/m2 isf : 13 mJ/m

2
100 100
20 mJ/m2
10 10
30 mJ/m2
2 : 15 mJ/m2
1 1
0 5 10 15 20 25 0 5 10 15 20 25
Number of overlapping stacking faults Number of overlapping stacking faults
(a) (b)
Fig. 21. Effect of the intrinsic stacking fault energy gisf on the equilibrium width of faults consisting of stacking faults overlapping on every second {111} plane for 2$sg/ε ¼ 15 mJ/m2
[124]. (a) The width diverges when gisf < 15 mJ/m2. (b) The width increases gradually with N when gisf > 15 mJ/m2. The case of a twin, gisf ¼ 2$sg/ε, corresponds to the red line in (a)
and (b).
Table 2
Comparison of CALPHAD-based models used to determine the magnitude of gisf of TWIP steels.
Reference TWIP Steel Alloy System Strain energy term Grain size excess term Surface energy, mJ/m2 Database employed
Allain et al. [71] Fe-Mn-C Not Included Not Included 9 SGTE

Dumay et al. [125] Fe-Mn-C-Al-Cr-Cu-Si Not Included Not included 8 Literature
0 < mass-% alloying < 8
Saeed-Akbari et al. [117] Fe-Mn-C-Al Included Included 15 Literature
Mn: 10e30 mass-% CALPHAD
C: 0e1.2 mass-%
Al: 0e7 mass-%
Curtze and Kuokkala [127] Fe-Mn-Al-Si-C Included Not included 8 Literature
SGTE
Nakano and Jacques [128] Fe-Mn-C Included Not included 16 Literature
CALPHAD
T0 data
Pierce et al. [93] Fe-Mn-Al-Si-C Included Not included 8e12 Literature
CALPHAD
of 15 mJ/m2. The figure also shows that the TWIP effect has been It is recognized that the magnitude of a single parameter, gisf,
observed for various TWIP steels within the measured gisf-range cannot be sufficient to predict the dominant plastic deformation
of 15 mJ/m2 to 45 mJ/m2. The data suggests that when mode of low and medium stacking fault energy metals and alloys.
gisf z 15 mJ/m2, DGg/ε is close to zero, as expected based on the The interpretation of mechanical twinning requires the computa-
Olson-Cohen work [102]. tion of intrinsic energy barriers which control the twinning process,
Fig. 22. Comparison of stacking fault energy calculations in C-Mn composition diagrams, as redrawn from (a) Allain et al. for Fe-Mn-C [72], (b) Saeed-Akbari et al. for Fe-Mn-C [26],
(c) Nakano and Jacques for Fe-Mn-C [128] and (d) Saeed-Akbari et al. [26] for Fe-x%Mn-y%C-1.5%Al. The limits for the room temperature transformation and the strain-induced
transformation to ε-martensite reported by Schumann [132] are also included in the graphs.
G 0 determination of these energy barriers is not possible and ab initio

G <0 TWIP-effect calculations are therefore absolutely necessary. A number of codes
70 are available for the ab initio computation of the electronic struc-
= 15mJ/m2 = 45mJ/m2 ture of solids, cf., e.g. Gebhardt et al. [137]. When applied to the
isf isf
problem of deformation twinning in TWIP steel, these simulation
60 programs use specific theoretical approaches, such as density
functional theory, to compute from first principles the correlation
mJ/m2
between specific shear displacements of a rigid lattice and the

50 corresponding free energy changes associated with the nucleation
of an intrinsic stacking fault, a two-layer twin or a multi-layer twin.
In this way the stacking fault energy can be computed explicitly. Ab
isf,
40
initio methods make it possible to calculate minimum energy
pathways to generating intrinsic, extrinsic, and twin faults. The
Calculated
30 outcomes of these calculations, which will be discussed in the next

section, provide the free energy landscapes for TWIP steels. Energy
barriers apparent in the GSFE, characterized by an ‘unstable fault
20 energy’, need to be overcome to produce stable intrinsic stacking
faults, extrinsic stacking faults, or twins. GSFE surfaces typically
have one of the following characteristic shapes: (i) a single
10 Saeed-Akbari et al.
maximum for an undissociated dislocation and (ii) a local minimum
Curtze et al.
corresponding to the formation of an intrinsic stacking fault in the
0 case of a dissociated dislocation. For case (ii) the local minimum on
0 10 20 30 40 50 60 70 the reaction trajectory is preceded by an energy barrier, which
Experimental isf, mJ/m2 represents the resistance to the nucleation of the leading partial
dislocation and the formation of an intrinsic stacking fault. This
Fig. 23. Comparison between experimentally measured [52,62,73,93,113,170e172] and energy barrier is quantified as the unstable stacking fault energy
CALPHAD-based calculated intrinsic stracking fault energy gisf [117,126,127]. gusf [138]. This energy barrier is reduced from gusf to gusf - gisf if
nucleation of the trailing partial of a dissociated dislocation is
considered.
in addition to the computation of gisf [133]. It is widely accepted Expanding on Rice's ideas, the unstable twinning fault energy,
that the conditions for the glide of dissociated dislocations or iso- gutf, was introduced by Tadmor and Hai [139]. It corresponds to the
lated partial dislocations trailing wide stacking faults, as well as resistance to a one-layer intrinsic stacking fault becoming a two-
deformation twinning occurring during the plastic deformation of layer stacking fault or twin embryo. Tadmor and Hai assume that
low and medium stacking fault energy metals and alloys, are the following sequence of events take place during twinning: first, a
controlled by four stacking fault energy parameters: (a) gisf, the leading partial dislocation is emitted; then, instead of emission of a
intrinsic stacking fault energy, (b) gusf, the unstable stacking fault trailing partial, a new leading partial dislocation with the same
energy, (c) gesf, the extrinsic stacking fault energy, and (d) gutf, the Burgers vector as the first one is emitted, yet on a neighboring glide
unstable twin fault energy. These four intrinsic energy barriers plane. The energy barriers associated with this sequence of shears
provide a description of the generalized planar fault energy (GPFE) is illustrated in the schematic GSFE curve shown in Fig. 24 (c).
or g-surface, i.e. the excess free energy when the fcc lattice is When the GSFE curve for an intrinsic stacking fault is calculated, the
sheared on a {111}-type plane. Ratios or differences between these starting configuration is the perfect fcc lattice. The GSFE curve is
four energy barriers can be used to predict the type of planar fault obtained by calculating the energy difference between the perfect
which is more likely to form. fcc lattice and the sheared lattice. The baseline zero energy of the
The concept of a generalized stacking fault energy (GSFE) sur- GSFE curve corresponds to the relaxed perfect crystal with no
face, or g-surface, and the generalized planar fault energy (GPFE) shift. In the GSFE, the first maximum encountered, for a shear equal
surface was originally proposed by Vitek [134,135]. The GSFE sur- to a/12<112>, is the unstable stacking fault energy, gusf, which
face of a TWIP steel represents the energy required to rigidly shear corresponds to the energy required to nucleate a partial dislocation
two parts of a supercell with the fcc crystal structure on a {111} at absolute zero temperature. An intrinsic stacking fault is obtained
glide plane in a 〈112〉 direction. It is defined by the difference be-
by shifting the top layers by 6a < 211 > on a {111}-type plane. When
tween the energy of the perfect lattice Efcc and the energy of the
this is repeated a second time on an adjacent plane, a two-layer
sheared crystal Esf, normalized by the sheared area A:
twin or an extrinsic stacking fault is obtained.
Note that once the GSFE surface has been calculated, it can be
. Esf Efcc ðeVÞ
2
GSFE J m ¼ 1:6$10 19
ðJ=eVÞ$ (30) used to derive the critical resolved stress for slip nucleation from
A m2 first principles [87]:
It should be noted that conceptually, twinning should be seen as

a result of a combination of shear and shuffle components of atomic
1 dE
tcrss ¼ $ (31)
b du max
movement [136]. However, unlike in ab initio simulations, in most
thermodynamic studies the shuffle component is disregarded.
A key parameter for mechanical twinning is the unstable
3.2.2. Ab initio approach twinning fault energy gutf, the saddle point energy associated with
Ab initio computational methods have made it possible to a rigid {111}〈112〉 shift in the presence of an intrinsic stacking fault
compute the GSFE surface, which can be used to understand the on an adjacent {111} plane. Whether dislocation activity is domi-
role of characteristic energy barriers in deformation twinning and nated by perfect dislocations or partial dislocations is determined
g g
predict the ability for twinning of a TWIP steel. Experimental by the g isf ratio. A low g isf value indicates a greater tendency for
usf usf
Saddle Saddle
point Intrinsic stacking fault point
(111) A B C A C A A
B C
a
[110]
2
a a
(a) [112] usf [211] isf usf
6 6
Saddle Saddle Extrinsic stacking fault
point Intrinsic stacking fault point (Two layer twin)
(111) A B C A C A A C A C B
a
[110]
2
a a
(b) 6
[112] usf
6
[112] utf
Stacking fault energy
utf
Emission twinning
usf Partial dislocation
utf isf
usf isf
isf
Emission leading Emission trailing esf isf
Partial dislocation Partial dislocation
(c) 0.0 2 T
0.0 0.5 1.0 1.5 2.0
Rigid lattice shear (units of a/ 6)
Fig. 24. Schematic of the supercell method, which consists in sliding a part of the supercell by a distance a/6<211>, i.e. by the Burgers vector of a partial dislocation, on a {111} plane,
to create an intrinsic stacking fault. A two-layer twin is obtained by shifting the next atomic plane by the same distance. (a) Rigid lattice shear corresponding to the emission of a
leading partial dislocation followed by the emission of the trailing partial dislocation, which restores the perfect lattice. (b) Rigid lattice shear corresponding to the passage of a
partial dislocation followed by the passage of a twinning partial dislocation, which results in the formation of a two-layer twin. (c) Schematic of the generalized planar fault energy
for the shear in (a) and (b). The solid GSFE curve corresponds to the energy landscape for slip by a dissociated dislocation. The dotted line is the energy landscape for the formation
of a two-layer twin by shearing on the adjacent plane.
plastic deformation by dissociated dislocations with a large stack- to one, the barrier for the nucleation of a partial dislocation and the
ing fault. The movement of the leading partial requires overcoming barrier for the nucleation of a two-layer mechanical twin are
an energy barrier of height gusf. The trailing partial encounters a g
similar. Twinnability therefore depends on both the g isf ratio, a
lower energy barrier of height gusfgisf. A large difference between
usf
measure for the energy barrier for partial dislocation glide, and the
gusf and gisf will therefore promote the formation of wide stacking gusf
g
faults. The ratio g isf is a much better parameter to predict the gutf ratio, a measure for the energy barrier for twinning. Twinning
usf g
occurrence of deformation twinning than gisf, deformation twin- will be observed when gusf is large, i.e. the energy barrier for the
utf
g g
ning being more likely when the g isf ratio has a low value. emission of the twinning partial is low, and, at the same time, g isf is
usf usf
Assuming that a two-layer twin acts as an embryo for a full size small. The schematic in Fig. 25 (b) shows the planar stacking fault
deformation twin by thickening, the tendency for twinning by energy variations for twin nucleation and thickening path.
nucleation of a partial dislocation on a slip plane adjacent to a Jin et al. [141] have proposed the following universal relation
g between the planar fault energy barriers in fcc metals and Cu-Al
stacking fault is actually better predicted if both the g isf ratio and
usf
g g alloys based on ab initio GSFE computations:
the gutf ratio are considered [140]. The g isf ratio characterizes the
usf usf
energy barrier for slip only, and the parameters related to the en- gutf 1 gisf
z $ þ1 (32)
ergy barrier for twinning should also be taken into account. The key gusf 2 gusf
to twin nucleation is therefore the energy landscape for the
nucleation of the trailing partial dislocation as compared to the They also confirmed the well-known relation between gisf, gesf
energy landscape for twin nucleation, i.e. the emission of a second and gT, viz.gisf zgesf z2,gT . Ab initio computations have also shown
g that the local composition and the local atomic configuration
leading partial on an adjacent slip plane. When the gutf ratio is close
usf
affected the energy barriers, and confirmed the Suzuki effect, i.e.
a a a
[112] [112] [112]
6 6 6
C A B C
B C A B
A B C A
C A B B
B C C C
A A A A
C C C C
B B B B
(a) A A A A
C
Stacking fault energy
utf utf utf

usf
isf 2 T 2 T
Emission leading Emission twinning
Partial dislocation Partial dislocation
0.0
Perfect Intrinsic Extrinsic Twin nucleation Twin
lattice stacking fault stacking fault (Three layer twin) thickening
(b) (Two layer twin)
Fig. 25. (a) Rigid lattice shears for the formation and thickening of a twin nucleus. In the first stage, the energy path from a perfect fcc crystal supercell to one with an intrinsic
stacking fault is calculated. The saddle point of the transition is gusf, the unstable stacking fault energy. In the second stage, the energy path is calculated for the transformation of an
intrinsic stacking fault to an extrinsic stacking fault, i.e. a two-layer twin. The saddle point is gutf, the unstable twin energy. In the third stage and the subsequent ones, the saddle
point remains equal to gutf as the twin thickens. (b) Schematic of the corresponding planar stacking fault energy for twin nucleation and thickening.
the influence of solute segregation on gisf [142]. Li et al. [143] have twinnability parameter TAS proposed by Asaro and Suresh quan-
shown that the above relation is obeyed by Cu-X (X: Al, Ni, Zn, Ga), tifies the tendency for partial dislocation nucleation at grain
Pd-X (X: Ag, Au), Ni-X (X: Nb, W, Mn, Fe, Cu) and Al-X (X: Mg, Ga, boundaries [145]:
Zn, Si, Cu) alloys. It is worth mentioning that in the work of Jo et al.
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
[144] another interesting form of the same universal relation be-
3$gusf 2$gisf
tween gisf, gusf and gutf was derived: TAS ¼ (36)
gutf

gusf gisf þ gusf gisf When TAS>1, mechanical twinning is favored, while TAS<1 sig-
gutf gisf ¼ 0/gutf gusf ¼ nifies the occurrence of dislocation glide. Li et al. [149] have pointed
2 2
(33) out that, if Jin's relation [141] given by Eq. (32) is, indeed, a uni-
versal law as they claim, then the twinnability should only depend
Equation (33) implies that when gisf ¼ 0, i.e. for DGg/ε ¼ 0, on a single parameter, the g isf ratio:
g
twinning and stacking fault formation leading to hcp ε-phase for- usf
mation, are equally likely to occur. As the universal relation vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u
expressed by Eq. (33) corresponds to idealized conditions which do u3 2$ gisf
gusf gisf u gusf
not take strain contributions into account, Li et al. [133] revised it as TAS ¼ 3$ 2$ ¼t g (37)
gutf gutf 1 þ 12$g isf
follows: usf
gisf g
Accordingly, twinning is favored (TAS>1) if the ratio g isf is smaller
gutf ¼ gusf þ þd (34) usf
2 than 0.8.
Here d is a deviation parameter. The critical upper limit of gisf, for Jo et al. [144] have introduced a different parameter, according
the transition from the strain-induced formation of the ε-phase to to which the deformation mechanisms can be classified. This
deformation twinning, denoted gisf,cr, is determined by the devia- parameter, rd , again depends only on the gisf/gusf ratio:
tion from the relation expressed by Eq. (33):
gisf
gusf
gisf ;cr ¼ 2,d ¼ gutf gusf (35) rd ¼ g (38)
1 g isf
usf
This equation implies that stacking fault formation is preferred
over deformation twinning when gisf is smaller than the difference Low values of rd favor twinning. If rd is less than 1/2, only
between the energy barrier for twinning and stacking fault for- stacking faults are formed. If rd is in the range of 1/2 to 0, stacking
mation. The value of gisf,cr for typical TWIP steels is in the range of faults and dislocation glide occur. Although no orientation depen-
10e20 mJ/m2. dence enters Eq. (38) explicitly, the occurrence of stacking fault or
Various ‘twinnability’ parameters have been proposed. They are dislocation glide in this range of rd does depend on the grain
based on the characteristic energy barriers in the GSFE. The orientation. For rd values ranging between 1 and 2, mechanical
twinning and dislocation glide occur. Finally, for rd in excess of 2, edge components of the partial dislocations of a screw dislocation
dislocation glide prevails. This is a useful concept as it allows one to point in opposite direction. This increases their separation and
set limits on the magnitude of gisf/gusf leading to twinning and partial dislocation breakaway can occur [146] if the following
through that determine the alloy compositions which will promote condition is fulfilled:
deformation twinning:
t$bP ¼ 2$gisf (41)
gisf 1 g
1 < rd ¼ < 20 < isf < 1 (39) Here t is the resolved imposed shear stress, and bP is the
gusf gisf 2 gusf
magnitude of the partial dislocation Burgers vector. When the angle
Combining these inequalities with the ‘universal’ relation pro- between the dislocation line and the dislocation Burgers vector is
posed by Jin et al. [141] Eq. (32), one arrives at the following con- larger than 30 , the edge components of the partial dislocations of
dition for the ratio of the saddle point energy values for twinning the dislocation point in same direction, i.e. the trailing partial al-
and stacking fault formation: ways follows the leading partial dislocation and eliminates the fault
created by the leading partial dislocation. Fig. 26 also shows that
1 gutf 1 this occurs despite an increase of the dissociation width with
< < (40) increasing imposed stress. This also implies that dislocation seg-
4 gusf 2
ments in the screw orientation play a key role in the emergence of
While it can be considered as established that deformation deformation twinning.
twinning is controlled by the intrinsic slip barrier, gusf -gisf, and the The twinnability measure introduced by Bernstein and Tadmor
intrinsic stacking fault energy, gisf, crystallographic orientation is [147,148],
also essential, as shown by Jo et al. for fcc metals by molecular
dynamics (MD) simulations [144]. The direction of the externally ! rffiffiffiffiffiffiffiffiffiffiffi
applied shear is also essential as it can override the intrinsic energy g gusf
TBT ¼ 1:136 0:151$ isf $ (42)
barrier conditions. This implies that twinning also depends on the gusf gutf
angle between the externally imposed shear stress and the orien-
tation of the dislocation line. While screw dislocations will readily indicates the tendency for the occurrence of mechanical twinning
form twins when the energy barrier conditions for twinning are during the deformation of a polycrystalline fcc metal or alloy. A TBT
met, edge dislocations will never form twins. The geometrical value of 0.987 (e.g. for Ni) corresponds to the absence of twinning, a
reason for this orientation dependence is illustrated in Fig. 26. The value of 1.05 (e.g. for Cu) indicates that twinning is still not
Screw orientation Edge orientation
b1e
s s
b b
2 1 b1s b
b b
e e e
b 0 b b
b2e b1e 2
s
2 1
b 2
1=-30º 2=+30º =0º 1=0º 2=+60º =+30º 1=+60º 2=+120º =+90º
Twinning No twinning
Dislocation separation
100 m
10 m
1 m
100 nm
10 nm
1 nm
0 10 20 30 40 50 60 70 80 90
Angle , degrees
Fig. 26. Stacking fault width as a function of the angle q between the perfect dislocation line direction and the Burgers vector for different values of the normalized imposed shear
stress R ¼ 2,t,bP [146]. For g ¼ 30 mJ/m2, R ¼ 1 corresponds to t ¼ 409 MPa. For q < 30 , the edge components of the Burgers vectors of the partial dislocations point in opposite
gisf isf
directions. The partial dislocations thus move in opposite directions, and the widening of the stacking faults leads to twin formation in areas where intrinsic stacking faults overlap.
For q > 30 , the partial dislocations move in the same direction. In this case the stacking fault trailed by the leading partial will always be removed by the trailing partial, i.e. no
twinning will ensue. The three curves represent the stress-free, intermediate and breakaway situation.
Table 3
Twinnability parameters. The twinnability parameter values, assuming gisf ¼ 20 mJ/m2, gusf ¼ 120 mJ/m2, and gutf ¼ 140 mJ/m2, predict that
deformation-twinning will occur during straining.
Twinnability Parameter Parameter Value Reference
dut=us ¼ gutf gusf dut/us ¼ 20 mJ/m2 > 0 Tadmor-Hai [139]

!
qffiffiffiffiffiffiffiffiffiffi TBT ¼ 1.08 Bernstein-Tadmor [147,148]
TBT ¼ 1:136 0:151,ggisf , ggusf
usf utf
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
TAS ¼ 3,gusf 2,gisf TAS ¼ 1.56 Asaro-Suresh [145]
gutf
gisf rd ¼ 0.2
rd ¼ g Jo et al. [144]
usf gisf
predominant, while TBT value of 1.1 or higher points to an easy 3.2.3. First principles computations
activation of twinning. It should be noted that, unlike the Jo et al.’s In the previous paragraphs the GSFE concept was introduced.
parameter rd, the Bernstein-Tadmor parameter TBT involves not one, The GSFE of a TWIP steel can be obtained by first principles com-
but two energy ratios, ggisf and ggusf . In that regard, it is closer to the putations of the structural stability and the fault energy landscape
of fcc g-Fe, Fe-Mn, Fe-Mn-C, Fe-Mn-Al-C, and other TWIP alloys.
usf utf
Asaro-Suresh parameter TAS.

In other publications [131,139] the difference between gutf and Although the results apply to zero absolute temperature, they can
gusf was considered as an appropriate quantitative measure for the provide valuable information about the effect of alloying on the
tendency for deformation twinning: GSFE. The calculation of the GSFE involves complex computations
of solid state properties related to the electronic structure. This
requires solving the Schro €dinger equation self-consistently for
dutf=usf ¼ gutf gusf (43)
electrons in a periodic lattice. This is done by means of ab initio
The parameter dutf/usf compares the barrier for the nucleation of methods, which do not require any empirical input. Only a very
a second leading partial, when a first one has been emitted, relative limited number of inputs, viz. the atomic numbers and the crys-
to the energy barrier for the emission of the trailing partial. If this tallographic structure, are required. The ab initio electronic struc-
parameter is positive, the barrier for the nucleation of a trailing ture computations used to determine the GSFE are based on
partial is lower than for the formation of a two-layer twin nucleus, Density Functional Theory (DFT).
and dislocation glide will be predominant. The different twinn- Analysis of ab initio calculations for g-Fe-Mn alloys [149,150]
ability parameters are compiled in Table 3 and Fig. 27, which il- and g-Fe-Mn-C alloys [151,152] shows that Mn has only a minor
lustrates the relationship between the twinnability parameters and effect on the GSFE [153,154]. Binary g-Fe-C and g-Fe-N have
the shape of the generalized planar stacking fault energy curve. therefore been used to represent the Fe-Mn-C and Fe-Mn-N alloys.
usf utf utf

Energy
Energy
usf
isf
isf 0
Lattice shear Lattice shear
60
Angle , degrees
40
Stacking Perfect dislocation
fault glide
20
Twinning
Pure metals isf
Jo et al. 0
2014 -1.0 -0.5 0.0 0.2 0.5 1.0 1.5 1.8 2.0 2.5 3.0 3.5
usf isf
ut f
Universal scaling law
0.5 1.0 1.08 1.17 1.25 1.30 1.321.33 1.36 1.38 1.39
usf
isf usf
Bernstein and Tadmor
2004
(1.136 0.152 )
1.82 1.14 1.07 1.01 0.95 0.92 0.9 0.90 0.88 0.87 0.86
usf utf
Increasing tendency for twinning
usf isf
Asaro and Suresh
2005
3 2
2.83 1.73 1.56 1.38 1.18 1.05 1 0.96 0.88 0.83 0.78
utf utf
Mechanical twinning favored Dislocation glide favored
Fig. 27. Schematic illustrating the relationship between proposed twinnability parameters and the shape of the generalized planar stacking fault energy curve [141,145,149,150,155].
The periodic supercell approach mentioned above has been approach requires prior DFT calculation of the total energy of
implemented for g-Fe alloys in the commercial quantum simula- defect-free unit-cells with different stacking sequences. In the axial
tion software Vienna Ab Initio Software Package (VASP). The peri- next-nearest-neighbor Ising (ANNNI) model proposed by Denter-
odic supercell method avoids the problem of surfaces and involves neer and van Haeringen [164], the energy of a particular stacking
the use of a cell which explicitly contains a specific fault. On this fault gsf is defined as an excess energy per unit area [165,166]:
basis, zero Kelvin calculations for pure g-Fe correctly predict large
negative values, e.g. 380 mJ/m2 [155,156] or 347 mJ/m2 [157] for Esf E0
gsf ¼ (45)
gisf of TWIP steels when no spin magnetization is considered. This A
comes as no surprise, as at 0 K the hcp ε-phase is more stable than Here Esf and E0 are the energy of the crystal with and without a
the fcc g-phase, even if different magnetic states (non magnetic, stacking fault, respectively, and A is the area associated with a
ferromagnetic, and anti-ferromagnetic) are considered for g-Fe. pffiffiffi
single atom in the slip plane, i.e.A ¼ 3,ag =4. Using the ANNNI
Another important group of ab initio methods are based on the
approach, Denteneer and Soler derived the following equations for
so called Muffin-Tin Approximation (MTA) [158]. The ab initio DFT
gisf and gesf [167]:
method can be used to determine the total energy and electronic
density of a solid [158,159]. Whereas the use of supercell first 4$ðJ1 þ J2 þ J3 Þ
principles approach is common in ab initio DFT calculations, it is gisf z
A
challenging to use it for modeling all possible concentrations and (46)
4$ðJ1 þ 2$J2 þ 2$J3 Þ
configurations. The exact muffin tin orbital (EMTO) method is gesf z
considered to be the most powerful technique to obtain the elec- A
tronic structure of random substitutional alloys with a close packed Here J1, J2, and J3 are the nearest-neighbor, next nearest-
crystal structure, such as g-Fe alloys [158]. It replaces the alloy by an neighbor and next-next nearest-neighbor interaction parameters.
ordered effective medium. In this method the solid is subdivided In can be shown that the Ji parameters are related to Ghcp, the free
into spherically symmetrized regions around the atomic nuclei and energy of ε-Fe (… ABAB … -type stacking), Gdhcp, the free energy of
interstitial regions between the atoms. The DFT Kohn-Sham equa- double hexagonal close packed Fe (dhcp-Fe with … ABACABAC …
tions are solved using the EMTO method and the full charge density -type stacking) and Gfcc, the free energy of g-Fe (… ABCABC … -type
(FCD) technique, which is combined with the coherent-potential stacking). The calculation of gisf requires the computation of the
approximation (CPA). CPA is a method of performing configura- free energy of just these three defect-free structures:
tional averaging which replaces a multi-component random alloy
with a translationally invariant system with a single component Ghcp þ 2$Gdhcp 3$G fcc
gisf z (47)
representing an effective medium. It is used in conjunction with a A
mean-field approximation called the disordered local moment
Local changes in atomic positions or magnetic order resulting
(DLM), which takes into consideration the effects of magnetic dis-
from the presence of the stacking fault are not taken into account in
order on the electronic structure [160]. Results of the combined
the ANNNI approach.
EMTO-CPA method, which is computationally efficient, have almost
It is important to note that in g-Fe with low gisf alloys defor-
the same accuracy as the full potential methods that do not involve
mation twins, ε-martensite and a’-martensite are often observed
approximations. In the EMTO-CPA method the stacking fault en-
experimentally. At large strains a’-martensite nucleates at in-
ergy at 0 K is defined as an excess free energy per unit area [147]:
tersections of ε-martensite, twins and shear bands. This has also
been observed in TWIP steels [168,169]. The formation of ε and a’-
2$ Ehcp Efcc martensite have not yet been related to the GSFE surface approach
gisf zgesf z (44)
ASF discussed in the previous section, and the energy barriers associ-
ated with the formation of ε and a’-martensite have not been
Here Ehcp and Efcc are the total energy per atom of the hcp and
studied as yet.
the fcc phase, respectively. ASF is the area of the stacking fault.
Ab initio methods based on density functional theory are in
principle very suitable to support materials design, and the 3.3. Effect of alloying elements
approach has been useful for the calculation of 0 K ground state
properties. While this is very valuable at a fundamental level, it is In the original work of Schumann [132] on the mechanical
not sufficient for the actual design of g-Fe alloys, as these can have a properties of Fe-Mn-C alloys, the composition was mainly consid-
variety of magnetic states which will significantly influence the ered as a means to achieve room temperature phase stability for the
outcome of DFT computations. Despite the popularity of the ab fcc crystal structure. The observation of the microstructure formed
initio calculations, a careful approach is required in their imple- also enabled the author to draw a characteristic boundary line
mentation and application to high Mn TWIP steels. Ekholm and between two regions in the Mn-C composition plane: one of stable
Abrikosov [161] noted that standard ab initio computational ap- compositions deforming by dislocation glide and twinning and
proaches do not always describe 3d electron systems, such as g-Fe- another one of unstable compositions, i.e. those leading to the
Mn based alloys, sufficiently accurately, Reyes-Huamantinco et al. deformation-induced formation of the ε phase:
[162] have shown that magnetism and thermal lattice expansion
%Mn ¼ 32 20$%C (48)
play a role in the determination of gisf for a paramagnetic random
Fe-22.5%Mn alloy, thereby illustrating that gisf is sensitive to the Here %C and %Mn denote the weight percentage of C and Mn in
magneto-volume coupling at finite temperatures. This is in the alloy, respectively. In the range of 15e25%Mn and 0.4e0.8%C,
disagreement with the results by Vitos et al. for Fe-Cr-Ni alloys, this equation corresponds approximately to alloy compositions
who report that the temperature dependence of the gisf is deter- with an iso-gisf value of about 20 mJ/m2, according to Saeed-Akbari
mined only by the contribution of magnetic fluctuations to the free et al. [117], and 30 mJ/m2, according to Nakano and Jacques [128]
energy [163]. (Fig. 22).
The fault energy barriers can also be calculated using the energy It is clear that the alloy composition plays a key role in deter-
of crystal structures with periodically repeated stacking faults. The mining the magnitude of gisf, but the fundamental reasons for this
influence are often more complex than the composition depen- a reduction of the probability that an interstitial octahedral site in
dence of stability of the g and ε phases. In what follows, the in- the vicinity of a C-occupied octahedral site will also be occupied by
fluence of the most important alloying elements on the properties a C atom. As a result, the C diffusivity will rise with increasing C/Fe-
of TWIP steels is therefore reviewed in detail. atomic ratio [172,173]. The 0 K first principles analysis of Fe-Mn-Al-
C alloys has also shown that there is a pronounced repulsion be-
3.3.1. Carbon tween interstitial C atoms in TWIP steels [155,157,174e178]. Sub-
Carbon plays a central role in the properties of TWIP steels as: stitutional solutes, such as Al and Mn, affect the properties of C
(a) it increases the lattice parameter, (b) it enhances the stability of (Fig. 28). Al has been shown to have a small effect on the C diffu-
the austenite relative to ε-martensite (i.e. C increases DGg/ε), (c) it sivity in g-Fe-C-Al alloys, but Shun et al. [179], who studied the
promotes paramagnetism by reducing the Ne el temperature TgN, (d) temperature dependence of the properties of austenitic Fe-30%Mn-
it substantially increases gisf, with a larger effect for a lower Mn 1.0%C (gisf ¼ 27 mJ/m2), report that the apparent activation energy
for the onset of DSA-related serrations on the deformation curves
content in Fe-Mn-C TWIP steels, and (e) it leads to a strong solid
solution hardening effect. In g-Fe and alloys, carbon occupies was significantly lower than the activation energy for bulk diffusion
of carbon. According to Shun et al., this suggests that the serrations
octahedral interstitial sites. The small lattice distortion caused by C
is isotropic. The nearest neighbor (NN) and next nearest neighbor are related to the short range diffusion of C in the dislocation core,
rather than its bulk diffusion. Addition of 2.7% Al increased the
(NNN) interactions between carbon atoms in austenite have been
€ssbauer Spectroscopy (MS), thermodynamic esti- activation energy for the onset of serrations, indicating that Al re-
studied by Mo
duces the C mobility at dislocations.
mations and Monte Carlo simulations [170,171]. The NN and NNN
Carbon also strongly affects the local magnetic interactions in
C-C interactions have been found to be repulsive in g-Fe-C alloys
(Fig. 28). The results imply that the C distribution should be ho-
g-Fe-Mn-C alloys. Various magnetic states have been observed
experimentally for g-Fe (non-magnetic, low-spin ferromagnetic,
mogeneous. The repulsive character of C-C interactions also leads to
Fe C Mn Al
NNN-repulsion
NN-attraction
(-7kJ/mol)
(+26kJ/mol)
NN-repulsion
(-35kJ/mol)
NNN-attraction
(+13kJ/mol)
(a) (b)
NN-repulsion
(-10kJ/mol)
NNN-attraction
(+25kJ/mol)
-carbide
(c) (d) Fe3AlC
Fig. 28. Schematic for the NN and NNN interactions of solute atoms in fcc g-Fe alloys. (a) Repulsive NN and NNN C-C interaction. (b) Attractive NN and NNN C-Mn interaction. (c)
Repulsive NN and attractive NNN C-Al interaction. (d) Crystal structure of k-carbide, which is stabilized by the attractive NNN C-Al interaction.
high-spin ferromagnetic, and collinear and non-collinear anti- width by 40e60%, corresponding to a decrease of the gisf by
ferromagnetic) [180e183]. The collinear spin, anti-ferromagnetic approximately 9 mJ/m2, during in situ heating of Fe-22%Mn-0.6%C
double layer (AFMD) state is the lowest energy state in g-Fe, g- TWIP steel from 158 C to þ40 C. They refer to this C ‘nano-
Fe-C and g-Fe-Mn-C alloys [151]. In the lowest energy AFMD diffusion’ out of the stacking fault plane as an anti-Suziki effect, to
state the Fe6-octahedra comprise five Fe atoms which are ferro- highlight the fact that the normal Suzuki effect involves a reduction
magnetically coupled and one Fe atom which has an opposite of gisf resulting from the diffusion of substitutional solutes to the
spin. The presence of C induces the formation of local ferro- stacking fault plane. The solute C interstitials cause a lattice
magnetic Fe clusters in g-Fe-C alloys, i.e. the ferro-magnetic expansion of g-Fe by about 0.0041 nm/mass-%, which affects gisf. In
coupling of the spins of nearest Fe atoms is favored by the addition, the shear of the atomic layers during the glide of the
presence of C. In g-Fe-Mn-C alloys there is a strong tendency to leading partial dislocation of a dissociated perfect dislocation will
form Mn-C pairs. The g-Fe-Mn-C alloys are magnetically non- cause an iron atom to pass on top of the carbon atoms which are
homogeneous, because C also affects the local magnetic in- located in the stacking fault plane. This creates an unstable high
teractions in this case. C also frustrates the local magnetic energy configuration, forcing the C atoms to carry out a diffusional
ordering in g-Fe-Mn-C alloys by the re-orientation of the mag- jump to a nearby vacant octahedral interstitial site. Both effects will
netic moments of the nearest Fe and Mn atoms, thereby stabi- cause the carbon atoms to diffuse away from the stacking fault,
lizing the Fe-Mn ferro-magnetic coupling. The results of the thereby decreasing gisf and widening the stacking fault. Fig. 29,
above mentioned studies suggest that the preferred location of C which is based on the work of Gholizadeh [74] and Limmer [185],
atoms is away from the stacking fault plane. The stacking fault illustrates an increase of gisf when C-atoms are located in the
energy is the highest when C atoms are situated in the stacking stacking fault plane. Their calculations do not take into account the
fault plane. The 0 K value of gisf is increased by 300 mJ/m2 by the interaction of carbon atoms with the partial dislocation which may
addition of about 1 mass-% C. The effect is very short range. Thus, influence the carbon-stacking fault interaction.
the influence of C on the stacking fault energy rapidly decreases The affinity of Fe and Mn for C is related to the Fe-C and Mn-C
when the C atoms are positioned further away from the fault chemical bond strength in the solid. According to von Appen and
plane. This dependence of gisf on the distance of the C-atom from Dronskowski [67], the C-Fe and C-Mn bonds are strong local co-
the stacking fault plane is shown in Fig. 29 based on the work of valent bonds. The formation of Mn-rich octahedra with C in the
Gholizadeh [74]: the further C atoms are from the stacking fault interstitial positions cannot be explained in terms of bond strength,
plane, the smaller their influence on gisf. Only a few atomic hops as the binding energy of the C-Fe bond (4.43eV) is slightly larger
of the C atom away from the stacking fault are required to obtain in magnitude than that of the C-Mn bond (4.31eV). Since there are
a decrease of gisf to its C-free stacking fault value. Abbasi et al. relatively few C-Fe and C-Mn bonds in the alloy, they are not ex-
[155] also reported a significant increase of the stacking fault pected to have a significant influence on its electronic structure and
energy with increasing C content, along with a strong depen- lattice properties which depend on it, such as the elastic constants.
dence of the stacking fault energy on the position of the C atoms The solid solution strengthening effect of C in TWIP steel, with
relative to the stacking fault. an increment of approximately 279 MPa/mass-% C [186], is less
Experimental evidence for the short-distance (nano-scale) pronounced than in common stainless steels [187]. Allain [66] and
diffusion of C atoms away from the stacking fault has been provided Kusakin et al. [188] reported a similar solid solution strengthening
by Hickel et al. [184], who report a transmission electron micro- effect of 250 MPa/mass-% C.
scopy observation of an irreversible increase of the stacking fault
2500 isf max -Fe 2500 Fe24C supercell

initial 0K initial 0K
2000 2000
1500 1500
GSFE, mJ/m2
mJ/m2
1000 diffused 1000 diffused
u
isf,
500 500
0 0
C-free
-500 -500
0 1 2 3 0 1 2 3
Displacement (a/6<112>) Distance from SF plane
(a) (b)
Fig. 29. (a) Generalized stacking fault energy for g-Fe at 0 K, illustrating the effect of C on the magnitude of gisf. The value of gisf is very high when a C atom is in its initial state,
i.e. positioned on the stacking fault plane during the passage of a dissociated dislocation. As the C atom diffuses to a more favorable position inside the stacking fault, gisf is
reduced but is still high compared to a C-free stacking fault. (b) Effect of the distance of solute C from the stacking fault plane on gisf. Data redrawn from the work of Gholizadeh
[74] and Limmer [185].
3.3.2. Manganese faults during dislocation glide [193].

As manganese increases the ε/g transformation temperature,
it is expected that addition of Mn results in a decrease of gisf. This
3.3.3. Aluminium
appears to make Mn rather exceptional as all other transition
The role of Al in TWIP steel is complex. Ab initio calculations
metals increase gisf [177]. The Mn content dependence of gisf is non-
show that Al is a potent stabilizer of the g-Fe-Mn random alloys
linear. In binary Fe-Mn alloys, in low concentrations, Mn reduces
below as well as above the Ne el temperature, which is consistent
gisf.The lowest gisf value is obtained in the range of 10e16 at.% Mn.
with an increase in the gisf upon alloying with Al. The influence of Al
At higher Mn content, in the concentration range from 16 at.% to
appears to be independent of the magnetic state, and the volume
33 at.%, alloying with Mn increases gisf by 18 mJ/m2 per 1 at.% Mn.
effect outweighs the chemical effect [137]. Jin et al. [62] have
This non-linear dependence of gisf is due to the effect of Mn on the
studied the influence of Al addition on gisf for a Fe-18%Mn-0.6%C-x%
fcc-hcp energy difference: an increase of the Mn content sup-
Al TWIP steel by XRD and TEM. Their findings are expressed by the
presses the formation of the hcp structure. The experimentally
following linear Al-content dependence of the intrinsic stacking
observed non-linear dependence has been confirmed by first
fault energy:
principles calculations for disordered Fe-Mn alloys. With regard to
the stacking fault formation, first principles computations predict .
that alloying with Mn should result in a continuous decrease of gusf.
gisf mJ m2 ¼ 20 þ 7:8$%Alðmass%Þ (49)
A further prediction is that Mn atoms should have a tendency to
Using gisf measurements based on neutron diffraction, Kang
segregate to stacking faults.
et al. demonstrated an increase of gisf by 10 mJ/m per mass-% Al, i.e.
The strong attractive NN C-Mn (þ26 kJ/mol) and NNN C-Mn
from 17 mJ/m2 for Fe-18%Mn-0.6%C to 37 mJ/m2 for Fe-18%Mn-0.6%
(þ13 kJ/mol) interaction between Mn and C atoms in TWIP steels
C-2%Al [194]. These results are in reasonable agreement with the
results in the formation of Mn-C pairs [157,177]. Kang et al. [189]
increase by approximately 8 mJ/m2 per mass-% Al predicted by
carried out ab initio calculations of the energy change related to
thermodynamic calculation [117].
the addition of manganese atoms to Fe6-x-Mnx-C (0 x 6)
Alloying with Al inhibits dynamic strain aging in C-alloyed TWIP
octahedra in which a carbon atom occupies the central octahedral
steels. Based on their study of the temperature dependent properties
interstitial position. They report a free energy decrease by 0.0528eV
of austenitic Fe-30%Mn-x%Al-1.0%C alloys, Shun et al. [195] report
per Mn atom (or 5.09 kJ/mol), i.e. the C-Mn interaction is attractive.
that Al additions increased the apparent activation energy, as defined
The attractive C-Mn interaction leads to an increase of the solubility
by the onset strain for the onset of discontinuous yielding, from 14.4
limit of C in g-Fe when Mn is added.
to 22.3 kcal/mol due to the addition of 2.7 mass-% Al, suggesting that
Below we consider the effects of Mn on the properties of g-Fe-C
Al-additions reduce the C diffusivity within the dislocation core.
alloys, in addition to the effect on gisf discussed above, namely, (i)
Al promotes short range ordering and carbide formation. This is
magnetic properties and (ii) diffusivity.
related to the strong attractive NNN C-Al interaction, which has a
binding energy of 25 kJ/mol. It results in an increase of the solu-
(i) The presence of Mn suppresses ferromagnetism and ex-
bility limit of carbon in g-Fe when Al is added. As the NN C-Al
tends the temperature range in which anti-ferromagnetism
interaction (10 kJ/mol, or 0.104 eV/at) is repulsive, the formation
occurs by raising the Neel temperature. In the temperature
of local order resembling the crystal structure of perovskite Fe3AlC
range for which C diffusion has typically been studied, i.e. in
k-carbide is favored [173]. This compound was experimentally
the high temperature range of 720e1200 C, dilute fcc g-Fe-
identified by Park et al. in Fe-22%Mn-0.6%C-(0-3-6)%Al TWIP steels
C alloys are paramagnetic, i.e. they are characterized by a
[61]. The NN C-Al repulsive interaction was also reported by Kang
random orientation of local magnetic moments. Depending
et al. [189] who carried out ab initio calculations of the energy in-
on the composition, fcc g-Fe-Mn-C alloys and other TWIP
crease associated with the addition of an Al atom to Fe6-x-Alx-C
steels are either anti-ferromagnetic or paramagnetic at
octahedra, in which a carbon atom occupies the central octahedral
room temperature. Serrated stress-strain curves observed
interstitial position. An energy increment of þ0.382eV per Al atom
at room temperature suggest the possibility of the occur-
(þ36.86 kJ/mol) means that the interaction is strongly repulsive. As
rence of short-range diffusion of carbon atoms. Ab initio
a result of Al-Mn interaction being repulsive, Al atoms avoid Mn
calculations by Jiang and Carter [75] mentioned above have
atoms as nearest neighbors. Zero Kelvin first principles computa-
shown that the activation energy for diffusion of C in g-Fe
tions [157,177] also suggest that the Al distribution in TWIP steel is
strongly depends on its magnetic state. According to that
non-uniform.
study, the ground state of g-Fe is the anti-ferromagnetic
Ab initio calculations have revealed that alongside the mentioned
double layer phase with a bi-layer … [[YY … in the
increase of the intrinsic stacking fault energy gisf, alloying of TWIP
〈001〉 orientation.
steels with Al leads to a systematic decrease of gusf. The ensuing
(ii) The formation of Mn-C pairs reduces the C diffusivity. Using
reduction in the difference gusf-gisf with the growing Al content is
the nudged elastic band method, Jiang and Carter [75]
expected to facilitate the emission of a trailing partial relative to the
calculated the activation energy for C diffusion and found a
emission of a twinning partial dislocation, provided the magnitude
value of 0.99 eV, which is close to the experimental value for
of gutf is unaffected by the Al content or decreases with it.
very dilute Fe-C alloys [190e192]. The details of carbon
diffusion in g-Fe based alloys are complex, the activation
energy being temperature and composition dependent. 3.3.4. Silicon
Thus, the activation energy for C diffusion in g-Fe in the C Si is a strong solid solution strengthening element in g-Fe alloys.
concentration range of 0e1.4 mass-% is 1.01e1.19 eV; spe- Its addition to an alloy reduces the Ne el temperature and affects the
cifically, at 0.6% mass-% C it amounts to about 1.17 eV [70]. magnetic order below the Ne el temperature. No volume change is
observed when Si is added to Fe-Mn alloys, indicating that the ef-
The room temperature dynamic strain aging of TWIP steel, fect of Si on the stability of the g-phase is governed by electron
which manifests itself in serrated stress-strain curves, has been contributions to the interactions between the atoms. In the pres-
linked to the re-orientation of Mn-C pairs in the vicinity of stacking ence of C, the NN Si-C interaction is strongly repulsive. The NNN Si-
C interaction is attractive. In the presence of both C and Mn, which
is the case for a TWIP steel, there is a repulsive Si-Mn interaction. over Cu-Fe and Ni-Fe bonds. The effect of Cu on the stacking fault
According to Tian and Zhang [196], an increase in the Si content energy of TWIP steels was studied by considering the variation of
results in a decrease of gisf in Fe-Mn-Si-C alloys with up to 8 at-%Si. gisf of Fe-Mn-C-Al TWIP steel from 22.8 mJm2 to 26.0 mJm2 due
This was confirmed by Jeong et al. [172] who measured a drop in gisf to an addition of 2% of Cu [104]. Cu additions also retard the kinetics
at a rate of 4 mJm2 per mass-% of Si for Si concentrations up to 1.5 of twin formation [104]. Alloying with copper also has an effect on
mass-%. They also reported that the decrease of gisf in a 1.5% Si- DSA of Fe-Mn-C-Al TWIP steels, which is similar to the effect of N
added TWIP steel resulted in a higher strain hardening, which with respect to the type of serrations on the stress-strain curves
they attributed to enhanced activation of primary and secondary and the number of PLC bands. A beneficial role of Cu additions in
twinning. increasing total elongation without a loss of strength was reported
There is no consensus about the effect of Si on gisf, however. by Lee et al. [200].
Based on calculations, Dumay et al. [125] reported an increase of gisf At the atomistic scale, an important effect of Cu and Ni on gisf
for Si concentrations of up to 6 at.-% and a smooth decrease at was established by the ab initio calculations of Limmer [185]. It was
higher concentrations. This is at variance with the mentioned re- found that Cu and Ni atoms located in the stacking fault plane in-
sults of Tian and Zhang [196] and Jeong et al. [172]. The latter group crease gisf and decrease gusf. Specifically, the following figures for
of authors suggested that this discrepancy is due to insufficient the rate of increase of gisf were found: 9.9 mJ/m2 per atom-% Cu and
accuracy of the Mn-Si and Si-C interaction parameters used in the 10.0 mJ/m2 per atom-% Ni.
model calculations.
3.3.7. Chromium
3.3.5. Nitrogen While Cu and Ni atoms prefer to cluster in fcc austenite, the
Nitrogen has been reported to promote the formation of interaction between Cr atoms appears to point to a preference for
ε-martensite in Fe-16%Mn-(0.015, 0.05)%N TWIP steel, indicating their planar segregation, favoring Cr-Fe over Cr-Cr bonds. In the
that its presence reduces gisf [104]. Using a combination of presence of both C and Mn, as is the case for a TWIP steel, the
neutron diffraction, XRD, and TEM Lee et al. [104] have shown, attractive C-Cr bond remains attractive. Due to a weak Cr-Mn
however, that gisf varies with the concentration of nitrogen at a repulsion, the preferred atomic arrangement is the 180 -oriented
rate of approximately 100 mJm2/mass-%N. The authors explained Cr-C-Mn octahedral complex. Ab initio computations for binary g-
the observed retardation of twinning kinetics and reduction of the Fe-Cr alloys predict that the effect of Cr is similar to that of Mn.
thickening of the twins to this increase of gisf with increasing N Alloying with either of them diminishes gusf and gisf. Cr atoms
content. Huang et al. [197] showed that N increases gisf in Fe- located at a stacking fault reduce gisf by 1.7 mJ/m2 per atom-% Cr
(20e22)%Mn-0.01%C-(0.7e2.5)%Al-(2e3)%Si-(0e0.05)%N TWIP within the range of Cr concentrations up to 5.6 atom-% Cr [177].
steel. The observed reduction in the probability of the occurrence This is consistent with the predictions of thermodynamic stacking
of stacking faults with increasing N content in Fe-30%Mn-6%Si- fault energy models, which also predict a decrease gisf upon addi-
(0e0.047%) N steel also suggests that alloying with N raises gisf tion of Cr.
[198]. A general observation that refers to the alloying additions
According to Lua et al. [199], the main effect of interstitials on gisf considered above (including Mn, Al, Ni, Cu, and Cr) is that their
is through the variation of the lattice parameter they cause. Their influence on gisf extends only over a region of two {111} planes
calculations of the effect of N and C additions on of Fe-22%Mn TWIP adjacent to the stacking fault [185].
steel showed the variation of gisf at a rate of 29 mJm2/mass-% N
and 32 mJm2/mass-% C, in agreement with the results of ther-
modynamic modeling [117]. 4. Mechanical properties of TWIP steel
The lowering of gisf by N has been used to explain the activation
of twinning in Hadfield steel (12%Mn-1.2%C). In High Nitrogen The yield strength of TWIP steels is relatively low, and a large
strain is needed to see the strength benefits associated with strain
Steels (HNS) and in austenitic stainless steels N inhibits twinning in
favor of planar slip of dissociated dislocations when N < 0.4 mass-%. hardening. This implies that additional strengthening mechanisms
have to be introduced to attain a well-balanced property profile. An
For N > 0.7 mass-%, the addition of N promotes extensive twinning.
Thus, alloying with 1 mass-% of N (4 at-%) was shown to increase obvious option is pre-straining, but it results in lower tensile
elongation under subsequent loading. Alternative strengthening
the gisf of these steels. Lee et al. [104] have studied the effect of
nitrogen on gisf of a Fe-15%Mn-2%Cr-0.6%C-x%N TWIP steel by mechanisms involve grain refinement and precipitation hardening
by micro-alloying.
means of XRD. They established the following phenomenological
dependence of the stacking fault energy on the nitrogen
concentration: 4.1. Elastic properties of TWIP steel: Young's modulus anomaly
.
gisf mJ m2 ¼ 12 þ 100$%Nðmass%Þ (50) The magneto-elastic properties of anti-ferromagnetic high Mn
Fe-alloys are associated with a pronounced Young's modulus
A similar trend, i.e. a linear increase of gisf from 404 mJ/m2 to anomaly. The Young's modulus E is higher in the high-temperature
179 mJ/m2 for an increase of the N content by 2.14 mass-% in Fe-12% paramagnetic state than in the low-temperature magnetically or-
Mn-x%N alloys was found by means of ab initio DFT calculations dered anti-ferromagnetic state. The origin of this “DE”-effect is
[156]. This corresponds to the rate of variation of gisf of 272 mJ/m2/ spontaneous magnetostriction. A volume expansion caused by the
mass-% N. They also report that the proximity of N to the stacking anti-ferromagnetic ordering that occurs when the alloy is cooled
fault increases gisf, which implies that N should have a tendency to below the Ne el temperature leads to a deviation from the normal
diffuse away from the stacking fault. temperature dependence of the modulus of elasticity [201,202].
The decrease in the modulus can be rationalized in terms of an
3.3.6. Copper, nickel increase in the interatomic distances associated with the volume
Both Cu and Ni stabilize the fcc g-phase relative to the hcp expansion, which leads to a weaker interatomic bonding. That is to
ε-martensite. Cu atoms also tend to cluster in fcc austenite, as N say, the attendant lattice softening results in a stiffness reduction.
atoms do as well. That is to say, Cu-Cu or Ni-Ni bonds are favored The fundamental reason for this magneto-elastic effect lies in the
effect of the changing interatomic distances on the electronic The weak sensitivity of the Young's modulus of TWIP steels to
structure, i.e. the density of states and the topology of the Fermi alloying with carbon can be rationalized in terms of the nature of
surface. This kind of “DE”-effect associated with the formation of the atomic bonds. According to von Appen and Dronskowski [67],
antiferromagnetic order below the Ne el temperature has been the electronic structure of Fe-Mn-C TWIP steels consists of covalent
observed for TWIP steels. Fe-C and Mn-C bonds and metallic Fe-Fe, Mn-Mn, and Fe-Mn
A systematic study of Young's modulus of relatively highly bonds. Since the number of covalent bonds is small, the addition
Mn-alloyed ternary face-centered cubic Fe-(28.0e37.5)at.% Mn- of C has only a local effect and the electronic structure is not
(1.5e3.0)at.% C steels was carried out by Reeh et al. [203] by affected by C additions in the concentration range used for TWIP
means of nano-indentation and ab initio calculations. According steels. The effect of C on the Young's modulus is therefore also
to their findings, within the compositional range studied, the expected to be minimal.
carbon content significantly affected the lattice parameter, Allain [66] stated that the shear modulus is reduced substan-
whose value was varied in the range of 0.3597e0.3614 nm. The el temperature for the anti-ferromagnetic transition
tially if the Ne
ab initio calculations using the VASP code predict that the happens to be close to room temperature or to exceed it. The Ne el
addition of carbon leads to a slight lattice expansion by 0.4% temperature of Al-free Fe-22%Mn-0.6%C TWIP steel is 321 K (at
relative to the binary Fe-Mn alloys. Reech et al. also reported 48 C). The magnitude of the shear modulus is reduced from
that carbon has no effect on Young's modulus. The measured 72 GPa at 107 C to as low a value as 60.5 GPa at 74 C. At room
Young's modulus values were in the range of 185(±12)-220(±20) temperature it takes the value of 62 GPa. For the same steel
GPa for bulk samples. Assuming a Poisson ratio value of composition, Steinmetz et al. report a shear modulus value of as
v ¼ 0.29, this corresponds to a magnitude of 72e85(±20) GPa low as 52.5 GPa [41].
for the shear modulus G ¼ 2ð1þ E Fig. 30 shows that Al-additions, which increase the lattice
nÞ at room temperature. These
values are in broad agreement with Jung and De Cooman who parameter, also result in a reduction of the shear modulus of Fe-18%
measured the temperature and Al-content dependence of the Mn-0.6%C-x%Al TWIP steel. The values of the elastic constants for
shear modulus for Fe-18%Mn-0.6%C-x%Al TWIP steels (Fig. 30) high Mn Fe-alloys are compiled in Table 4. As the Mn content is
[105]. The value of the room temperature shear modulus they reduced from 40 to 5 at. %, C44 is unaffected, while C11 and C12 in-
quote is in the range of 76.5e79.0 GPa. It should be noted that crease by 25.6% and 39.2%, respectively. This behavior of the elastic
the Neel temperature of Fe-18%Mn-0.6%C-x%Al TWIP steels is constants is due to the magneto-volume effect which softens the
below room temperature. lattice.
200 Fe-18%Mn-0.6%C
80
Young's Modulus E, GPa
79
Shear Modulus G, GPa
180
78
77
160
76
140 75 Fe-18%Mn-0.6%C-x% Al
0.0% Al
74 1.5% Al
TNéel
2.5% Al
120 73
-100 0 100 200 300 400 500 600 -60 -20 +20 +60 +100 +160
(a) Temperature, ºC (b) Temperature, ºC
el
Fig. 30. (a) Temperature dependence of Young's modulus of Fe-18%MMn-0.6%C TWIP steel. The anti-ferromagnetic ordering results in a reduction of the modulus below the Ne
temperature. (b) Effect of Al on the shear modulus of Fe-18%Mn-0.6%C-x%Al TWIP steel [210].
Table 4
Single crystal elastic constants for selected high Mn Fe-alloys. Average isotropic shear modulus G and bulk modulus B (compressive modulus) were computed according to the
following equations.
GVoigt þ GReuss C þ 2C12

G¼ ; B ¼ 11
2 3
C11 C12 þ 3C44 5ðC11 C12 ÞC44
GVoigt ¼ ; GReuss ¼
5 3ðC11 C12 Þ þ 4C44
C11 C12 C44 C12C44 G B B/G Reference

GPa GPa GPa GPa GPa GPa
Fe-25%Mn-3%Al-3%Si 174 85 45 14 71.8 114.7 1.60 Pierce et al. [206]

Fe-22%Mn-3%Al-3%Si 175 83 46 14 71.9 113.7 1.58 Pierce et al. [206]
Fe-20%Mn-1.2%C 236 139 117 þ22 82.2 171.3 2.08 Idrissi et al. [91]
Fe-24%Mn-3%Al-2%Si-1%Ni-0.06%C 167 110 140 30 75.0 128.9 1.72 Saleh et al. [43]
Table 5 with a cubic crystal structure is characterized by B/G < 1.75,

Room temperature shear modulus of selected TWIP steels. ductility is associated with B/G > 1.75. The B/G ratio is mainly
Composition G Reference determined by the difference C12C44, which is positive for
GPa ductile materials [209]. In first-principle calculations of elastic
Fe-24%Mn-0.7%Ca 62.4 Kang et al. [184] and magnetic properties of Fe-Mn-X alloys, the calculated B/G
Fe-22%Mn-0.5%Ca 63.6 ratio can therefore be used as a useful criterion to predict which
Fe-18%Mn-0.5%Ca 73.2 alloying element might improve the intrinsic ductility of a TWIP
Fe-22%Mn-0.6%Ca Choi et al. [186]
steel. In this light, the data shown in Table 5 suggests that high
[5 4 22] single crystal 67.0
[16 6 23] single crystal 77.0 Mn Fe-alloys are more likely to display brittle behavior. Based on
Fe-22%Mn-0.6%Ca,b 62.0 Bouaziz et al. [6] the Pugh and Pettifor criteria, only the Fe-15%Mn alloy is ex-
Fe-18%Mn-0.6%Cb 79.0 Jung and De Cooman [105] pected to have a ductile behavior. Reeh et al. [203] predict that
Fe-18%Mn-0.6%C-1.5%Alb 78.0
Cu additions, which significantly reduce G while only slightly
Fe-18%Mn-0.6%C-2.5%Alb 76.7
increasing B, should give rise to a substantial improvement of
a
Anti-ferromagnetic. ductility of high Mn Fe-alloys, such as TWIP steels. Their calcu-
b
Polycrystal.
lations, based on the EMTO formalism as implemented in the
EMTO5.7 Package, suggest that the addition of Cu increases the
valence electron occupancy in the d-bands, which are not fully
Considerable grain to grain stress incompatibilities can be
occupied. This enhances the metallic component of the bonds
generated in polycrystals deformed in the elastic regime and during
relative to the covalent component, and leads to a reduction of
the elasto-plastic transition when the elastic anisotropy coefficient,
the shear modulus. A similar trend was observed by Jung [210] in
A ¼ 2C44/(C11-C12), is large [204]. The ab initio single crystal elastic
experiments on Fe-18%Mn-0.6%C-x%Al TWIP steels. It was found
coefficients of C-free Fe-Mn-Al and Fe-Mn-Si TWIP steels deter-
that an increase in Al content lowers the magnitude of the shear
mined by Gebhardt et al. [205] give very large A values in the range
modulus, cf. Fig. 30b.
of 4.6e7.4. In contrast, experimentally determined single crystal
elastic coefficients for the Fe-22%Mn-3%Al-3%Si, Fe-25%Mn-3%Al-
3%Si and Fe-18%Mn-1.5%Al-0.6%C TWIP steels result in a much 4.2. The elasto-plastic transition
lower value for A in the range of 2.11e2.22 [206]. This difference is
very likely due to a magneto volume effect [207]: while anti- Fig. 31 compares stress-strain curves for polycrystalline TWIP
ferromagnetically ordered TWIP steels have a large elastic anisot- steel samples tested in tension and single crystals of TWIP steel
ropy, the anisotropy is significantly lower for magnetically disor- tested in compression. Using the Taylor factor (M ¼ 3.06) to
dered TWIP steels. compute the critical resolved shear stress for plastic flow initiation,
According to Pugh [208], the B/G ratio, i.e. the ratio of the bulk one obtains the values of 76 MPa and 72 MPa for single crystals
modulus B to the shear modulus G, can be used as a guidance to with a twinning-suppressing [011]-orientation and a twinning-
predict the ductility of a solid. The bulk modulus B is a measure enhancing [001]-orientation, respectively. Bracke et al. [17] re-
of the resistance of a material against volume compression. It is ported very similar values, viz. a critical resolved stress for dislo-
related to the metallic bond strength and the resistance to cation slip of 89 MPa, and a critical resolved twinning shear stress
rupture. The shear modulus G is a measure of the resistance to a of 77 MPa. It is unlikely that partial dislocation sources can be
shape change by shear deformation. Materials which deform activated at such low stresses, since Frank-Read type partial dislo-
easily commonly have a lower shear modulus compared to hard cation sources required to generate twins cannot achieve instability
brittle materials. A ductile material will therefore have a high B/G and expand at such a low stress level. This suggests that the elasto-
ratio, as the grains at the tip of a crack will deform by having plastic transition is controlled by a dislocation slip mechanism
their atomic planes slip rather than being pulled apart. which does not involve twinning. The deformation of single crystal
Conversely, a brittle material is characterized by a low B/G ratio. TWIP steel micro-pillars was investigated by Choi et al. [186] and
At the crack tip in a brittle material the atomic planes will be Wu et al. [211]. The well-defined testing conditions used enabled a
pulled apart instead of slipping. Brittleness in metals and alloys study of plastic deformation in the absence of strain gradients,
(a) (b)
Fig. 31. (a) Stress-strain curves for polycrystalline TWIP steel deformed in tension and single crystals TWIP steel deformed in compression. (b) Stress-strain curves for submicron
diameter TWIP steel micropillars deformed in compression. The arrows indicate small strain bursts at low stress levels. sYS: yield strength, tYS: yield shear strength [186].
grain boundaries, grain size effects, etc. A complicating issue in

micro-pillar measurements is the size effect. A size effect is also of tcrss ðMPaÞ ¼ 74:5 þ 61:11$%C 0:65$%Mn ¼ 97 MPa
(52)
relevance to the critical stress for twinning [212]. In fcc metals and sys ðMPaÞ ¼ 296 MPa
alloys this effect is associated with source starvation, when the
Choi et al. [186] put forward an alternative equation for
pillars “run out of dislocations” and the dislocation-free state re-
composition-dependent solid solution strengthening in terms of
quires high strength for commencement of plastic flow. Alterna-
the critical resolved shear stress and the lattice friction stress for Fe-
tively, truncation of the dislocation Frank-Read source size may be
Mn-C-Al-Si TWIP steel:
the reason for a size effect. It is interesting that Wu et al. [211]
concluded that twins were nucleated at the surface of the pillars,
tcrss ðMPaÞ ¼ 31:7 þ 91:3,%C 0:49$%Mn þ 16:2$%Si þ 6:7$%Al
whereas Choi et al. [186] observed that dislocationedislocation
s0 ðMPaÞ ¼ 97 þ 279$%C 1:5$%Mn þ 49:6$%Si þ 20:5$%Al
interactions were needed for twin formation in the bulk. The yield
strength of micro-pillars oriented for dislocation glide was (53)
425 MPa, while that associated with an orientation favoring twin- Using the equation proposed by Choi et al. [186], one obtains for
ning was 316 MPa. Wu et al. measured a critical resolved shear Fe-15%Mn-0.7%C-2%Al-2%Si TWIP steel:
tcrss ðMPaÞ ¼ 31:7 þ 91:3$%C 0:49$%Mn þ 16:2$%Si þ 6:7$%Al ¼ 134MPa

(54)
s0 ðMPaÞ ¼ 410MPa
stress of 488 MPa for a Fe-22%Mn-0.6%C micropillar 0.705 mm in It is recognized that the overall effect of solutes on the charac-
diameter with a [001] orientation favoring twinning. The large teristics of plastic yielding is quite appreciable. It should be noted
difference, i.e. 488 MPa in Wu et al. study versus 316 MPa in Choi that Mn additions provide a negative contribution to this effect
et al. is due to the fact that Wu et al. report the maximum regardless of the specific form of the heuristic relation for solute
compression stress, whereas Choi et al. consider the average value solution strengthening chosen by the different authors.
of the stress at the first strain burst as a signature of plastic flow There are conflicting reports on the solid solution strengthening
initiation due to the activation of a single source. The size- of Al and further investigation of the effect of Al on the yield
independent critical resolved shear stress measured on a 7.6 mm strength of TWIP steel are needed. In contrast to Jung and De
diameter Fe-22%Mn-0.6%C TWIP steel micro-pillar was 61.1 MPa, Cooman who report a hardening effect of 6.7 MPa/mass-%Al [105],
i.e. only slightly smaller than the critical resolved shear stress of Pasakin et al. [188] find a much lower solid solution strengthening
76 MPa obtained experimentally for polycrystalline TWIP steel. effect of 0.8 MPa/mass-% Al.
Table 6 reviews reported values for the critical resolved shear stress
and indicates whether a particular orientation favors dislocation 4.4. Thermally activated plastic flow in TWIP steels
glide or deformation twinning.
Allain et al. [213] and Jung and De Cooman [105] have investi-
gated the thermally activated plastic flow in austenitic high Mn Fe-
Mn-C steels. The Peierls shear stress, tP, is the resolved shear stress
4.3. Solute solution strengthening required to move a straight dislocation from one valley of the Peierls
relief to the next one a thermally, at absolute zero. This quantity
The strengthening effect of solute atoms in austenitic fcc Fe al- cannot be measured experimentally, but theoretical equations can
loys is known to be much less pronounced than in the case of ferritic be used to calculate the Peierls shear stress at 0 K [214,215]. Using
bcc Fe alloys. Bouaziz et al. have proposed the following heuristic the characteristic parameter values for {111}-ag/2<110> dislocation
equation for the lattice friction stress, which includes the solid so- glide in g-Fe, viz. an effective inter-planar spacing equal to
lution strengthening effect of C and Mn in Fe-Mn-C TWIP steel [53]: pffiffiffi pffiffiffi
d111 ¼ 6 3a, a dislocation Burgers vector of magnitude b ¼ 2 2a, a
lattice parameter a of 0.36 nm, a zero Kelvin shear modulus Gg(0 K)
s0 ðMPaÞ ¼ 228 þ 187$%C 2:0$%Mn (51)
of 78.24 GPa, and Poisson's ratio n(0 K) ¼ 0.313, one obtains the
Using the Taylor orientation factor M ¼ 3.06, the critical resolved following estimates for the zero Kelvin Peierls stress for an edge and
shear stress and the corresponding yield strength for Fe-22%Mn- a screw dislocation, respectively:
0.6%C TWIP steel are given by:
Table 6
Overview of the reported critical resolved shear stress values for dislocation glide and twinning in TWIP steels.
Steel Compression axis Sample type Deformation tCRSS MPa Reference
Fe-22%Mn-0.6%C [0 0 1] Micro-pillar (0.7 mm diameter) Twinning 488 Wu et al. [211]

Fe-22%Mn-0.6%C [16 6 23] Micro-pillar (2.0 mm diameter) Dislocation glide 206 Choi et al. [186]
Fe-22%Mn-0.6%C [1 3 6] Micro-pillar (3.9 mm diameter) Twinning 188 Wu et al. [211]
Fe-22%Mn-0.6%C e Polycrystal (EN10002) e 160 Choi et al. [186]
Fe-22%Mn-0.6%C [5 4 22] Micro-pillar (2.0 mm) Twinning 154 Choi et al. [186]
Fe-18%Mn-0.6%C e Polycrystal (ASTM E8) e 144 Dumay et al. [125]
Fe-22%Mn-0.6%C [0 1 1] Single crystal (4.0 mm) Dislocation glide 76 Choi et al. [186]
Fe-22%Mn-0.6%C [0 0 1] Single crystal (4.0 mm) Twinning 72 Choi et al. [186]

G 2$p d 78 GPa 2p
tP;edge ¼ $exp $ ¼ $exp $1:1547 ¼ 2:9MPa
1n 1n b 1 0:313 1 0:313

(55)
d
tP;screw ¼ G$exp 2$p$ ¼ 78 GPa exp½2$p$1:1547 ¼ 55 MPa
b
Allain et al. [213] suggested that, in addition to its influence on exhibits a clear temperature dependence below room temperature.
the TWIP effect, C also affects the dislocation mobility. The tem- By analyzing the data from various literature sources, Allain et al.
perature dependence of the yield strength of various Fe-Mn, Fe- [213] showed that the temperature dependence of the contribution
Mn-C, and Fe-Mn-C-Al steels, modified by subtraction of the yield to the yield stress associated with thermally activated dislocation
stress at room temperature, is presented in Fig. 32. The yield stress motion was more pronounced in TWIP steels with a high C content.
500 Fe-18%Mn-0.6%C 500

Fe-18%Mn-0.6%C-1.5%Al
Fe-18%Mn-0.6%C-2.5%Al 432.1MPa
400 400
MPa
MPa
300 300
ys(298),
ys(298),
~0.76MPa/K Higher C content

200 200
-
-
ys(T)
ys(T)
100 100
~0.20MPa/K ~0.25MPa/K
0 0
~0.76MPa/K
-100 -100
0 100 200 300 400 500 600 0 100 200 300 400 500 600
(a) Temperature, K (b) Temperature, K
0 Fe-18%Mn-0.6%C-1.5%Al 0 25ºC
-10 25ºC -10 45ºC
65ºC
-20 -20
-30 -30 Vapp~30 b3

, MPa
, MPa
-40 -40
-50 -50
-60 -60
-70 -70
-80
3 kB T t
-80 ln(1 )
Vapp
-90 -90
0 20 40 60 0 2 4 6
(c) Time, s (d) Ln(1+t/ )
Fig. 32. (a) Temperature dependence of the difference sys(T)- sys(298) for TWIP steels with 0.6%C. (b) Temperature dependence of the difference sys(T)- sys(298) for TWIP steels
with a higher C content (>0.6%). (c) Isothermal stress relaxation test results for Fe-18%Mn-0.6%C-1.5%Al TWIP steel. (d) Ds vs. ln(1 þ t/t) plot for isothermal relaxation tests carried
out at 25 C, 45 C and 65 C to determine the apparent activation volume Vapp [66,216].
1000 the atomic volume, or ~ b3110, is indicative of dislocation climb

433 control. An activation volume in the intermediate range of
900 ys ( MPa) 184 10e100 b3110 is likely to be associated with the lattice resistance
800 d ( m) (Peierls mechanism) or dislocation-point defect interaction. Allain
Yield strength, MPa
reported that while the activation energy for TWIP steels (~0.5eV)
700 was independent of the C content, the activation volume did
600 depend on it, decreasing with the carbon concentration [71]. This
suggests that the low temperature dislocation dynamics are
500 controlled by a strong interaction of mobile dislocations with car-
bon atoms facilitated by a low activation energy for C diffusion in
400
the vicinity of the dislocation core. However, in the observed range
Fe-17Mn-0.45C
300 of the apparent activation volume there may be an overlap of
Fe-17Mn-0.60C
solute-controlled and Peierls relief-controlled mechanisms of
200 Fe-18Mn-0.60C
Fe-22Mn-0.60C thermally activated dislocation dynamics [217,218]. To untangle
100 Fe-25Mn-0.02C-3Si-3Al these effects, a detailed analysis of the temperature, stress, and
solute concentration dependence of the plastic strain rate would be
0 required.
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
In the cited studies by Allain [213] and Majidi et al. [216], the
-1/2 -1/2
d , m magnetic state of the alloys (anti-ferromagnetic or paramagnetic),
the Mn content, and the Al content did not appear to influence the
Fig. 33. Verification of the Hall-Petch relation for various TWIP steels (Fe-22%Mn-0.6% activation volume. Above the room temperature, the yield stress of
C [218], Fe-17%Mn-0.45%C [221], Fe-25%Mn-3%Si-3%Al-0.02%C [225], Fe-17%Mn-0.6%C all the TWIP steels studied showed a reduced temperature sensi-
[228]). tivity of approximately 0.20 to 0.25 MPa/K, cf. Fig. 32, which
may be associated with the temperature dependence of the shear
modulus, rather than a thermally activated character of plastic flow.
They also determined the attendant activation volume, which was
in the range of 14e60 b3110, where b110 is the atomic spacing in the
close-packed direction 〈110〉. Majidi et al. [216] carried out careful 4.5. Grain size strengthening in TWIP steels
isothermal measurements of the apparent activation energy of Fe-
18%Mn-0.6%C-1.5%Al TWIP steel using stress relaxation experi- In numerous publications, the yield stress, sYS, of polycrystalline
ments. The results are shown in Fig. 32. The time dependence of the TWIP steels was shown to follow the Hall-Petch relation:
stress decrement Ds was found to roughly follow a logarithmic
dependence and be linear in ln(1 þ t/t). Here t is the time, and t is a kHP
characteristic relaxation time, whose temperature dependence sys ðMPaÞ ¼ s0ys þ pysffiffiffi (57)
d
presented in an Arrhenius form can be used to determine the
activation energy of the process. Such a logarithmic dependence, in Here the parameter s0YS includes the lattice friction stress, the
the form solid solution strengthening contribution of the alloying elements,
and the strain hardening contribution of the initial dislocation
pffiffiffi kB ,T density, kHP

YS is a material parameter, and d is the average grain size.
t
Ds ¼ 3$ $ln 1 þ (56) Fig. 33 compiling some data for TWIP steels demonstrates that they
Vapp t obey the Hall-Petch relation. Available literature data on the pa-
rameters s0ys and kHP
ys for a number of TWIP steels are reviewed in
would, indeed, follow for a stress relaxation curve under the
Table 7 [6,49,217e221]. For polycrystalline Fe-22%Mn-0.6%C TWIP
assumption of a constant apparent activation volume, Vapp, in the
steel, the reported values of s0YS and kHP YS are 132 MPa and
Arrhenius relation for the plastic strain rate (or individual dislo-
cation velocity). The activation volume is an informative footprint 449 MPa mm1/2 (14.2 MPa mm1/2), respectively. However, the s0YS
of the rate-controlling mechanism of plastic deformation, which value given by Bouaziz et al. [6] in their original manuscript seems
can be used to identify the mechanism. While dislocation dynamics to have been misprinted since the yield stress level is significantly
controlled by thermally activated breakaway from forest disloca- lower than the value calculated using the parameters from the
tion junctions is characterized by a large activation volume of the same reference. Therefore, the value of s0YS for polycrystalline Fe-
order of 1000 b3110, a very small activation volume, of the order of 22%Mn-0.6%C TWIP steel was re-evaluated. The corrected
Table 7
Parameters of the Hall-Petch relation for TWIP steels.
s0ys, MPa ys , MPa.mm

kHP 1/2
Alloy Reference
Fe-31%Mn-3%Al-3%Si 53 764 Dini et al. [217]

Fe-22%Mn-0.6%C 137 449 Bouaziz et al. [6]
170 428 Scott et al. [218]
Fe-22%Mn-0.6%C 157 357 Gil Sevillano-Cuevas [49]
Fe-22%Mn-0.5%C-0.08%N 219 478 Lee-De Cooman [214]
Fe-20%Mn-0.6%C 158 485 Shen et al. [220]
Fe-17%Mn-0.45%C-1.5%Al-1%Sia 208 445 Gwon [221]
Fe-17%Mn-0.45%C-1.5%Al-1%Sib 403 445
Fe-15%Mn-0.7%C-2%Al-2%Si 305 330 Rahman et al. [20]
a
Hot rolled.
b
Hot rolled and recrystallization annealed.
magnitude for s0YS is 242 MPa, rather than 132 MPa, the value re- The yield strength of TWIP steels is generally rather low, and
ported by Bouaziz et al., and the corresponding critical resolved relatively large deformations are required to achieve a high flow
shear stress is 79 MPa, assuming a Taylor factor of 3.06 [6]. Lee and stress. This limits the use of TWIP steels to complex press-formed
De Cooman [214] have derived the following Hall-Petch equations parts whose manufacturing involves large amounts of plastic
for the yield strength and the ultimate tensile strength for Fe-22% strain. Micro-alloying of TWIP steels was therefore considered as a
Mn-0.5%C-0.08%N: way to increase the yield strength by grain size reduction and
precipitation hardening. Results are available for low carbon Fe-25%
15:1MPa$mm1=2 477:5MPa$mm1=2 Mn-3%Si-3%Al-0.02%C TWIP steel [225], and high carbon Fe-17%
sys ðMPaÞ ¼ 219 þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 219 þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi Mn-0.6%C [226] and Fe-22%Mn-0.6%C TWIP steel [218]. The
dðmmÞ dðmmÞ
observed strengthening was accounted for in terms of the Orowan
18:8MPa$mm1=2 594:5MPa$mm1=2 precipitate bypassing mechanism. Unfortunately, no information
sUTS ðMPaÞ ¼ 754 þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 754 þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi on the equilibrium solute solubility and precipitation kinetics for
dðmmÞ dðmmÞ
nitrides, carbides or carbo-nitrides in TWIP steels is currently
(58) available in relation to micro-alloying additions of Ti, Nb or V. It can
Kang et al. [222] analyzed the influence of the C content on the therefore not be ruled out that some of the observed improvements
Hall-Petch coefficient kHP ys using a slow cooling procedure to ensure of strength were due to solid solution strengthening by the micro-
equilibrium C segregation to the available grain boundary sites. alloying additions.
Kang et al. reported that kHP ys increased from 218 MPa mm
1/2 Vanadium appears to be the most efficient micro-alloying
to
344 MPa mm1/2 when the C content was doubled from 0.3 mass-% to element, as it both reduces the grain size and leads to the forma-
0.6 mass-%. They offered two possible explanations for this pro- tion of very small precipitates, which can be 3.4 nm in size, ac-
nounced increase of kHP ys . The effect may be associated either with an cording to Scott et al. [218]. The solid solution strengthening effect
increased stress for the activation of dislocation sources in the bulk of V is approximately 17 MPa/mass-% V. Vanadium also inhibits the
of a grain when the C content is increased or with an increased recrystallization kinetics. According to Chateau et al. [227], who
density of grain boundary ledges acting as dislocation sources studied the effect of V micro-alloying additions on the properties of
[223]. The second explanation is erroneous as the mechanism Fe-22%Mn-0.6%C TWIP steel, the VC particles do no change the
would result in a decrease of kHP ys upon an increase in C. strain hardening behavior. Based on their TEM analysis, Chateau
The Hall-Petch diagram can be affected by alloying additions, as et al. concluded that VC precipitates do not act as strong obstacles
demonstrated in Fig. 34 for the case of alloying TWIP steels with Ti, to the propagation of microtwins. They report that the addition of
V, and Ni. It also depends on the pre-strain at which the mea- 0.21 mass-% V was found to bring about a strengthening effect of
surements are taken for a particular material. While the slope kHP 140 MPa. Yen et al. [228] associated the strengthening effect of V in
ys of
the Hall-Petch diagram decreases with increasing strain for a-Fe Fe-21.6%Mn-0.63%C-0.87%V TWIP steel with precipitation. Indeed,
and ferritic steels, it increases with strain for austenitic steels. The a high density of V4C3 carbide precipitates with the average
latter trend is also observed for TWIP steels. Gil Sevillano et al. [44] diameter of 13 ± 7.7 nm were formed during a recrystallization
and de las Cuevas [224] reported a Hall-Petch slope of about annealing treatment (60 s at 850 C). A stress increment of 198 MPa
350 MPa mm1/2 for the yield strength, and a Hall-Petch slope of by precipitate strengthening was attributed to the Orowan
630 MPa mm1/2 for the ultimate tensile strength. The data of Wang bypassing mechanism. This is a rather modest strengthening effect
et al. [225] for Fe-24.8%Mn-0.022%C-3.17%Si-3.12%Al confirms that considering the high V content in the steel studied. A reduction of
the Hall-Petch slope is higher for the ultimate tensile strength, the precipitate size to below 5 nm, as observed by Scott et al. [218]
amounting to 638 MPa mm1/2 vs. 568 MPa mm1/2 for the yield enables a more pronounced strengthening effect due to the Orowan
strength. bypassing mechanism. To quantify the effect for the case of VC
700 700
600 600
Yield strength, MPa
Yield strength, MPa
500 500
400 400
300 300
KHP=433 MPa m1/2 KHP=433 MPa m1/2

200 200
Fe-22Mn-0.6C-0.16V Fe-22Mn-0.6C-0.1Ti
100 Fe-22Mn-0.6C-0.21V 100 Fe-22Mn-0.6C-0.2Ti
Fe-22Mn-0.6C-0.40V Fe-22Mn-0.6C-0.3Ti
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
-1/2 -1/2 -1/2 -1/2
(a) d , m (b) d , m
Fig. 34. Influence of micro-alloying additions of (a) V and (b) Ti on the Hall-Petch diagram for TWIP steel [218,221].
precipitates, the Ashby-Orowan equation 22.3Mn-0.19Si-0.14Ni-0.27Cr-0.61C TWIP steel deformed by equal

channel angular pressing (ECAP) were investigated. It was shown
rffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffi that a good balance between strength (1702 MPa) and tensile
1:2 3 f r
p elongation (24%) in the steel with a resulting UFG structure can be
Dsys ¼ 0:847$M$ $G$b $ VC $ln (59)
2,p 2,p rp b achieved by a suitable choice of the processing schedule. This was
associated with the formation of deformation microbands and
can be used. Here fVC is the volume fraction of vanadium carbides twins (including nano-twins) in the microstructure during the
and rp is their mean radius. The least-square fitting of the data ECAP processing.
shown in Fig. 35 to Eq. (56) yields rp ¼ 10 nm for M ¼ 3.06. This is in
excellent agreement with the precipitate radius range of 7e12 nm
5. Twinning mechanisms in TWIP steels
reported by Chateau et al. [227].
It is generally accepted that grain refinement down to the sub-
As deformation twinning is at the core of the mechanical
micron grain size, 100 nm < d < 1 mm, results, with a few excep-
on this subject will be
properties of TWIP steels, a detailed expose
tions, in a reduction of strain hardening and hence lowered uniform
given in this section.
tensile elongation [229,230]. Plasticity as such is not suppressed, as
the ultrafine grained (UFG) materials sustain a large post-uniform
elongation in the necking region. TWIP steel does not appear to 5.1. Mechanisms of deformation twinning
be susceptible to a negative influence of small grain size on the
uniform elongation. Grain growth after recrystallization annealing Deformation twinning results from a homogeneous shearing of
makes it possible to obtain a broad grain size range for a high Mn the matrix by a highly coordinated glide of partial dislocations with
TWIP steel. Ueji et al. [231,232] reported that grain size reduction in the same Burgers vector, with exactly one dislocation gliding on
C-free Fe-31%Mn-3%Al-3%Si TWIP steel (with the intrinsic stacking each successive {111}-type twinning plane. In polycrystalline ma-
fault energy gisf ~42 mJ/m2) resulted in a strong inhibition of terials twin formation by {111}〈112〉 slip results in a volume
deformation twinning and a significant drop in ductility. In fine orientation change of a certain fraction of the twinned grains which
grained TWIP steel twinning was still prevalent in grains oriented affects the crystallographic texture. {111}〈112〉 slip requires com-
with a 〈111〉 direction close to the tensile axis. Gutierrez-Urrutia plete separation of partial dislocations, or the existence of partial
et al. [233] report that grain refinement did not suppress defor- dislocation sources. The spontaneous formation of twins is very
mation twinning in a Fe-22%Mn-0.6%C TWIP steel (gisf ~23 mJ/m2) unlikely as it would require very high stresses. The model originally
altogether, but made it more difficult. They also associated the proposed by Fontaine [235] for deformation twinning is generally
smaller grain size obtained to a reduction of the twin volume accepted. In this model wide stacking faults are first formed,
fraction. Lee et al. [226] compared the tensile deformation behavior overlapping stacking faults then generate twin embryos, and twin
of UFG C-free Fe-17%Mn and Fe-17%Mn-0.6%C TWIP steel with that growth occurs by the addition of a stacking fault on the twin plane.
of the C-added TWIP steel and concluded that the grain refinement The wide stacking faults that form twin embryos can either be a
induced reduction in elongation was smaller for the latter. It should result of suitable dislocation reactions or they can be emitted from
be noted that TWIP steel with the grain size down to about 2 mm interfaces such as grain boundaries. The conditions for twin
still retained a large uniform elongation of approximately 50% en- nucleation from stacking faults emitted from grain boundaries are
gineering strain. In a recent work [234] the properties of a Fe- not well established. Grain boundaries containing interfacial
700 Fe-22%Mn-0.60%C-V 300 Fe-22%Mn-0.60%C-V

0.31% V
Fe-17%Mn-0.45%C-V Fe-17%Mn-0.90%C-V
, MPa
0.21% V P
600 0.11% V
250
0% V
500
Yield stress increment
Yield stress, MPa
433
0.40% V ys 184 d 200
400 0.21% V
0.16% V ys 2270 fVC
0% V 150
GS
300
100
200
ss
50
100 rp=10nm
0
0 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14
(a) 1/d1/2, m-1/2 (b) fVC1/2

Fig. 35. (a) Analysis of the grain size dependence of the yield strength of V-added Fe-22%Mn-0.6%C and Fe-17%Mn-0.45%C TWIP steels, indicating the various contributions to
strength. (b) Analysis of the contribution of the VC precipitates through the Orowan bypassing mechanism [218,221].
dislocations which can emit a twinning partial dislocation with a twinning in imparting enhanced plasticity to UFG and nano-
Burgers vector out of the boundary plane, e.g. dissociated screw- crystalline materials with high gisf [238e241]. The present section
type misfit dislocations in a twist boundary, could act as areas will review only those deformation twinning models which have
where twin embryos are formed. Random grain boundaries could been reported to be compatible with the TEM analysis of twins in
also contain pre-existing multi-layer twin nuclei which can easily TWIP steels. Six twinning mechanisms which have been discussed
be activated at low strains. in relation to deformation twinning in TWIP steel are as follows:
Models for deformation twinning should be compatible with
the basic feature of the deformation in TWIP steel, which is (1) the Venables pole mechanism [95].
generally considered to be initially by dislocation slip. During this (2) the Cohen-Weertman Frank cross-slip mechanism [242].
early stage secondary slip systems are already activated, and long (3) the Fujita-Mori stair-rod cross-slip mechanism [243].
stacking faults and secondary slip systems are visible in TEM. (4) the Mahajan-Chin extrinsic fault mechanism [244].
Deformation twins are formed soon after yielding and at low (5) the Miura-Takamura-Narita Frank primary slip mechanism
strains in areas where two slip systems are activated. Twins are [245].
mainly formed on the primary twin system and the contribution of (6) the Copley-Kear-Byun partial dislocation breakaway mech-
secondary twins systems to deformation twinning is limited. A anism [146,246].
twinned region consists of very thin twin plates. Twinning is sen-
sitive to the crystal orientation. It is easy when a grain has a 〈111〉 These models have in common that deformation twinning re-
direction or a direction on the 〈111><100〉 tie line close to the quires achieving a high enough dislocation density and a high local
tensile axis. stress to trigger the twin nucleation and growth. The schematics in
Reviews offering an in-depth description of classical models for Fig. 36 illustrate the key aspects of the deformation twinning
deformation twinning are available in the literature [236,237], but mechanisms which have been reported in relation to TWIP steels in
it should be mentioned that several new twinning mechanisms literature. In the figure, the Thompson tetrahedron is used to
have recently been proposed due to the importance of deformation describe the twinning mechanisms. Perfect and partial dislocations
B B B
D D Extrinsic D
A A Stacking A
fault
D
CB D
C C A C
A
CB B
N2
A N1 DA B B A
D
(B A) ( B D )
DA D A BA B A
B D
D
N2
N1
C A
A
CB C B B
A A
D
A
D B
D d111
d111 D
D
D A
d111
A
A C d111 A
B C B B
A
D A
Fujita-Mori-Liu Venables Mahajan-Chin

(a) (b) (c)
Fig. 36. Schematics of (a) the Fujita-Mori-Liu stair-rod cross slip twinning model [243], (b) the Venables pole dislocation mechanism model [95], and (c) the Mahajan-Chin three
layer twin nucleus [244].
slip on {111}-type slip planes only. There are four {111} slip planes dislocation, taking into account gisf and the drag force. The twin
in total. Adjacent {111} slip planes intersect at an angle of 70.53 . thickening mechanism is based on a chance overlap of wide
The edges of the tetrahedron are a/2<110>-type perfect dislocation stacking fault generated by the breakaway process.
Burgers vectors [247]. Much of the early work on deformation In the Mahajan-Chin model two dislocations with coplanar
twinning was done on single crystals of metals or alloys strained Burgers vectors form a set of three Shockley partials on three
beyond stage II and into stage III strain hardening range, i.e. in consecutive slip planes, resulting in an extrinsic fault configuration.
relatively strongly deformed material subjected to high stresses. In This three-layer extrinsic stacking fault is considered to act as a
the case of TWIP steels, deformation twinning is believed to start in twin nucleus. The mechanism requires the cross-slip of a partial
early stages of the deformation. It is generally accepted that twin- dislocation, and twin thickening requires the chance overlap of
ning requires the presence of dissociated dislocations, i.e. a low gisf, similar three-layer twin nuclei [244]. Bracke et al. [17] report that
and that the nucleation of the twins requires dislocation reactions their TEM observations agree with the Mahajan-Chin twinning
leading to the formation of twinning partial dislocations trailing an model. Steinmetz et al. [41] considers a strongly pinned segment of
intrinsic stacking fault, and a process which suppresses the for- a pre-existing three-layer Mahajan-Chin stacking fault as the twin
mation or the movement of the corresponding trailing partial nucleus in their model for the TWIP effect. Mahato et al. [19]
dislocation. The twinning partial dislocations involved in the pro- observed extrinsic stacking faults by TEM of TWIP steel. Taking
cess of twin thickening must all have the same Burgers vector and the viewpoint that extrinsic stacking faults serve as precursors to
move on consecutive {111} slip planes. the formation of twin nuclei, they conclude that the Mahajan-Chin
In the Venables model [95,248,249], a twinning partial, formed mechanism operates in their case. Kibey et al. [156] also favor the
on a long jog on the primary slip plane, rotates around a specific Mahajan-Chin model for the deformation twinning in Hadfield
pole dislocation. The model requires the formation of a large steel.
enough jog, which implies that multiple intersections must have Based on the direct observation of twin formation, Liu et al.
taken place in multi-slip conditions. The obvious advantage of the [251] concluded that a slightly revised version of the Fujita-Mori
Venables pole mechanism is that a perfect twin can be formed, model [243] was compatible with their in situ observations made
without the need to create new overlapping twinning partials, as in a Fe-24%Mn-0.5%C TWIP steel. The Fujita-Mori-Liu model im-
the twin gradually thickens with strain. The mechanism requires plies prior activation of slip on two slip systems. In the model, the
that a long jog is first formed on a dislocation of the secondary slip leading partial dislocation of a primary slip dislocation splits into a
system plane by intersection of primary slip dislocations. The jog stair-rod dislocation and a twinning partial dislocation on the
then dissociates into an a/6<112>-type twinning partial and an a/ conjugate slip plane.
3<111>-type sessile Frank partial dislocation. The Venables A key feature of the Fujita-Mori model [243] involves the “stair-
mechanism implies that a certain amount of plastic strain must rod cross slip” mechanism, which enables cross-slip of partial dis-
precede twin formation. An alternative process has been proposed locations. The model requires that slip is activated on at least two
by Niewczas and Saada [250]. They assumed conditions of high slip systems. A a/6<112>-type partial dislocation on the primary
stress and large dislocation density and envisaged that a primary slip plane dissociates into a sessile a/6<110>-type stair-rod dislo-
dislocation reacts with one arm of a faulted Frank dipole, forming a cation and a glissile a/6<112>-type partial dislocation on the con-
twinning partial dislocation that is free to move on the conjugate jugate slip plane. The model is based on a dislocation reaction
slip plane. In the case of the Venables mechanism, the jog, which originally proposed by Cohen and Weertman: a perfect a/2<1110>-
lies on the primary slip plane, should be large enough, i.e. result type dislocation at the head of a dislocation pileup in front of a
from about 20 intersections of the dislocation by primary disloca- strong barrier, e.g. a Lomer-Cottrell lock or dislocation dipole, can
tions (~5 nm in length) and subjected to a considerable stress dissociate in a sessile a/3<111>-type Frank dislocation and a a/
concentration in order to operate as a source. Accordingly, high 6<112>-type partial dislocation. The partial dislocation can glide
passing stresses must be overcome for the operation of the twin away from the Frank dislocation on the conjugate slip plane trailing
source. The poles of the source should be on opposite sides of the a wide stacking fault. In the Cohen-Weertman deformation twin-
twinning partial. The Venables pole model [95] offers an elegant ning mechanism additional partials have to be emitted for a twin to
explanation for how a perfect mirror twin, which implies that thicken, but this would require a highly ordered arrangement of
exactly one partial dislocation glides on each consecutive slip plane, stacking faults to be produced by chance. In the Fujita-Mori model,
is formed. Whereas there is no direct experimental evidence for the the cross-slip of the partial dislocations occurs in an orderly
Venables mechanism, Niewczas and Saada have published weak- manner.
beam dark field transmission electron micrographs of twin sour- In the Miura-Takamura-Narita mechanism (Fig. 37) primary
ces in high deformed Cu-8%Al which support their model. While dislocations first form a dislocation pileup at a Lomer-Cottrell
the Venables and Niewczas-Saada models have been used to dislocation barrier located at the junction of the primary slip
explain the observation made by Choi et al. [186] for TWIP micro- plane and a cross-slip plane. A two-layer twin nucleus, consisting of
pillars, no convincing experimental evidence is available, to the two twinning partial dislocations and a sessile Frank partial dislo-
best of the authors' knowledge, for the operation of Venables or cation, is formed. Karaman et al. [252] used a modified version of
Niewczas-Saada twin sources in TWIP steel. the Miura-Takamura-Narita model to compute the critical twinning
Glide-type twin nuclei models are attractive due to their stress for Fe-12%Mn-1.2%C steel.
simplicity. Pure glide type models for twin formation do not require Idrissi et al. [91] studied the twin nucleation mechanism in Fe-
multi-slip conditions. The simplest glide-type twin nuclei model is 20%Mn-1.2%C TWIP steel by TEM. Their results appear to support
based on the Copley-Kear-Buyn's dislocation break-away mecha- the Cohen-Weertman (Fig. 38) or the Miura-Takamura-Narita
nism. The Mahajan-Chin model for the formation of a three-layer nucleation mechanisms, whereas twin thickening appears to be
twin nucleus, formed by two dissociated co-planar dislocations, is due to the pole mechanism as proposed by Venables [95].
also a glide-type source. It requires the cross-slip of partial dislo- Considering the large variety of possible mechanisms and in the
cations. In the Copley-Kear-Byun model twinning is assumed to be absence of clear microscopic evidence for the previous models in
initiated by partial dislocation breakaway, solely based on many other reports, it remains very plausible that grain boundaries
straightforward application of the Peach-Koehler equation for the may be playing a key role in the nucleation of deformation twins as
partial dislocation on either side of a dissociated primary sites for emission of twinning partials. This is the position of de las
Cross slip plane

Screw dislocations pileup
D B
A B
A
A
B
A C BA
BA
BA DC 2 A C
C A Twin
nucleus
Fig. 37. The Miura-Takamura-Narita twinning mechanism [245]. Primary dissociated screw dislocations BA form a pileup against a Lomer dislocation DC. The reaction between the
leading dislocation of the pileup and the Lomer dislocation results in the formation of a sessile Frank partial dislocation and a two-layer twin nucleus.
Cuevas et al. [224] who assert that the deformation twins are of fault formation is entirely controlled by random events in TWIP
nucleated at high-angle boundaries. In their scenario, the emitted steel with an initially perfect fcc crystal structure with an …
twins have a sub-micrometer or nanometer thickness and they ABCABC … stacking sequence in the cubic [111] direction. Electron
traverse a grain until they run into a grain boundary, another twin, diffraction patterns of the perfect g-phase observed along 〈110〉-
or some strong obstacle. type zone axes show only the sharp Bragg reflections of the cubic
The variety of reported microstructural constituents in structure. When a number of random stacking faults is created in
deformed TWIP steel, i.e. faulted austenite, ε-martensite, and this crystal structure, the volume fraction of the regions with the
twinned austenite may be a result of apparent occurrence of hcp stacking sequence will initially be extremely low, but the vol-
ε-martensite and twins in the electron and X-ray diffraction pat- ume fraction of regions with a twin stacking sequence will statis-
terns that stem from clustering of stacking faults [253]. tically be greater. This results in the appearance of Bragg reflections
Indeed, deformation twins are often identified on the basis of corresponding to the twin orientation. There will also be diffuse
electron diffraction in transmission electron microscopy. The effect streaks connecting the Bragg reflections of the matrix and the
of overlapping stacking faults causes diffuse scattering, and the twins. With a higher stacking fault density the probability of hex-
coincidental overlapping of wide stacking faults can result in the agonal stacking increases. The random stacking of cubic and hex-
formation of thin twins and ε-martensite plates even if the process agonal sequences will result in a presence of continuous diffuse
Dislocation pileup
D
D
D
A D A
A
LC lock
C DA
B DA
A
Twin
Dislocation pileup A nucleus
C C
AC
C
Fig. 38. The Cohen-Weertman twinning mechanism [242]. In one of the two dislocation pileups forming a Lomer-Cottrell (LC) lock a primary dislocation at the head of the pileup
dissociates into a Shockley partial dislocation and a sessile Frank partial dislocation. The Shockley partial dislocation moves away from the Frank partial dislocation trailing a
stacking fault. The chance overlap of stacking faults formed in a similar manner results in the formation of a twin.
streaks in the electron diffraction pattern. These gradually develop with the 〈111〉 orientation in tensile direction. At room tempera-
maxima at positions corresponding to Bragg reflections for the ture, deformation twinning in TWIP steels is initiated at the flow
hexagonal crystal structure with an … ABAB … stacking sequence stress corresponding to the onset of stage B of strain hardening
in the hexagonal [0001] direction. The interpretation of the elec- introduced in Section 2.1. A review of the literature shows that
tron diffraction patterns taken from thin foils of strained TWIP there is a small difference between the yield stress and the twin-
steels is therefore subject to ambiguity. These considerations also ning stress of TWIP steels.
suggest that deformation twins may be nucleated at pre-existing Bouaziz et al. [35] reported that for Fe-22%Mn-0.6%C TWIP steel
multi-layer defects, and that the increase to the volume fraction the twin nucleation stress was independent of the grain size, at
of twins is due to the activation of more and more pre-existing twin least in the grain size range of 1.3e25 mm. The magnitude of this
nuclei as the applied stress is increased with strain. The chance stress, approximately 550 MPa, corresponds to the critical twinning
overlap of similar twins will result in an apparent twin thickening. shear stress oftTz180 MPa. Rahman et al. [20] evaluated the
The saturation of the winning process would then occur when all twinning stress and its grain size dependence in cyclic loading with
the pre-existing twin nuclei have been activated. These observa- the stress amplitude below the yield strength. They reported a
tions are consistent with the model for deformation twinning due critical twinning stress of 67 MPa for a single crystal. This value is
to Beyerlein and Tome [254]. These authors proposed a probabi- close to the critical resolved shear stress of 72 MPa reported by Choi
listic theory for twin nucleation in hcp metals, which relies on the et al. [186] for a Fe-22%Mn-0.6%C single crystal oriented to favor
dissociation of pre-existing multilayer grain boundary defects un- deformation twinning, i.e. in compression along the [001]-axis.
der stress. The model emphasises the importance of twin nucle- Classical twinning theories generally predict that the critical
ation, rather than twin growth in deformation twinning. Their resolved shear stress for twinning is proportional to gisf:
mechanism has the following features, which could in principle
also explain the salient features of deformation twinning in TWIP tT $b112 gisf (60)
steel: (a) the number of twin nuclei is related to the grain boundary
Specifically, Narita and Takamura [256] report the following
structure, rather than the dislocation density, (b) twin nucleation
equation for the twinning stress:
does not require a large amount of pre-deformation, (c) the twin-
ning kinetics are nucleation-controlled, rather than growth- gisf
controlled, and (c) the twin volume fraction at saturation is gov- tT ¼ (61)
2$bP
erned by the initial density of grain boundary twin sources. Similar
to the model by Beyerlein and Tome [254], the phenomenological where bP is the Burgers vector of a partial dislocation. Application of
model proposed by Vinogradov et al. [255] is based on the idea that Eq. (58) to Fe-22%Mn-0.6%C (gisf ¼ 23 mJ/m2, bP ¼ 0.147 109 m)
deformation twinning is stress-driven, rather than strain-driven. and Fe-18%Mn-0.6%C-1.5%Al (gisf ¼ 30 mJ/m2, bP ¼ 0.147 109 m)
The model provides a closed-form constitutive description for yields 78 MPa and 102 MPa, respectively. Suzuki and Barrett [257]
materials exhibiting twinning-induced plasticity and is considered assumed that a small segment of dislocation consisting of a sessile
suitable for TWIP steels. partial and a twinning partial pinned at two points act as a twin
nucleus. They defined the twinning stress as the critical stress for
the loss of stability of the twinning partial bowing out under the
5.2. Critical stress for twinning
shear stress:
The formation of deformation twins is generally considered to gisf G$bP

be preceded by dislocation activity which creates the conditions for tT ¼ þ (62)
2$bP L0
dislocation-mediated twin nucleating processes, such as the Ven-
ables' pole mechanism [95], the Cohen-Weertman deviation pro- Here L0 is the length of the sessile partial dislocation engaged in
cess [242], or the Mahajan-Chin stacking fault process [244]. While the twin nucleus. Assuming a pole mechanism is operating, Mahato
various equations are available to determine the twinning stress et al. [19] also used the Suzuki-Barrett equation to calculate the
theoretically for specific deformation twinning models, the exper- twinning shear stress, but they omitted the factor 2 in the de-
imental determination of the twinning stress of TWIP steels is nominator of the first term in Eq. (62).
notoriously difficult. Experimentally measured values of the critical Grain size reduction is generally believed to make deformation
stress for the onset of deformation twinning are listed in Table 8. twinning more difficult. Meyers et al. [258] quantified that trend in
Under tensile deformation, twinning tends to be initiated in grains terms of a grain-size dependent twinning stress equation:
Table 8
Reported values of the critical twinning stress for TWIP steels.
Alloy Twinning stress Grain size gisf Reference

MPa mm mJ/m2
sT (tT)a
Fe-27%Mn-2.5%Si-3.5%Al 272 (89) e 18 Mahato et al. [19]
Fe-22%Mn-0.5%C 272 (89) - e Bracke et al. [17]
Fe-22%Mn-0.6%C 220 (72)b e 23 Choi et al. [186]
Fe-22%Mn-0.6%C 550 (180) 1.3e25 23 Bouaziz et al. [6]
Fe-22%Mn-0.1%C-1.2Al 251 (82) 15 16 Present work
Fe-17%Mn-0.4%C-1.3%Al 312 (102) 15 17 id.
Fe-18%Mn-0.6%C-1.5%Al 422 (138) 5 25 id.
Fe-15%Mn-0.7%C-2%Al-2%Si 190 (62) 84 30 Rahman et al. [20]
Fe-15%Mn-0.7%C-2%Al-2%Si 352 (115) 10 30 id.
Fe-15%Mn-0.7%C-2%Al-2%Si 563 (184) 4.3 30 id.
a Assuming sT ¼ 3:06,tT .
b [001] single crystal in compression (twinning favored).M

gisf K 2$q ð1 nÞ$L 2 gisf
tT ¼ þ pTffiffiffiffi (63) 1 þ $q $tT $tT ¼ (67)
bP D 3$l 1:84$G$b bP
where D is the grain diameter. The influence of the grain size on the Here q is an orientation-dependent term. It is the ratio of the
critical shear stress for twinning in TWIP steel is not entirely clear. Schmid factors for slip and twinning. bP is the Burgers vector of a
While according to Phiu-on [259] the grain size has no effect on tT , partial a/6<112>-type dislocation, and b is the Burgers vector of an
Mohammed et al. reported an increase in the critical shear stress for un-dissociated a/2<110>-type dislocation. l is a parameter gov-
deformation twinning with decreasing grain size [260]. Gutierrez- erned by the forest dislocation density and the critical resolved
Urrutia et al. [27] also observed that, in the grain size range of shear stress for slip. In practice, qzl z 1. A simplified version of the
3e50 mm, a grain size reduction resulted in an increase of the (parabolic) Venables equation for the gisf-dependence of the critical
twinning stress for a Fe-22%Mn-0.6%C TWIP steel. They proposed resolved shear stress for twinning has been derived by Remy [82]:
the following grain size dependent twinning stress equation:
1 1 t gisf
pffiffiffi$ þ K$ ¼ (68)
gisf
G$bp 3 3 G Gb
tT ¼ þ (64)
bp D
where K is a numerical constant. The results for a Fe-18%Mn-0.6%C-
The first term in the equation is the stress required to achieve 1.5%Al TWIP steel are compared to Eq. (65) in Fig. 39. The symbols
partial dislocation separation to create a twin and the second term represent the measured twinning stress at 25 C, 50 C and 75 C
is the stress it takes to nucleate a twin at a grain boundary. Using plotted against the calculated values of gisf. The twinning stress is
tT ¼ 0.447 sT, gisf ¼ 22mJ/m2, G ¼ 62 MPa and bP ¼ 0.147 nm, the defined as the stress at the onset of stage B strain hardening. The
twinning stress for Fe-22%Mn-0.6%C according to Eq. (64) due to twinning shear stress was calculated by using the average Schmid
Gutierrez-Urrutia et al. [27] is given by: factor of 2.238 for fcc crystals. The data agree well with the Ven-
gisf G$bp 22$103 Nm=m2 62$109 Pa$0:147$109 m

tT ¼ þ ¼ þ ¼ 154:3 MPa
bp D 0:147$109 m 2$106 m (65)
sT ¼ 2:238$tT ¼ 345:3 MPa
It is reasonable to assume that Mahajan-Chin three-layer twin ables equation for K ¼ 200. Note that the parameter K was calcu-
nuclei are preexisting multi-layer defects, residing, for instance, at lated to be 900 by Venables, and that Remy and Pineau gave a value
grain boundaries, which are activated when the stress reaches a of 600 and 700 [96]. According to Bouaziz et al. [35] the twinning
critical value. Steinmetz et al. [41] assumed that a small segment of initiation stress of a Fe-18%Mn-0.6%C-1.5%Al TWIP steel is in the
a Mahajan-Chin three-layer twin between two pinning points acts range of 501e574 MPa, i.e. tT is in the range of 164e188 MPa. That is
as twin nucleus. The twinning stress, defined as the critical stress to say, for G ¼ 65 GPa the ratio tT/G is in the range of 2.5 103-
for the formation of an unstable bow-out of the three twinning 2.9 103 - in excellent agreement with an estimate in terms of the
partials, is expressed by the following equation: Venables model. As mentioned in Section 3.2.2, according to Buyn
[146] twinning is activated when the applied stress is high enough
to initiate breakaway of partial dislocations into which a straight
gisf 3$G$bP screw dislocation segment is split. The twinning stress is defined as
tT ¼ þ (66) the critical stress at which the distance between the partial dislo-
3$bP L0
cations diverges. It is given by Eq. (41).
This equation yields very reasonable values for the critical Fig. 40 illustrates the effect of an externally imposed shear stress
twinning stress for Fe-22%Mn-0.6%C TWIP steel, suggesting that on the separation of two partial dislocations in a TWIP steel with
twins are initiated in the early stages of deformation. The reported gisf ¼ 30 mJ/m2 based on Buyn's equation for the distance d be-
values for the critical twinning stress based on Eq. (66) are tween the partials:
compiled in Table 9.
According to Venables, the twinning stress tT is comprised by a ð2 3$nÞ G$b2P
twin nucleation stress and a twin growth stress [248,249]. It can be d¼ $ (69)
8$p$ð1 nÞ g t$bP
calculated by solving the following equation: isf 2
Table 9
Reported values of the twinning shear stress of TWIP steels based on Eq. (63).
tT sT sT gisfe G L0 Ref
MPa 2.24 tT 3.06 tT mJ/m2 GPa nm
MPa MPa
Fe-22%Mn-0.5%C 160 358 490 23 63.6 260 Kang et al. [189]

Fe-18%Mn-0.5%C 160 358 490 16 73.2
Fe-24%Mn-0.7%C 183 410 560 24 62.4
Fe-22%Mn-0.6%C 192 430 588 29 52.5 260 Steinmetz et al. [41]
Fe-27Mne2.5Sie3.5Al 201 450 615 18.8 66.0 200 Mahato et al. [19]
7
25ºC 50ºC 75ºC
6
5
G, x103
4
T/
1 Fe-18%Mn-0.6%C-1.5%Al
0
0 1 2 3 4 5 6 7 8
isf / Gb, x103

Fig. 39. Graphical representation of the Venables equation for the gisf-dependence of the critical resolved shear stress for deformation twinning in a Fe-18%Mn-0.6%C-1.5%Al TWIP
steel (solid line). The following parameter values were used: gisf ¼ 31 mJ/m2, G ¼ 79 GPa, n ¼ 0.312, a/6<112>-type Burgers vector bP ¼ 0.147 nm, based on the lattice parameter
equation aðnmÞ ¼ 0:3575 þ 2:89,105 ,%Mn þ 8:44,105 ,%C, dislocation pileup length L ¼ 134 nm, and K ¼ 200.
from which Eq. (41) follows when the denominator vanishes. The
figure shows that the twinning stress predicted by Byun's model is
p p gisf

much larger than the experimentally measured critical twinning tT ¼ $ gutf 2$gtsf z $ gusf (71)
stresses. The Steinmetz et al. [41] model is closest to the experi- bP bP 2
mentally determined values. This is an important observation as it
where gutf is the unstable and 2gtsf the stable twin fault energy. As
implies that (a) the critical dislocation density required for twin-
nucleating dislocation-dislocation interactions is achieved after
gisf ¼ 2gtsf , and assuming the universal scaling law expressed by
g
small amounts of strain and that (b) twinning is very easily initiated Eq. (32), one has gutf ¼ gusf þ 2isf . Using the zero Kelvin values
in TWIP steel. Alternatively, the result can also be interpreted as an available for critical energy barriers in TWIP steels listed in Table 10,
indication that there are pre-existing multi-layer twin nuclei in the one obtains an estimate tT ¼ 14 GPa for homogeneous nucleation of
recrystallized microstructure. An obvious likely location for these twins. This figure is excessively large compared with the experi-
pre-existing nuclei are grain boundaries where Mahajan-Chin type mental value for the critical twin nucleation stress. It is thus
three-layer twin nuclei may be part of the grain boundary defect obvious that twin formation requires a heterogeneous twin nucle-
structure, as suggested by Steinmetz et al. [41]. ation at pre-existing lattice defects.
More recent deformation twinning theories have proposed that A quantitative equation for the onset stress for twinning, which
it is gusf, the unstable twinning fault energy, rather than gisf, which incorporates the barriers to twinning obtained from ab initio cal-
governs twinning nucleation: culations of the GPFE surface by Kibey et al. [261] reads
tT $b112 gutf (70)

2 3$N 2$gtsf þ gisf
This relation has been verified for pure metals [142] and for Cu-
tT ¼ $ 1 $ gutf þ
3$N$bP 4 2
Al alloys [261] by ab initio simulations using the VASPePAW-GGA (72)
2
software. Such analysis has not been applied to TWIP steels as yet. $ gusf þ gisf
3$N$bP
The ‘ideal’ twining stress associated with the homogeneous
nucleation of a deformation twin in a defect-free crystal, not Here N denotes the number of planes involved in a nascent twin.
involving a heterogeneous twin nucleus such as a three-layer twin Taking into account that for a three-layer Mahajan-Chin twin nu-
nucleus, can be computed using calculated generalized planar fault cleus N ¼ 3 holds, and assuming that gisf ¼ 2gtsf applies, Eq. (72)
energy (GPFE) curves [140,142]. The relevant equation reads can be simplified to
bP2 100
Experimental
d( )
90
bP1
80
Buyn 70
Sessile dislocation
Pinning point 60
d, nm
L
50
Suzuki-Barrett
Steinmetz
bP
Venables
40
Byun
Suzuki-Barrett 30
Sessile dislocation Pinning points

20
L
10
0 , MPa
0 50 100 150 200 250 300 350 400 450
3 bP
, MPa
Steinmetz 0 200 400 600 800 1000 1200 1400
(a) (b)
Fig. 40. (a) Schematics of twin nucleation models due to Buyn [146], Suzuki and Barrett [257], and Steinmetz et al. [41]. (b) Stress dependence of the partial dislocation separation
d according to Buyn, cf. Eq. (41). The values for the twinning stress according to the Buyn, Suzuki-Barrett, Venables and Steinmetz models are compared with the experimentally
measured value for a Fe-18%Mn-0.6%C-1.5%Al TWIP steel. The following parameter values were used for the calculated twinning shear stress: gisf ¼ 30 mJ/m2, G ¼ 79 GPa,
bP ¼ 0.147 nm, n ¼ 0.312, L ¼ 261 nm [41], and L ¼ 135 nm [95].
5 4 5.3. Twinning kinetics and twinning saturation

tT ¼ $ gutf þ gisf $ gusf þ gisf (73)
18$bP 18$bP
The paramount role of twinning in the deformation of TWIP
steels is unquestionable. However, the direct contribution of
deformation twins to plastic strain is not a significant factor
The first term in Eq. (73) describes the stress it takes to over-
responsible for their exceptional mechanical properties. Under-
come the twin nucleation barrier. The second term accounts for
standing the exact nature of the twinning-related phenomena in
cross-slip effects, as increasing gusf results in the inhibition of cross-
the microstructure and texture development and the attendant
slip, thus promoting twinning through a decrease of tT. Eq. (73) also
mechanical properties is therefore a formidable task. Part of it is
implies that tT increases with increasing gutf, so that twinning be-
understanding the evolution of the volume fraction of twins in the
comes less favorable.
course of plastic deformation. This evolution is known to terminate
Using first principle calculations, Medvedeva et al. [157] showed
in saturation. While there are a few cases in which high saturation
that Al short-range order strongly reduces the unstable stacking
volume fractions of deformation twins [217,262,263] were re-
fault energy gusf, i.e. it facilitates the formation of partials trailing a
ported, most researchers agree that deformation twinning in TWIP
stacking fault, while inhibiting an increase of gisf, as discussed in
steels saturates at relatively low volume fractions. In early work on
Section 3.3. Medvedeva et al. therefore conclude that Al additions
TWIP steels it was not uncommon for authors to report that
should promote planar slip prior to twinning, as observed experi-
deformation twinning was the primary deformation mechanism in
mentally. Progress in the calculation of the non-0K critical energy
low gisf austenitic TWIP steels. Vercammen et al. [88] argued that
barriers for TWIP steel should make it possible to determine an ab
the superior mechanical properties were due to the formation of a
initio-value for tT.
Table 10
Generalized stacking fault energy (GSFE) parameters for pure fcc Fe: gusf, unstable stacking fault energy at b112/2; gisf, intrinsic stacking fault energy at b112; gtsf, unstable twin
fault energy at 1.5b112; 2gtsf, stable twin fault energy at 2b112. It was assumed that the relation 2gtsf ¼ gisf applies.
gisf mJ/m2 gisf mJ/m2 gmax mJ/m2 2gtsf mJ/m2 Reference
393 Kibey et al. [156]

464<gisf < 452 Abbassi et al. [155]
350 Gholizadeh [74]
359 þ490 þ3191 359 Limmer [185]
347 Medvedeva et al. [157]
layered nano-scale microstructure. Early TWIP effect models by Here FT,sat is the saturation twin volume fraction. Reported
Bouaziz et al. [33] also led to the conclusion that of all possible saturation values, FT,sat, are usually less than 0.20, and typically
deformation mechanisms in TWIP steel, deformation twinning had about 0.15. Obviously, these values are higher than the overall
the most beneficial effect on strain hardening. In their original volume fraction of twins, which is calculated by considering the
model, the evolution of the twin volume fraction with strain is entire population of grains, including those that do not twin. A
described by the equationFðεÞ ¼ 1 em,ε , m being a phenome- larger grain size, which promotes twinning, may result in greater
nological parameter. The twin saturation volume fraction was twin volume fractions inside twinned grains. εi is a critical strain for
found to be 0.69, based on the substitution of m ¼ 1.95 (their the nucleation of deformation twins. Reported values for εi are
published value) and the uniform elongation εu ¼ 0.6 in the usually small. For example, Bracke et al. [17] observed micro-twins
equationFðεu Þ ¼ 1 em,εu . The above saturation value was, how- already at a true strain of 0.02. b and mT are adjustable parameters
ever, revised downward to 0.2 in later work [35]. In reality, the which need to be determined experimentally. Table 11 lists typical
volume fraction of deformation twins in TWIP steels remains small parameter values for Fe-18%Mn-0.6%C-x%Al TWIP steel at 293 K
and εT, the contribution of the twins to the macroscopic strain, is [266]. Fig. 41 shows twinning kinetics data reported in the litera-
very small. Indeed, assuming a twin volume fraction fT z 0.1 and a ture for various TWIP steels. While Eq. (73) is used in models for the
pffiffiffi
twinning strain of 2=2, it can be estimated as mechanical properties of TWIP steel, it is difficult to apply it to
pffiffiffi experimental data, as the twins are commonly very small in size.
1 2 Instead, the fraction of twinned grains is often used to describe the
εT ¼ $ $fT z0:023 (74)
M 2 twinning kinetics.
Similarly, deformation twinning does not appear to contribute As discussed above, deformation twins are often nucleated
directly to the evolution of the crystallographic texture during preferentially at grain boundaries, which are known to act as sources
deformation. The contribution of deformation twinning to strain for overlapping stacking faults. Most grains tend to develop one
hardening is therefore chiefly due to its influence on the dislocation main twinning system in addition to the dislocation glide. Often in a
density evolution through the reduction of both the dislocation TWIP steel long primary twins occur, which span a grain from one
mean free path and the dislocation annihilation rate. The latter grain boundary to another. Secondary twins may be nucleated
effect is associated with planar glide promoted by twinning. within the twins of the primary system, where they remain
Three major factors have been reported to affect deformation confined. The deformation twins in TWIP steels are usually very thin,
twinning: the stacking fault energy, the grain size, and the crys- typically 20e30 nm in thickness. When observed in transmission
tallographic orientation of a grain. The role of these factors is a electron microscope, most primary twins appear to have propagated
subject of intense experimental and theoretical research. Karaman easily across entire grains without encountering many obstacles. It is
et al. [264] proposed a model for the deformation of Fe-12%Mn-1.2% therefore rather surprising that the deformation twins in TWIP
C Hadfield steel using a viscoplastic self-consistent approach. Their steels do not thicken appreciably once they are formed. TEM ob-
constitutive model incorporates the twin lamellae spacing and the servations have revealed that what appears to be wide deformation
grain size. In the spirit of the Kocks-Mecking-Estrin model [11], twins in optical metallography are in fact bundles of nanometer
theirs employs the dislocation density as a state variable within a thick twins. The relation between nP, the number of partial dislo-
self-consistent approach and on this basis predicts the flow stress cations, and tT, the thickness of the twin they form is given by:
as a function of strain. The predicted saturation twin volume frac-
pffiffiffi tT
tion for coarse grained polycrystals with a grain size of 100 mm was nP ¼ 3$ z4:37$tT (76)
approximately 0.3. The self-consistent approach used is potentially ag
suited for the prediction of polycrystalline plasticity, particularly The twins do not thicken to any sizeable degree and retain their
texture evolution. Although the authors reported good agreement mean thickness of 20e30 nm. The formation of twins with a thick-
between experimental measurements and the model predictions ness of this order of magnitude involves glide of a/6<112>-type
with regard to the variation of the twin volume fraction, it may twinning partial dislocations on approximately 100 successive {111}
have been overestimated in the experimental measurements by planes each. This value is surprisingly close to the number of dislo-
light microscopy, since bundles of very thin twins formed may have cations emitted by Frank-Read sources in the avalanches predicted
been identified as thicker twins. by the Fisher-Hart-Pry model for slip band formation [269]. In this
Other literature data suggests saturation volume fraction of model the avalanches are terminated by their own back stress.
deformation twins of lower magnitude, about 10% [71,265e267], Transmission electron microscopy observations show that
while Haase et al. report a saturation volume fraction of twins in a deformation twining does not result in perfect twins with straight
cold rolled TWIP steel as low as 5% [268]. coherent twin boundaries. Usually, the twins are faulted and they
It should be stressed that not all grains undergo twinning even often have a variable thickness. Their formation appears to be a
under the conditions when deformation twinning is possible in result of a disorderly overlapping of stacking faults. Although no
principle. Thus, according to Kim et al. 20% of the grains do not twin pronounced strain dependence of the twin thickness was found in
during deformation in uni-axial tensile tests [18]. The twinning the carbon-free Fe-31%Mn-3%Al-3%Si TWIP steel [262], close in-
kinetics in the grains which do exhibit deformation twinning sat- spection by means of TEM reveals some variation of thickness. Still,
urates at a volume fraction of approximately 15%. This saturation of
the twinning activity is very likely due to an increase of the dislo-
cations density. Table 11
The strain-dependent kinetics of the evolution of the twin vol- Typical parameter values for the strain dependence of the twin volume fraction in
TWIP steels at 293 K.
ume fraction inside the grains which do undergo deformation
twinning, FT(ε), has a characteristic S-shape dependence on the Parameter Fe-18%Mn-0.6%C Fe-18%Mn-0.6%C-1.5%Al Fe-22%Mn-0.6%C
applied strain. FT(ε) has usually been described by means of the [266] [266] [35]
following empirical equation: FT,sat 0.15 0.14 0.15

mT b 7.50 8.00 3.00
FT ðεÞ ¼ FT;sat $ 1 eb$ðεεi Þ (75) εi 0.04 0.03 0.06e0.12
mT 1 1 2
Fe-20%Mn-1.2%C ( isf=39mJ/m2) Fe-18%Mn-0.6%C-1%Al

Fe-18%Mn-0.6%C ( isf=17mJ/m2) RT
Fe-18%Mn-0.6%C-1.5%Al ( isf=31mJ/m2) 100ºC
100 Fe-18%Mn-0.6%C-1.5%Al ( isf=31mJ/m2) 100
Volume percentage of twinned grains, %

200ºC
Percentage of twinned grains, %
isf=31mJ/m
2
80 80
isf=44mJ/m
2
60 60
isf=63mJ/m
2
40 40
20 20
RT
0 0
0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
20 Fe-22%Mn-0.6%C (Allain, 2004) 10 Fe-23%Mn-0.3%C-1.5%Al (Haase, 2014)

Fe-22%Mn-0.6%C (Barbier, 2009) Cu-Twin texture component volume percentage
Twinned volume percentage, %
Twinned volume percentage, %

Cu-Twin volume percentage, %
Fe-18%Mn-0.6%C-1.5%Al (Kim, 2009) Twinned volume percentage in all grains

16 Fe-17.5%Mn-0.56%C-1.4%Al (Soulami, 2011) 8 Room temperature
Room temperature
12 6
8 4
4 2
0 0
0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
(c) True strain (d) Cold rolling reduction, %
Fig. 41. (a) Strain dependence of the number fraction of twinned grains in TWIP steels based on data by Renard et al. [23], Beladi et al. [78], and Kim [266]. (b) Strain dependence of
the volume fraction of twinned grains in TWIP steels at different temperatures based on data by Shterner et al. [270]. (c) Strain dependence of the volume fraction of twins in TWIP
steels based on data by Allain [71], Kim [266] and Soulami et al. [267]. (d) Effect of the thickness reduction by cold rolling on the measured volume fraction of the Cu-Twin texture
component (cf. Section 5.4) and the calculated value of the corresponding deformation twin volume fraction [268].
the twin thickness typically does not exceed 50 nm, its mean value and cold rolling of TWIP steels has been studied extensively.
staying within the range of 20e30 nm. An increase in the twin Although the development of the Brass-type texture is relatively
volume fraction with strain should therefore be associated with the well documented, there is still some uncertainty about the exact
formation of new twins. Continual nucleation of new deformation contribution of dislocation glide, deformation twinning and shear
twins results in an increasing fragmentation of the grains by twin banding to the development of this specific texture in TWIP steels.
lamellae in the process of straining. Since TWIP steels are low-to-medium gisf fcc alloys, there has been
The temperature has a significant influence on the stress a considerable interest in analyzing whether the material under-
required for deformation twinning. According to Shterner et al., an went the well-known texture transition between the two main
increase in temperature from room temperature to 100 C results in types of fcc rolling textures, i.e. the Cu-type and the Brass-type
an increase of the strain at which deformation twinning is initiated textures, and how the texture development was affected by the
from 4% to 20%, a lower density of deformation twins, as well as a steel composition, the initial texture, the rolling conditions, and the
decrease of the fraction of twinned grains [270]. recrystallization processes. A Cu-texture typically develops
during rolling in high gisf metals and alloys which deform by
5.4. Influence of dislocation slip and twinning on texture octahedral {111}〈110〉 slip and cross-slip. The Cu-texture is a com-
development bination of several texture components: a relatively high fraction of
S-{123}〈634〉 component, and equal amounts of the Cu-{112}〈111〉
The texture development during uniaxial tensile deformation and the Brass-{110}〈112〉 components. Deformation twinning in
TWIP steels appears to transforms a Cu-type texture into a Brass- with the 〈100〉//td undergo deformation twinning [78,270]. An in-
type one. Analysis of this transition provides insights into the intermediate twinning behavior is observed for the 〈110〉//td oriented
fluence of deformation twinning on the development of texture grains. In contrast to 〈111〉//td and 〈100〉//td grains, those with the
and mechanical properties. The Brass-texture is the stable end 〈110〉//td orientation are less stable and characterized by a
rolling deformation texture when both octahedral slip and defor- considerable re-orientation during straining. The grains rotate to-
mation twinning are active deformation mechanisms. It develops in wards orientations with a low and a high Taylor factor.
low gisf fcc metals and alloys and consists mostly of the Brass-{110} In the classical picture of the response of single crystals of high
〈112〉 and Goss-{110}〈001〉 texture components, with a minor, gisf metals and alloys with an fcc crystal structure, the tensile axis
almost negligible, fraction of Cu-{112}〈111〉 and S-{123}〈634〉 rotates to the primary slip direction during tensile deformation.
texture components. The Cu-type to Brass-type texture transition Slip occurs on one primary slip system until the crystal orientation
can therefore be evaluated on the basis of the presence vs. absence reaches the [100]- [111] symmetry line, after which both the pri-
of these two texture components. mary and the conjugate slip system should be equally activated. The
Most reports on the development of crystallographic texture of simultaneous slip results in a crystal rotation towards the [211]-
TWIP steels focus on texture evolution during uniaxial tensile direction. In the [211]-orientation the two rotations cancel each
testing and, to a lesser extent, plane-strain cold rolling. As dis- other and the rotation stops, i.e. [211] is a stable final orientation. In
cussed in the previous section, microstructural observations of low gisf fcc materials the tensile axis often continues to rotate into
TWIP steels have clearly shown that the volume fraction of defor- the conjugate triangle for a while before the conjugate slip starts, a
mation twins is relatively small. As the accumulation of the volume phenomenon usually referred to as “overshooting”. Normally
fraction occurs through the addition of nano-twins of almost non softening should occur due to crystal rotation, but overshooting
variable thickness, the contribution of twinning to crystallographic causes latent hardening, i.e. it appears as if the slip on the primary
texture development is therefore not expected to be a volume ef- system had hardened the conjugate system. This latent hardening
fect, but rather to stem from an indirect effect of the twins on the persists when the conjugate slip system is activated with further
conventional {111}〈110〉-type slip. Models for the crystallographic deformation, as the dislocation mean free path of the primary
texture development in TWIP steel, for example in rolling, should dislocations remains larger than for dislocations of the conjugate
not rely on a large volume fraction of deformation twins to capture slip system. The secondary slip and twinning systems therefore
the observed evolution to a predominantly Brass-type texture require higher stresses for their activation and they are expected to
during deformation [16]. occur at higher strains. In low gisf TWIP steels, the grains develop a
Most authors agree that under uniaxial tensile deformation set of deformation twins which belong to the primary twinning
TWIP steels with a weak initial texture typically develop a double, system. The active primary dislocations and the primary twins act
〈111〉//td and 〈100〉//td, fiber orientation and that a pronounced as barriers to the dislocations of the conjugate slip system which is
mechanical twinning takes place in grains oriented close to the non-coplanar with the primary slip and twinning systems. Defor-
〈111〉//td fiber [21,25,232]. Usually the 〈111〉//td fiber is strong as mation twins thus control the texture evolution at low strains by
compared to the 〈100〉//td fiber (Fig. 42). suppressing multiple slip owing to the inhibition of the conjugate
The grains with a 〈111〉//td orientation develop a high density of slip system. The suppression of multiple slip favors the Cu-texture
deformation twins. By contrast, only a small fraction of the grains generated by standard {111}〈110〉 slip. Hence, deformation twins
max: 3 random max: 5 random
111 111
direction
tensile
Cu
{110}<111>
Brass
TD Goss
Cube RD
{110}<110>
TD
ND
<111>//td
(a) RD (c) RD
100 100
100
TD TD
<100>//td
(b) RD (d) RD
Fig. 42. (a) {111} and (b) {100} pole figures for cold rolled and recrystallization annealed Fe-18%Mn-0.6%C-1.5%Al TWIP steel. The texture is weak and consists of the Cu orientation
and a-fiber 〈110〉//ND orientations. (c) {111}and (d) {100} pole figures for the same steel tested in tension parallel to the rolling direction to an engineering strain of 20%. The grain
re-orientation caused by uniaxial tensile deformation results in the formation of the double fiber texture typical of fcc materials, consisting of a strong 〈111〉//td fiber and a weaker
〈100〉//td fiber (td: tensile direction).
promote overshooting. Piercy et al. showed that a reduction of gisf dislocation glide:
enhances the latent hardening and the overshooting effect [271].
Leffers also argued that the texture transition from Cu-type to
Brass-type is delayed by pronounced planar slip brought about by
tT ¼ tcrss ¼ mT $sA < tG ¼ tcrss ¼ mG $sA /mT < mG (77)
the occurrence of twinning due to the mentioned suppression of Here the Schmid factor has a conventional definition as the
multiple slip by deformation twins [272,273]. product of the cosines of the angles between the direction of the
The initial grain orientation has been reported to have a pro- applied tensile stress sA and the slip plane normal and the direction
nounced influence on twinning and the development of the crys- of shear for a perfect dislocation (mG) or a partial dislocation (mT).
tallographic texture. The effect of grain orientation on twinning has This view is mainly based on the mentioned observation that
been given three different explanations, which are outlined sche- during uniaxial tensile deformation of TWIP steels a strong 〈111〉//
matically in Fig. 43. According to Gutierrez-Urrutia et al. [27] and td and a weak 〈100〉//td fiber orientations develop. Fig. 44 shows
Sato et al. [274] the effect of the grain orientation observed at low that while 〈111〉//td-oriented grains are profusely twinned, 〈100〉//
strain is governed entirely by the relative value of the Schmid td-oriented ones are mostly twin-free. Using the electron chan-
factor, as twinning occurs when the resolved shear stress for neling contrast imaging (ECCI) technique to analyze Fe-22%Mn-
twinning is smaller than the critical resolved shear stress for 0.6%C TWIP steel (with gisf ¼ 22 mJ/m2), Gutierrez-Urrutia and
0.88
0.92 Type I
0.96
mT>mG
1.00
1.10
1.20 Type II
1.40
mT<mG Type III
1.60
(a) (b)
mG mT
3.67
0.36
3.60
0.40 3.40
0.42
0.44 3.20
0.46
[113] 0.48 3.00
0.40 2.80
0.42 2.60
0.44
0.46
2.40
2.45 3.67
0.48 [102] 3.0 3.4 3.6
(c) (d) M
deff < d0 deff > d0 M<2.6

M>2.60
Fig. 43. (a) Schematic illustrating the effect of grain orientation on deformation twinning according to Gutierrez-Urrutia et al. [25] and Sato et al. [274]. (b) Orientation range of Type
I and Type II grains, which behave as expected on the basis of their Schmid factor. (c) Schematic illustrating the effect of grain orientation on deformation twinning according to
Geissler et al. [276] and Kuprekova et al. [277]. (d) Schematic illustrating the effect of grain orientation on deformation twinning according to Beladi et al. [78].
Grain B: no twinning Grain C: moderate twinning Grain F: Pronounced twinning

ND ND ND
{111} {111} {111}
RD RD RD
Cube: {100}<001> Cu: {112}<111> -fiber: {111}<112>
Fig. 44. Illustration of the influence of the grain orientation on twinning in a tensile sample of TWIP steel strained to 50% engineering strain. The pole figures of the individual grains
reveal that the C-grain, which has a Cube {100}〈001〉-orientation, does not contain deformation twins, the B-grain, which has a Cu {112}〈111〉-orientation, exhibits an intermediate
twinning behavior, and the F-grain, which has a {111}〈112〉 (g-fiber) orientation, is heavily twinned. Here RD and ND denote the rolling direction and the normal direction,
respectively.
Raabe [275] evaluated this mechanism experimentally. They shear stress, t, for a grain with a specific orientation and subjected
showed that this strict Schmid law orientation dependence of slip to a specific deformation is defined as the ratio of the sum of the
and twinning in a tensile test does not apply in practice. Only type I shear increments over the active slip systems and the axial strain
grains, which were favorably oriented for twinning due to their increment:
orientation very close to 〈111〉//td, and type III grains, which were
favorably oriented for dislocation glide with an orientation very P
dgi
close to 〈100〉//td, behaved in a manner expected solely on the basis sA i
of their Schmid factor. Indeed, micrometer long twins nucleated at M¼ ¼ (79)
t dε
grain boundaries in the type III grains were also observed.
Furthermore, type II grains, which cover a wide range of orienta- Here sA is the applied uniaxial tensile stress. The larger the
tions unfavorable for twinning based on their Schmid factor, were Taylor factor for a particular grain, the higher is the stress for its
also found to contain deformation twins. Gutierrez-Urrutia and plastic deformation. Grains with larger Taylor factors experience a
Raabe associated the deformation twinning in these unfavorably faster evolution of the dislocation density and strain harden more
oriented grains with stress concentrations and strain gradients at than grains with a low Taylor factor. For Fe-18%Mn-0.6%C-1%Al
grain boundaries. TWIP steel, Beladi et al. [78] report that, under uniaxial tensile
Other authors, e.g. Geissler et al. [276] and Kuprekova et al. testing, grains that had rotated towards 〈111〉//td had a high Taylor
[277], considered the Schmid factor for the individual partial dis- factor, M z 3.67. These grains twinned in the early stages of
locations. They noted that for specific grain orientations the force straining. An example of this behavior is shown in Fig. 44. Grains
on the leading partial dislocation is larger than that on the trailing with a 〈100〉//td orientation having a low Taylor factor, M < 2.6,
partial dislocation of a dissociated perfect dislocation. When this remained twin-free up to fracture.
happens, the effective stacking fault width deff is larger than d0, the Another confirmation of the thesis that multiple slip precedes
equilibrium stacking fault width in the absence of an applied stress. twinning is the paper by Yang et al. [278] on texture develop-
This can result in breakaway of the leading partial dislocation of a ment in a Fe-33%Mn-3%Al-3%Si TWIP steel during uniaxial ten-
dissociated screw dislocation [146] when the applied stress sA ex- sile deformation. The very thin deformation twins formed did
ceeds the level of not contribute to texture evolution. The twinned volume was
reoriented to CuT-{552}〈115〉, Fig. 45, which is an orientation
g favoring slip. Yang et al. also reported that uniaxial tensile
sA ¼ 6:14$ (78) deformation resulted in a strong 〈111〉//td and a weak 〈100〉//td
bp
fiber orientation. Slip caused grain rotation towards the
Beladi et al. [78] observed that whereas grains with an orien- 〈111><100〉 line. They also noticed that tension along a 〈111〉//
tation close to the Goss and Cube orientations, i.e. low Taylor factor td-type direction promoted twinning, in contrast to tension
grains, were free of mechanical twins, those having an orientation along 〈100〉//td, which did not favor twinning. They report that
close to the Brass, S, or Cu, orientations i.e. high Taylor factor grains, orientations near those with the highest Schmid factor for
were twinned. They therefore correlated the tendency for twinning twinning did not twin. Grains with 〈110〉//td orientations were
of individual grains with the magnitude of the Taylor factor for their seldom observed to contain twins even after an engineering
orientation. According to Beladi et al. [78] the nucleation and strain as high as 20%, despite their high Schmid factors for
growth of twins requires the activation of multiple slip systems and twinning. Yang et al. suggested that there were two contributions
the presence of a high stress resulting from dislocation pile-ups. of the TWIP effect to strain hardening: (a) the interaction be-
They report that deformation twinning is initiated at a strain of tween twin variants in the 〈111〉-oriented grains, and (b) the
0.04. The Taylor factor M, the ratio of the axial flow stress and the dislocation-dislocation interactions within the deformation
111 111 111
Deformation Slip
twinning (rotation)
Dislocation Deformation
glide twinning
001 101 001 101 001 101
Strain: 0% 5% 10%
Fig. 45. Orientation changes within a single grain with an orientation close to 〈111〉//td. Twinning is clearly seen to give rise to a volume reorientation. The initially Cu-{112}〈111〉-
oriented grain twins easily to a CuT-{552}〈115〉 orientation. In this new orientation, slip is preferred relative to deformation twinning. {111}〈110〉 dislocation slip results in a gradual
rotation of the grain towards the 〈001><111〉 line in the basic stereographic triangle. The dotted line separates the orientations for which twinning and dislocation glide are the
preferred deformation modes under uniaxial tension.
twins, whose orientation favors dislocation glide. the 〈100〉//td fiber orientation (Fig. 46). In grains with this
Barbier et al. [21] studied the texture development in Fe-22% orientation, dislocation glide is favored relative to twinning as
Mn-0.6%C TWIP sheet steel with 2.6 mm grain size during uniaxial the Schmid factors for twinning and glide are 0.23 and 0.41,
straining along the transverse direction (TD) of rolling that pre- respectively.
ceded the tensile test. The initial texture of their material was weak. 7. As most twins are oriented close to 〈100〉//td, the increase in the
Similar to Yang et al. [278], they observed the development of a intensity of this fiber orientation during straining is directly
pronounced 〈111〉//td fiber and a weaker 〈100〉//td fiber texture. related to the evolution of the twin density.
Twinning started at a true strain of 0.02 (corresponding to a stress
of 550 MPa). In accord with the results of other authors cited above, The results by Prakash et al. are slightly different from the
twinning resulted in the formation of bundles of very thin nano- previous reports reviewed above [279]. The initial texture of their
twins, with a thickness in the range of 10e40 nm. Again, the TWIP steel sheet material had mainly Brass and Goss orientations,
thickness of the twins remained essentially unchanged, whilst the and the S texture component was not a major one. Tensile defor-
bundles thickened by adding on more new twins in the course of mation with the loading direction along the rolling direction raised
straining. The maximum thickness of the bundles was 250 nm. the intensity of the {110}〈111〉 texture component and decreased
Barbier et al. estimated that the twinned volume fraction was 9% at the intensity of the Brass orientation. Goss continued to be a
a true strain of 0.55, again supporting the observation that satu- principal component. Uniaxial deformation along the transverse
ration levels of the twin volume fraction is typically of the order of direction (TD) resulted predominantly in the Goss and Brass texture
10%, see Section 5.3. Four texture components dominated the components, with the Brass texture spreading towards {110}〈111〉.
texture development during straining: Brass-{110}〈112〉, Rotated It has been pointed out that some of the texture analysis was
Cu-{112}〈110〉, Rotated Goss-{110}〈110〉 and Cube-{001}〈100〉. At carried out on TWIP steels which are susceptible to room temper-
high strain, the Brass-{110}〈112〉 component had the highest in- ature dynamic strain aging (DSA), such as Fe-22%Mn-0.6%C [280].
tensity. An important conclusion was that deformation twinning in The increased temperature in the localized deformation bands due
their steel did not generate new orientations, but that it increased to heat dissipation causes gisf to increase, which is expected to
the intensity of existing orientations to various degree. According to enhance the Cu-type texture. Al additions also increase gisf, and this
Barbier et al., deformation twinning affects the texture develop- results in a slightly more pronounced Cu-texture. It is believed that
ment in the following ways: these effects and the pronounced planarity of slip may explain the
persistence of the Cu-{112}〈111〉 texture component in tensile
1. The grain rotation resulting from {111}〈110〉 slip during uniaxial straining, as illustrated by Figs. 47 and 48.
tensile deformation generates a pronounced 〈111〉//td-fiber. The
Brass-{110}〈112〉 and the Rotated Cu-{112}〈110〉 belong to this 5.5. Rolling texture
fiber.
2. The initial recrystallization texture affects the slip activity prior The orientation changes in rolling deformation are more com-
to twinning that sets in when the critical stress for twinning is plex than those occurring under uniaxial tensile deformation. All
attained. Deformation twinning is influenced by this prior slip high gisf fcc metals and alloys are similar in that they deform by
activity and the initial grain orientation. homogeneous {111}〈110〉-type dislocation slip and easy dislocation
3. The Cu-{112}〈111〉 and Goss {110}〈100〉 grains have an orienta- cross-slip. The {111}〈110〉 slip and the grain rotations during rolling
tion favorable for twinning, and the development of the 〈111〉// result in the development of a b-fiber orientation. The b-fiber is an
td-fiber orientations continues to favor twinning. ODF-skeleton line going from the Cu-{112}〈111〉 component, via the
4. Eventually most grains are oriented with 〈111〉//td at a strain of S-{123}〈634〉 component to the Brass-{110}〈112〉 component. It
0.3. This promotes twinning at higher strains. usually consists of a relatively large fraction of the S-{123}〈634〉
5. Twinning also influences the texture development through the component.
generation of new orientations and the modification of the A Brass-{110}〈112〉-type texture is expected as the stable rolling
texture evolution by dislocation glide. texture in low-to-medium gisf TWIP steel, since both octahedral slip
6. New orientations are formed when twins form inside 〈111〉//td and mechanical twinning are active deformation mechanisms
or 〈110〉//td oriented grains. These new orientations are close to (Fig. 49). Deformation twinning and the attendant limited cross-
(a) [111]
IPF
[110]//RD
TD
(b)
IQ
RD//td
3 m
(c)
Slip
Deformation
Twins
True strain: 0 0.1 0.2 0.3

Fig. 46. In situ EBSD analysis of twinning in a grain close to the 〈111〉//td-orientation for a true strain in the range of 0.0e0.3. (a) Inverse pole figure evolution. (b) Image quality (IQ)
contrast of the same region as in (a). Note that the deformation twins are already visible in the IQ image at a strain of 0.1 (Arrows). (c) The pole figures taken at different strain reveal
the grain rotation and the formation of CuT-{552}〈115〉-oriented twinned regions at a strain of 0.2 and 0.3.
slip result in a texture characterized by a-fiber orientations Dillamore [284] associate the texture transition to Brass-{110}〈112〉
spreading form the Brass-{110}〈112〉 to the Goss-{110}〈100〉 with the limited cross-slip in low gisf metals and alloys.
orientation. When the Brass-texture is pronounced, the volume At large strains a lamellar microstructure of the matrix and the
fraction of the Cu-{112}〈111〉 and S-{123}〈634〉 texture components twins parallel to the rolling plane develops. According to Morii et al.
is negligible. According to the classical model for the emergence of [285], slip cannot continue when the slip planes are almost parallel
the Brass-texture in low gisf alloys, deformation twinning plays a to the surface, and this leads to the formation of shear bands.
key role in the texture development. During straining, grains with However, this alignment of the lamellar microstructure of matrix
the Cu-{112}〈111〉 orientation first twin into the CuT-{552}〈115〉 and twins parallel to the rolling plane does not appear to be
orientation. Dislocation glide then causes the CuT-{552}〈115〉-ori- necessary to form shear bands, and there is evidence for shear
ented grains to rotate around 〈110〉//TD towards the Goss-{110} bands forming early on when the rotation of the lamellae is initi-
〈001〉 orientation and the stable Brass-{110}〈112〉 final orientation. ated. Vercammen et al. [88] studied the evolution of microstructure
Hirsch et al. [281,282] proposed that the CuT-{552}〈115〉 oriented and texture of a Fe-30%Mn-3%Al-3%Si TWIP steel during cold roll-
lamellae formed by twinning first rotate to an intermediate Brass- ing. According to that study, a Brass-type texture was already
R-{111}〈112〉 orientation, and then to the Goss orientation during present at small rolling strains. The Brass orientation {110}〈112〉
the shear band formation. Smallman and Green [283] and was predominant at every strain level, whilst the Cu-orientation
1 Strain:0% 20%
fibre<110>//TD
Cu
fibre <111>//ND Cu Twin
Brass
Goss
Cube
Fibre <110>//ND
2 = 45 section
Fig. 47. Evolution of the orientation distribution function (ODF) during tensile straining of a Fe-18%Mn-0.6%C-1.5Al TWIP steel to an engineering strain of 20%, which results in an
enhancement of the Cu texture and S components. A schematic f2 ¼ 45 section of the ODF in the Euler angle space with the main texture components and fibers. Note the a-fiber
texture from Brass-{110}〈112〉 to Goss-{110}〈100〉.
12
{110}<111>
10
Volume percentage, %
Cu
6 Brass
S
4
Goss
Cube
2 CuTwin
0
0.0 0.1 0.2 0.3 0.4 0.5
(a) True strain
True strain:0.0 0.2 0.3
111
5 m
(b) 001 101
Fig. 48. (a) Evolution of the volume fractions of the various texture components during tensile straining of a Fe-18%Mn-0.6%C-1.5Al TWIP steel. (b) Corresponding RD-inverse pole
figure map at different strains. Note the a-fiber texture ranging from Brass-{110}〈112〉 to Goss-{110}〈100〉. The volume fraction of a texture component is obtained by assuming a 10
orientation tolerance around the exact component.
intensity remained low. The absence of a pronounced Cu- described as a potent way to predict their microstructure evolution
orientation texture was related to the early formation of defor- and deformation behavior. Crystal plasticity models are also an
mation twins, as supported by their TEM micrographs. The Goss equally powerful computational tool for describing and predicting
and Cu-Twin orientations developed due to an increase of the twin the crystallographic texture evolution for technologically impor-
volume fraction with strain. Most of the grains were twinned at tant processing operations, such as rolling [31,279,287]. In a poly-
larger values of sheet thickness reduction, the twinning planes crystal plasticity model, the interaction of the grains with their
being reoriented so as to be parallel to the rolling plane, making surroundings have to be accounted for. The classical Taylor
homogeneous deformation difficult. Shear banding was initiated assumption that strain and strain rate are the same in every grain is
because crystallographic slip was inhibited. At a thickness reduc- commonly used, but this full Taylor constraint can be relaxed in
tion corresponding to an engineering strain of 65%, non- more sophisticated models. A challenging part of the modeling is
homogeneous shear band formation became the preponderant that in the case of TWIP steel the models need to take into account
deformation mechanism. The deformation microstructure con- sudden changes in orientation, and make assumptions about the
sisted of bands oriented at approximately þ35 and 35 to the RD. behavior of the twinned volume. One possibility is to consider that
g-fiber {111}//ND texture components, i.e. {111}〈110〉 and {111} the grain population is subdivided in twinned and non-twinned
〈112〉, were formed due to the alignment of twin lamellae parallel to parts and assume that the twinned regions do not contribute to
the rolling plane. plasticity.
The rolling texture development in TWIP steels is akin to that in Barbier et al. [288] modeled the texture evolution and micro-
the austenitic stainless steel grade AISI 316L (gisf ¼ 64 mJ/m2) [286]. structure changes in Fe-22%Mn-0.6%C TWIP steel (gisf ~20 mJ/m2)
during uniaxial tensile and shear deformation. They considered a
5.6. Texture evolution modeling representative volume element (RVE) of 3000 grain orientations.
The texture evolution was found to be mainly controlled by crys-
In Section 2.2, crystal plasticity modeling for TWIP steels tallographic {111}〈110〉 slip, dislocation slip on multiple slip sys-
deforming by dislocation glide and deformation twinning has been tems governing the plasticity and strain hardening. Tensile
deformation along RD and TD induced twinning in most grains. As expected to influence the twinning kinetics in the deformation
the texture developed, more and more grains assumed an orien- bands, have not yet been implemented in crystal plasticity models.
tation favorable for twinning, while twin-twin interactions made it It is therefore not known how DSA-related strain localizations in-
more difficult to produce more twins. fluence the texture evolution in TWIP steels that deform at room
Dancette et al. [32] used idealized truncated octahedron-shaped temperature by the propagation of deformation bands causing
grains as representative volume elements (RVE) used in their cal- discontinuous yielding, such as e.g. the Fe-22%Mn-0.6%C TWIP
culations. By considering the effect of the twin thickness on the steel.
dislocation mean free path, and with the aid of a heuristic equation
for the twinning kinetics in the parent grains, they were able to 5.7. Recrystallization texture
calculate the texture evolution. One finding was that a simple
Schmid factor analysis does not predict the correct amount of Y.Lü et al. [291] studied the texture development in Fe-21.6%C-
twinning in the grains. Furthermore it was found that for the full 0.38%C TWIP steel cold rolled to a 50% rolling reduction and
constraint Taylor model the calculated texture is too sharp in the recrystallization annealed in the temperature range of 560 C to
Brass and Goss region. The results obtained by Dancette et al. at the 700 C. They report that the nucleation of new grains was site
grain level are very informative. Whereas a single slip Sachs-type saturated, i.e. all nuclei were generated at the start of the recrys-
behavior model predicts a Brass-type texture [289], a multi-slip tallization anneal. The nucleation sites were shear bands, grain
Taylor-type model predicts Cu-type textures [290]. boundaries and triple junctions. The growth rate of the nuclei
An important issue with the crystal plasticity modeling of C- decreased with increasing annealing time, which the authors
added TWIP steels is dynamic strain aging, which makes the associated with recovery by pipe-diffusion controlled dislocation
macroscopic deformation nonhomogeneous, see Section 6 below. climb and a decrease of the driving force for grain boundary
The associated serrations in the stress-strain curves are often migration. The grain structure after annealing was inhomogeneous
ignored or simply filtered out [32]. These effects, which are due to a non-random distribution of nucleation sites. In the initial
Cu
Fibre
Cu Twin <111>//ND
Brass Fibre <110>//ND
Goss
Cube
Rolling reduction: 0% 5% 10% 20% 30% 40%

(a)
Brass
5
4
random
Goss
3
Intensity,
Cu
1
Cube
0
0 10 20 30 40
(b) Cold rolling reduction, %
50μm
(c) Strain: 0% 5% 10%
Fig. 49. (a) Evolution of the ODF during cold rolling of a Fe-22%Mn-0.6%C-0.08%N TWIP steel. (b) Evolution of the volume fraction of the main texture components presented in (a).
(c) IQ and IPF maps showing that the main twinning activity occurred in the deformation range of 5e10%. Note the a-fiber texture spreading from Brass-{110}〈112〉 to Goss-{110}
〈100〉 and the weakening of the Cu-{112}〈111〉 component at large strains. The rolling texture can be described as a Brass-type {110}〈112〉-texture with a spread towards the Goss-
type texture {110}〈001〉.
stages of annealing, i.e. prior to the onset of recrystallization, the S- rate sensitivity parameter, m ¼ dd ln s ¼ S, is related to the activation
ln_ε s
{123}〈634〉 texture component increased in intensity for a short volume of the underlying mechanism governing dislocation glide,
period of time due to the transformation of the strain-induced which may be associated with thermally activated overcoming of
ε-martensite to austenite [292]. The intensity of the dominant dislocation forest junctions, stationary solute atoms or their clus-
texture components developed as a result of cold rolling, i.e. the S- ters, or the Peierls barriers. In all these cases m is positive. However,
{123}〈634〉, Brass-{110}〈112〉, and Goss-{110}〈100〉 components, under certain conditions, mobile solutes segregating to mobile
gradually dropped off during annealing. The texture after full dislocations temporarily arrested at localized obstacles may addi-
recrystallization was similar to the deformation texture, but was tionally pin them. This pinning effect, known as dynamic strain
very weak. Texture was also weak after annealing above 630 C in aging (DSA), is greater for lower dislocation velocities or plastic
not fully recrystallized state. Un-recrystallized domains had a Goss- strain rates, which means that its contribution to stress has a
{110}〈100〉 orientation. These grains are known to have a low stored negative strain rate sensitivity. Indeed, the contribution to stress
energy after cold work [293]. The texture was more pronounced due to this pinning effect increases with solute concentration cs on
after annealing at lower temperatures. For example, after annealing the dislocation, which in turn increases with the waiting time tw it
at 560 C the intensity of the a-fiber 〈110〉//RD and the 〈100〉-fiber takes the dislocation to overcome the obstacle it is arrested at. As tw
increased due to selective growth of grains with these orientations. is inversely proportional to the dislocation velocity, an increase in
Bracke et al. [294,295] studied the recrystallization texture of the velocity brings about a drop in the DSA contribution to stress.
annealed cold rolled Fe-22%Mn-0.6%C, an alloy with a low stacking- The competition between the two contributions to stress having
fault energy (gisf ¼ 15 mJ/m2). They contrasted it with the texture in different signs of the strain rate sensitivity may result in an overall
a hot rolled strip, which was weak with a low intensity Cube- negative one. The schematic in Fig. 50 illustrates how a negative
100<100> component, which is actually a recrystallization texture strain rate sensitivity component gives rise to an N-shaped s ln_ε
typical for high gisf fcc alloys. For cold rolling, the TWIP steel curve. According to the model proposed by Kubin and Estrin [37],
investigated showed a deformation texture evolution behavior the dislocations move slowly dragging along a Cottrell cloud of
characteristic of low gisf alloy. This assessment is based on two solute atoms in the low strain rate branch of the curve, i.e. in the
observed features: (a) a decrease of the intensity of the b-fiber solute drag region. In the high strain rate branch of the curve, i.e. in
component, and (b) the development of an a-fiber with a high in- the lattice friction region, the diffusivity of solutes is insufficient for
tensity between the Brass and Goss component. The presence of the them to catch up with the rapidly moving dislocations, which thus
CuT-552<115> texture component was also considered as a direct experience only the effect of the lattice friction and the stationary
proof for twinning and an evidence that twin formation leaves a solutes. In the intermediate strain rate range, represented by the
footprint in the texture. Bracke et al. [294,295] also observed a new descending part of the curve in Fig. 50, plastic deformation is
texture component they refer to as the “X-component”, which is inherently unstable. In this regime, propagating localized defor-
twin related to the Brass component. In their annealing experi- mation bands are formed [297]. A large volume fraction of the
ments on cold rolled steel, a slightly higher temperature (725 C) material deforms at a low strain rate whilst a small one, viz. the
than in the work by Y.Lü et al. [291] (700 C) was used. They found a material within the localized deformation bands, deforms at a high
recrystallization texture which was almost identical with the cold strain rate. This extreme manifestation of dynamic strain aging in
rolled texture, except for a marginal increase of the Goss-{110}〈100〉 the negative strain rate sensitivity of the flow stress, referred to as
component in the recrystallized steel. The texture did not depend the Portevin-Le Chatelier effect, also occurs in C-alloyed TWIP
on the reheating rate. The high nucleation density observed at the steels. Propagation of localized deformation bands accompanied
start of the recrystallization was associated with site-saturation with serrations on the stress-strain curve was observed in specific
nucleation. The nucleation was homogeneous throughout the ranges of temperature and strain rate where the strain rate sensi-
microstructure. At high rolling strains, the recrystallization was first tivity of the flow stress is negative, i.e. S < 0 holds. When a strain
initiated within the shear bands. Accordingly, no strong orientation rate from within this range, ε_ 1 < ε_ < ε_ 2 ; is imposed, the system will
selection occurred during nucleation and growth of the recrystal- follow the A-B-C-D stress-strain rate cycle shown in Fig. 50, thus
lized grains. Once recrystallized grains nucleated, their growth was avoiding the interval (_ε1 ; ε_ 2 Þ, while maintaining an average strain
sluggish, due to the impingement of the growing grains soon after rate equal to the externally imposed one. In TWIP steels, the strain
their formation. This process resulted in a small grain size. The rate within a localized band is typically one order of magnitude
results imply that the stored energy of deformation was distributed larger, while that outside the band is one order of magnitude lower,
homogeneously. Grain growth and the formation of recrystalliza- than the applied strain rate.
tion twins must have occurred in a random way, too. As the The strain rate sensitivity being slightly negative is one of the
nucleation process involved a random sampling of the orientations few disadvantageous mechanical properties of C-alloyed TWIP
in the deformed microstructure, the recrystallization texture steels. In other fcc metals and alloys micro-twins and nano-twins
replicated the initial deformation texture. were shown to increase the strain rate sensitivity considerably.
The observations of a small grain size and a pronounced cold For instance, for ultra-fine grained nano-twinned Cu, with a grain
rolled texture made by Bracke et al. [294,295] are not supported by size in the range of 400 nme500 nm, the strain-rate sensitivity
Barrales et al. [296]. The latter authors found a coarser hot rolled parameter m is significantly higher as compared to ultra-fine
grain structure, which they believe may have resulted in texture grained Cu without nano-sized twins, 0.037 compared to 0.005,
randomization and a coarser grain size after recrystallization respectively. The strain rate sensitivity parameter was also reported
annealing that followed cold rolling. to increase with decreasing twin spacing [298].
The main macroscopic manifestations of the Portevin-Le
Cha ^telier (PLC) effect are as follows: (a) the strain rate sensitivity
6. Dynamic strain aging and strain rate sensitivity in carbon- of the flow stress is negative, (b) the stress-strain curve is serrated,
alloyed TWIP steel (c) plastic strain is localized in PLC deformation bands, and (d) the
post-uniform elongation, as measured in a uniaxial tensile test, is
Plasticity controlled by thermally activated dislocation motion very limited.
discussed in Section 4.4 is normally associated with a positive strain The jerky flow associated with the serrated stress-strain curves
rate sensitivity of the flow stress defined as S ¼ ds=dlnε:_ The strain
d d d
0 0 0
d ln( ) d ln( ) d ln( )
A B
Stress
D C
Solute ln( 1 ) ln( 2 ) Lattice Strain rate

dragging friction
v1 v2 Dislocation velocity
Fig. 50. Schematic s ln_ε curve with a region of negative strain rate sensitivity, ds=d ln_ε < 0. In the strain rate interval between ε_ 1 and ε_ 2 , uniform deformation is unstable. In the
low strain rate condition, ε_ < ε_ 1 , the dislocations drag along an atmosphere of solute atoms. In the high strain rate condition, ε_ > ε_ 2 ; the dislocations move without a solute at-
mosphere, and experience only a resistance from the lattice friction and the stationary solutes. A regime characterized by localization of strain in deformation bands corresponds to
an imposed strain rate falling in the interval (_ε1 ; ε_ 2 Þ.
in uniaxial tension is an indication of non-uniformity of deforma- imposed strain rate [301]. Accordingly, the temperature was higher
tion. Strain localization takes place in propagating or static defor- inside the bands than outside of them. Close to fracture the strain
mation bands. From a technological point of view, it is important to within a band increased to ~0.090, whilst the band velocity drop-
avoid DSA, in order to ensure stable material flow during sheet ped to ~4 mm/s. The band displacement became progressively
forming processes. Strain non-uniformities also give rise to visible smaller as the strain rate decreased and eventually type B bands
surface defects on press-formed parts. DSA also leads to the occurred in a correlated manner.
inability of C-alloyed TWIP steels to sustain large post-uniform Type B bands are stationary, but as a new band emerges ahead of
strains after the initiation of necking. It has also been linked to a previously formed one in a ‘relay-race’ fashion, they appear to
premature ductile fracture occurring during stretch-flanging of have a hopping behavior. Whereas the propagation velocity of type
TWIP steels, as DSA leads to rapid edge cracking during hole A bands can be up to hundreds mm/s, the effective propagation
expansion [299,300]. It is therefore important to suppress the velocity of a B band pattern is very low, of the order of tens mm/s.
occurrence of DSA- or PLC-related instabilities of plastic flow by Mainly type A and type B bands have been observed for TWIP steel.
defining suitable strain rate and temperatures for processing. In Type C bands are static bands nucleated at random, in a non-
particular, the temperature dependence of the critical strain for the correlated way, over the entire sample. They are characterized by
onset of serrations on the stress-strain curve needs to be known. sharp stress drops. Type C bands are usually formed at very low
The temperature dependence of the critical strain for Fe-18%Mn- strain rates (typically 105s1). Qian et al. [302] observed type C
0.6%C-1.5%Al steel for a particular level of the imposed strain rate serrations on the flow curves of Fe-22%Mn-0.6%C TWIP steel tested
is shown in Fig. 51. in the strain rate range of 6 106s1-104s1.
Several types of PLC bands are considered, according to their DSA in Fe-24%Mn-0.6%C (gisf ¼ 31 mJ/m2) and Fe-27%Mn-0.6%C-
appearance, character of propagation, and statistics [297]. Three 3.5%Al (gisf ¼ 50 mJ/m2) was studied by Saeed-Akbari et al. [303].
major types observed during uniaxial tensile tests are labeled A, B, They detected persistent presence of instabilities even in appar-
and C. Type A bands appear as successive solitary bands which, in ently smooth stress-strain curves. As TWIP steels have a relatively
the case of TWIP steels, are usually nucleated at or close to one of low thermal conductivity, B€ aumer and Bleck suggested that adia-
the ends of the tensile sample and propagate across the entire batic heating plays a significant role during straining at higher
sample length, as illustrated by results on Fe-18%Mn-0.6%C TWIP strain rates [304]. They stress the importance of the effect of
steel in Fig. 52a. Type A bands are observed at relatively high strain adiabatic heating on the magnitude of gisf and the deformation
rates (typically 103s1). They are approximately 2 mm wide, mechanism. Their work also indicates that the serrations occurred
with a wider IR-thermography image of about 4 mm, and are in- earlier in the fine grain size state, and that the largest changes in
clined to the tensile axis at an angle of about 53 [311]. The local the flow stress occurred for the finest grain structure of TWIP steel.
strain within a type A band, as well as the band velocity, were strain A common scenario for the emergence of PLC related plastic
dependent. At low strains, the initial strain within a band was instabilities suggests that an initially stable uniform strain regime
~0.005 and the band velocity was ~36 mm/s. Local measurements prevailing when plastic deformation commences becomes unstable
of the characteristics of type A PLC bands showed that the strain once a specific critical strain εc is reached. For identifying defor-
rate in the deformation band velocity was ten to twenty times the mation conditions which are less conducive for the occurrence of
1200 50
70
1000 TEL
UTS 40
60
Engineering strain, %
800
Critical strain, %
30
50
600
UEL
YS
20
40
400
30 10
200
0 20 0
-100 0 100 200 300 -100 0 100 200 300 0 100 200 300
(a) Temperature, ºC (b) Temperature, ºC (c) Temperature, ºC
Fig. 51. Temperature dependence of (a) the yield stress and the ultimate tensile stress, (b) the uniform (UEL) and the total (TEL) elongation, and (c) the critical onset strain for
serrations on the stress-strain curve of Fe-18%Mn-0.6%C-1.5%Al steel. (Imposed strain rate: 103s1).
undesired PLC instabilities, it is important to know the strain rate increase of εc. This grain size effect is also seen in Fig. 53. In certain
dependence of the critical strain εc for the onset of serrated cases, though, no critical strain can be identified. This is the case
yielding. Commonly accepted models of the PLC effect [297] predict with Fe-20%Mn-1.2%C TWIP steel investigated by Renard et al.
a non-monotonic, V-shaped εc vs. ε_ curve. Experiments by Qian [305] at room temperature. In their material, stress serrations al-
et al. [302], who studied PLC instabilities in Fe-22%Mn-0.6%C TWIP ways appeared right from the start, as soon as plastic deformation
steel in the strain rate interval of 6 106s1-6 103s1, confirm commenced. Their material had a slightly negative strain rate
this dependence, Fig. 53. At high strain rates (104-102s1, type A sensitivity of the flow stress right from the start of plastic
regime), εc increased with strain rate (‘normal’ behavior). In the deformation.
lower part of the strain rate range considered (106-104s1, type C The entirety of the mentioned work suggests that plastic in-
regime), εc exhibited a negative strain rate dependence, referred to stabilities exhibited by TWIP steels are consistent with the idea that
as inverse, or ‘anomalous’. Grain size was shown to have an effect they are of the PLC type, i.e. take their origin in the dynamic strain
on the critical strain, as well, grain refinement leading to an aging effect. However, Lebedkina et al. [22] indicated that the
Extensometer
75ºC
60
Fe-18%Mn-0.6%C Fe-18%Mn-0.6%C-
58
Engineering Strain, %
0.086%N
56
54
52
40ºC
Extensometer
50 Band
initiation
48
46
44
(b) (c)
42
Fe-18%Mn-0.6%C
40
145 150 155 160 165
(a) Time, s
Fig. 52. (a) Example of a type A band in Fe-18%Mn-0.6%C TWIP steel which nucleates at one end of the tensile specimen and propagates across the entire sample length. The
passage of the band is only detected when it is moving within the extensometer range. This gives rise to characteristic type A serrations on the stress-strain curves. (b) High
resolution infrared image of a type A band; a secondary band is also visible. (c) Traces of multiple deformation bands which give rise to type B serrations on the stress-strain curve of
Fe-18%Mn-0.6%C-0.086%N TWIP steel.
100 Fe-22%Mn-0.6%C
Critical strain Qian et al.

10-1 Type A, grain size: 80 m
Type C, grain size: 80 m
Type A, grain size: 4 m
Type C, grain size: 4 m
Lebedkina et al.
Type A, grain size: 3 m
10-2
10-6 10-5 10-4 10-3 10-2
Strain rate, s-1
Fig. 53. Strain rate dependence of the critical strain for Fe-22%Mn-0.6%C TWIP steel, which exhibits an inverse behavior at low strain rates and a normal behavior at high strain rate
[22,302].
plastic instabilities observed in the case of TWIP steels are not due (b) Type A bands cause plateaus or steps in the strain vs. time
to a standard DSA effect because propagation of type A bands curves, which are due to the low deformation rate in the
prevails over an unusually wide strain rate range (2.1 105s1- gauge length of the sample when type A bands are nucleated
101s1). They argued that deformation twins will locally cause a and propagating outside of the measurement range of the
hardening and generate internal stresses which drive the persistent strain gauge.
propagation of localized deformation bands having the appearance (c) The deformation band velocity decreases with increasing
of type A bands. It should also be noted that the pseudo-PLC effect strain.
discussed in Section 2.2.5 cannot be ruled out as a mechanism for (d) The deformation band velocity increases with increasing
unstable, discontinuous plastic flow observed in TWIP steels. imposed strain rate.
The salient features of type A bands (Fig. 54) [301,306] can be (e) The localized strain inside a band increases in the course of
summarized as follows: straining.
(f) The strain rate within the deformation bands is typically at
(a) The deformation bands are observed over a wide range of least 10e20 times larger than the imposed strain rate.
strain rates (106-101s1) in C-alloyed TWIP steels, char- (g) A deformation band is inclined at an angle of 53 -55 to the
acterized by a negative strain rate sensitivity of the flow tensile direction.
stress, m < 0.
1200 0.7 120

Fe-18%Mn-0.6%C
110 isf=33mJ/m2
1150
0.6
100
1100
Engineering strain
Temperature, C
3
4 10 s 1
90
0.5
1050
80
1000
0.4 70
950
2 10 3 s 1
60
0.3
900 Type A serrations isf=21mJ/m2
50
850 0.2 40
120 160 200 240 280 120 160 200 240 280 120 160 200 240 280
Time, sec Time, sec Time, sec
(a) (b) (c)
Fig. 54. Characteristics of type A serrations in the stress-strain curve of C-alloyed TWIP steel Fe-18%Mn-0.6%C. (a) Serrated stress vs. time curve. (b) Strain vs. time curve with typical
strain plateaus and steps, which illustrates that the strain rate in the deformation band is higher than the externally imposed strain rate. (c) Temperature vs. time curve for a location
near the center of a tensile specimen. The rapid temperature increases are due to the passage of successive deformation bands through the point of measurement. The increase of
the stacking fault energy associated with the temperature rise is also indicated [306].
(h) A temperature increment caused by the passage of an indi- Chung et al. [9] the strain rate sensitivity parameter m for Fe-18%
vidual deformation band is typically in the range of 5 C- Mn-0.5%C-1.5%Al TWIP steel is 0.007. Shen et al. found larger
30 C. Combined heating by the deformation band and the negative values, e.g. 0.021, for the strain rate sensitivity of Fe-20%
background adiabatic heating of the overall tensile sample Mn-0.6%C TWIP steel [220] with a grain size of 3.5 mm, and 0.029
can result in a temperature rise of ~100 C. (Test conditions: for a steel with a grain size of 25 mm. Note that the strain rate
strain rate of ~103s1, temperature ~25 C, sample thickness sensitivity of low C ferritic steel, interstitial-free steel and austenitic
1e2 mm) stainless steel is typically about 0.015.
(i) A localized deformation band is accompanied by a weaker As the strain rate sensitivity measurements by strain rate jumps
secondary band which compensates for the axial misalign- are made when the stress-strain curve appears smooth, the results
ment of the two parts of the tensile specimen on either side presented in Fig. 56 also indicate that a TWIP steel can have a
of the main deformation band. negative strain rate sensitivity even when there are no visible
deformation bands. It should be noted in this connection that the
There is reasonable consistency between the data reported in necessary condition for PLC-type instabilities reads as follows
literature, and most of the deviations from the properties presented [308]:
above are likely due to differences in composition and grain size of
the TWIP steels considered.
DSA is influenced by the temperature, the strain rate, the S < U$s (80)
microstructure and the composition. It is also influenced by the Here U denotes an increment in strain (typically of the order of
grain size. Several metallurgical variable may affect DSA in TWIP 104) that would be produced if all mobile dislocations experienced
steels. Thus, alloying with Al [179,195] and V [307]. Alloying with a collective jump in strain rate corresponding to the A-B transition
nitrogen was also found to increase the critical strain for the onset in the diagram in Fig. 50. The condition expressed by equation (80)
of the PLC effect and modify the initiation of the deformation means that it is not enough for the strain rate sensitivity to be
bands. Lee et al. [200] in particular showed that the addition of negative: it has to have a sufficiently large negative value for a PLC
0.09% N to a Fe-18%Mn-0.6%C TWIP steel alters the appearance of instability to occur. The strain dependence of U stems from the
the serrated flow curves and the properties of type A bands variation of the mobile and forest dislocation density with strain.
considerably. This is illustrated in Fig. 55. Instead of a step-like Along with the variability in behavior and the types of deformation
pattern, stress discontinuities appear as spikes. The reason is the bands, this strain dependence adds to the complexity of the DSA
location of the deformation bands, which are initiated in the gauge, related plastic instabilities in TWIP steels.
rather than the grip part of the sample [200]. The band initiation Fig. 57 compares the occurrence of DSA-related serrations on
stress therein is higher than the stress required for band propaga- the stress-strain curve of a ferritic steel and a TWIP steel. The ser-
tion, which explains the spike-like appearance of the curve. rations on the room temperature stress-strain curve of the TWIP
Fig. 56 shows that the instantaneous strain rate sensitivity, as steel are unexpected. The activation energy for the interstitial
measured by various strain rate jump tests, is always positive. The diffusion of carbon in g-Fe and austenitic Fe-alloys is high,
quantity appearing in the PLC instability condition, Eq. (78), is a (Q ¼ 112.5 kJ/mol), implying that long range C diffusion is very
steady state strain rate sensitivity attained after a transient asso- unlikely at room temperature: the time required for a carbon atom
ciated with a strain rate jump [308]. This steady-state strain rate to diffuse from one interstitial site to the next covering a distance of
sensitivity may have a very small positive value or be negative. It 0.4 nm at room temperature is estimated at 1495 h (setting the pre-
typically varies in the range of þ0.002 to 0.010. According to exponential function in the diffusion coefficient at D0 ¼ 0.0156 cm2/
1100 Fe-18%Mn-0.6%C-0.09%N
1000
1020
900
Fe-18%Mn-0.5%C 1010
Stress (MPa)
800 1000
990
700
980
970
600
960
Strain rate: 10-3s-1 Fe-18%Mn-0.6%C-0.09%N
500 950
10 20 30 40 50 60 35 36 37 38 39 40 41 42
(a)
Engineering strain, % Engineering strain, %
(b)
Fig. 55. (a) Effect of alloying with N on the appearance of the serrations on the stress-strain curve of TWIP steel. (b) Relation between the stress discontinuities and the events of
nucleation and propagation of three type A deformation bands. The bands are nucleated inside the strain gauge length. Note the absence of plateaus between the stress spikes.
Serrated 0.09 Serrated

40 Stress-strain Stress-strain
Deformation band velocity

curve 0.08 curve
35
0.07
Strain of band
30
0.06
25
mm/sec
0.05
20 0.04
15 0.03
10 0.02
5 0.01
Fe-18%Mn-0.6%C Fe-18%Mn-0.6%C
0 0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Transient SRS 0.015 Transient SRS

0.015
Steady state SRS Steady state SRS
Strain rate sensitivity

Strain rate sensitivity
0.010 0.010
0.005 Serrated 0.005 Serrated

Stress-strain Stress-strain
0 curve 0 curve
-0.005 -0.005
-0.010 -0.010
Fe-18%Mn-0.6%C-1.3%Al Fe-18%Mn-0.6%C-1.3%Al
-0.015 -0.015
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
(c) True strain (d) True strain
Fig. 56. Main features of the room temperature DSA in various TWIP steels. Strain dependence of (a) the velocity and (b) the strain of type A deformation bands in Fe-18%Mn-0.6%C.
Strain dependence of the strain rate sensitivity (SRS) measured in (c) upwards and (d) downwards jumps in strain rate in Fe-18%Mn-0.6%C-1.3% Al. Note that the instantaneous
(transient) strain rate sensitivity is always positive [301].
s). Several micro-mechanical models for DSA have therefore been planarity. Koyama et al. [310] observed that deformation twinning
proposed to relate DSA to accelerated diffusion of solute C in the in 〈114〉-oriented grains was not influenced by an increase of the C
presence of twinning. These models are collated in Fig. 58. Lee et al. content, despite an increase in gisf carbon causes. They also noted
[64] suggested that the immobile C-Mn point defect complexes that deformation twins were formed in high-gisf Fe-18%Mn-1.2%C
interact with the stacking faults of gliding dislocations. They argue TWIP steel (gisf ¼ 55 mJ/m2), which had a serrated stress-strain
that C atoms located on the stacking fault plane will be re-located to curve. Both effects were explained by trapping of the trailing par-
a high energy tetrahedral position by the shear from the leading tial dislocations by solute C atoms when the velocities of the
partial dislocation. The C atoms then reorient themselves by a leading and trailing partial dislocations were significantly different.
single diffusive hop to an octahedral interstitial site out of the Their point was that when this occurs, an infinite separation of the
stacking fault plane. This process, which is associated with a local partials, and hence deformation twinning, can be effected.
reduction of the stacking fault energy, enhances planar glide. This Short range ordering discussed in Section 2.2.5 occurs when the
leads to larger dislocation pile-ups and a reduction of the critical number of unlike atom pairs in an alloy is larger than in the random
stress for deformation twinning, as implied by the Venables [95] solution. Using the McLellan model [191] for the distribution of
equation, cf. Section 5.1. Allain [66] suggested that the plastic interstitial solutes in a binary substitutional alloy, Kang et al. [189]
flow instabilities and the negative strain rate sensitivity observed have calculated the effect of SRO on the yield strength of TWIP steel.
for C-alloyed TWIP steel were due to a dynamic interaction be- They proposed the following equation for the yield stress:
tween solute C atoms and dislocations, which interact with defor-
mation twins to stabilize type A deformation bands. He proposed 11:3 11:3 Er ESRO
sys ðMPaÞ ¼ si þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ sSRO ¼ 90 þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ M$
that the C-related DSA may have an important effect on the strain dðmmÞ dðmmÞ b3
hardening mechanisms, including deformation twinning. Based on
(81)
earlier work by Andrews et al. [309], Allain suggested that carbon,
despite an increase of gisf it causes, may still raise the rate of Here si is the Peierls lattice friction stress, d the average grain
deformation twinning. As C increases the lattice friction, it inhibits size, and b the magnitude of the dislocation Burgers vector. The SRO
dislocations glide and promotes the activation of deformation contribution to stress, sSRO, was expressed in terms of Er and ESRO -
twinning as an alternative plasticity mechanism at higher stresses. the energy per atom in the alloy with a fully random C distribution
Allain argued that the presence of solute C atoms at the core of and with short range order, respectively. Ab initio values for Er and
partial dislocations also hinders the formation of constrictions on ESRO were obtained by computing the binding energy of C to three
screw dislocations, thus inhibiting cross-slip and favoring slip types of ternary octahedral cell configurations, i.e. Fe6-xMnxC1, Fe6-
600
-30ºC
1200 -30ºC
RT
500
RT

1000
+200ºC
400
+200ºC
800
300
600
200
400
100 200
AISI 430 Fe-18%Mn-0.6%C-1.5%Al

0 0
0 5 10 15 20 25 30 0 10 20 30 40 50 60 70
(a) Engineering strain, % (b) Engineering strain, %
Fig. 57. Illustration of the difference of the effect of DSA on the deformation curves of (a) C-alloyed ferritic stainless steel AISI 430 and (b) C-alloyed TWIP steel Fe-18%Mn-0.6%
C-1.5%Al.
xAlxC1 and Mn6-xAlxC1, in which the C atom occupies the center strength and ductility, is of primary interest in the context of the
position. The contribution of SRO to the yield strength calculated in present treatise. A fortunate property that distinguishes TWIP
this way was considerable: approximately 100 MPa and 200 MPa steels from non-TWIP alloys is that the occurrence of DSA in TWIP
for Fe-(15e25)%Al-0.6%C and Fe-(15e25)%Al-0.6%C-3%Al TWIP steels does not lead to a reduction of their ductility [22,305]. This
steel, respectively. unusual feature is also illustrated in Fig. 57 which compares the
The effect of DSA on the mechanical performance, notably temperature dependence of the flow stress of a ferritic steel and a
TWIP steel. Whereas there is a reduction of tensile elongation
associated with the occurrence of serrations on the stress-strain
SF curve in the case of the ferritic steel, the elongation is signifi-
C Cross slip cantly larger for the serrated stress-strain curve of the TWIP steel.
Mn plane
Similar to common alloys exhibiting DSA, an increase in me-
C chanical strength due to dynamic strain aging can also be expected
for TWIP steels [311]. Kim et al. [18] showed, however, that this
strength increase is rather small, approximately 20 MPa. The rela-
tion between DSA and strain hardening is complex. Renard et al.
C
[305] reported such a correlation. They observed a pronounced
increase in strain hardening when testing a TWIP steel at 170 C,
Lee et al. Allain which they relate to an observed reduction of the number of ser-
rations on the stress-strain curve. This report appears to be in
(a) (b) disagreement with the deformation curves shown in Fig. 59.
Indeed, the room temperature data presented therein indicates that
Ordered
the addition of Al to a Fe-18%Mn-0.6%C TWIP steel results in a
SF SF
C decrease of the strain hardening and a pronounced delay with the
onset of serrations. Both features are related to an increase of gisf by
alloying with Al, which has two effects: (a) inhibition of deforma-
tion twinning, and (b) a weakening of interaction between carbon
Random atoms and dislocations temporarily arrested at localized obstacles.
SF
The latter is associated with a decrease of the stacking fault width
C
resulting in a reduced interaction, if this interaction involves re-
orientation of C-Mn complexes in the stacking fault plane.
Koyama et al. Kang et al. While dynamic strain aging has been the subject of numerous
publications, little is know about the static strain aging of TWIP
(c) (d) steel. Allain et al. [312] carried out room temperature stress
relaxation tests on a Fe-22%Mn-0.6%C TWIP steel micro-alloyed
Fig. 58. Schematics of the micro-mechanical mechanism of DSA in C-alloyed TWIP
with V. They observed a yield point phenomenon and an associ-
steel according to (a) Lee et al. [64], (b) Allain [72], (c) Koyama et al. [310], and (d) Kang
et al. [189]. ated yield plateau upon reloading and ascribed these effects to
1200 10-4, s-1 1200 10-4, s-1

10-3, s-1 10-3, s-1
10-2, s-1 10-2, s-1
1000 1000

800 800
880 820
860 800
600 600
Eng. stress, MPa
Eng. stress, MPa

840 780
400 820 400 760
800 740
200 200
780 720
20 21 22 23 24 25 20 21 22 23 24 25
Eng. strain, % Eng. strain, %
0 0
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70
Engineering strain, % Engineering strain, %
(a) (b)
Fig. 59. Stress-strain curves for (a) Al-free Fe-18%Mn-0.6%C TWIP steel, and (b) Al-added Fe-18%Mn-0.6%C-1.5% Al TWIP steel. Both steels exhibit a negative strain rate sensitivity of
the flow stress.Type A serrations, characterized by sharp spikes in flow stress followed by stress plateaus, are present on deformation curves of both steels. Whereas type A
serrations appear on the stress-strain curve of the Al-free TWIP steel from the start of the plastic deformation, a large latent strain is required for serrations to set in for the case of
the Al-added TWIP steel.
aging by segregation of C atoms to dislocations during the stress 7. Recovery annealing

relaxation test. In addition, stress relaxation triggers the occur-
rence of serrations on the deformation curve. Based on an analysis Recovery annealing is an unusual microstructure optimization
of these experiments, Allain et al. argue that not only may twins process, consisting in a partial recrystallization annealing treat-
be responsible for the high strain hardening rate observed for ment in the temperature range of 575e625 C. It is effective for the
TWIP steels, but they could also be responsible for pronounced production of TWIP steels with a wide range of strength-ductility
plastic instabilities. They base their view on the consideration that combinations. The retention of the deformation twins in cold
deformation twins act as stress concentrators able to emit dislo- deformed TWIP steel during recovery annealing is utilized to obtain
cations which hinder the propagation of localized deformation a microstructure that provides a high yield strength and an
bands. The presence of twins may, however, be not as important appreciable tensile elongation. Through a suitable combination of
as envisaged by Allain et al. Fig. 60 provides clear evidence for cold rolling and annealing conditions (temperature and time) it is
static strain aging at room temperature in both Al-free and Al- possible to tailor the mechanical properties of recovery annealed
alloyed Fe-18%Mn-0.6%C TWIP. Since the yield drop phenome- TWIP steel over a wide range. The retained twin boundaries result
non was observed in conditions where twinning did not occur, it is in a higher yield strength after recovery annealing as compared to
likely that the static strain aging is due to an interaction between fully recrystallized TWIP steel. The recovery process also signifi-
dislocations and solutes in the initial as-annealed microstructure. cantly improves the ductility over that of the cold-rolled state.
Tsuji et al. [313] also reported that TWIP steel with a grain size less According to Dao et al. [298] the introduction of a large number of
than 2 mm exhibited a yield phenomenon, similar to ultra-fine nanoscale twin boundaries leads to a high strength, while retaining
grained (grain size <1 mm) fcc and bcc metals, but they did not ductility.
provide an explanation for their observation. Kang et al. [314] studied the recrystallization of 60% cold rolled
Kang et al. [189] suggested that the observed yield phenomenon Fe-18%Mn-0.6%C-1.5%Al TWIP steel in the temperature range of
can be rationalized in terms of SRO. The idea was that yielding is 550 C-1100 C. While the dislocation density was reduced during a
caused by the disruption of the SRO by the passage of dislocations, recovery annealing at 550 C, deformation twins were thermally
and that this would result in a softening effect manifesting itself in stable. The static recrystallization set in at 600 C and was complete
a yield drop. The data shown in Fig. 60 also proves that static aging at 700 C. The recrystallized grains had an average size of 2.38 mm.
occurs during stress relaxation test, as reflected in the stress The increase of the yield strength and tensile strength observed
overshoots and yield drops observed upon reloading. Very often the after annealing in the temperature range of 700 C-800 C was due
strain rate, as observed in strain vs. time plots, is temporarily lower to M3C carbide precipitation. At 840 C, grain coarsening occurred
than the imposed strain rate upon reloading. This is evidence for as a result of the dissolution of the M3C carbides. During recovery
local deformation band formation in the part of the sample outside annealing of a cold rolled TWIP steel below 600 C a low dislocation
the range of the strain gauge. density microstructure was obtained which still contained
1200 500
1000

Yield point
450 1.5%
800
1.2%
600 400
Fe-18%Mn-0.6%C-1.5%Al
Fe-18%Mn-0.6%C
400
350
200
Fe-18%Mn-0.6%C-1.5%Al
Fe-18%Mn-0.6%C
0 300
0 10 20 30 40 50 60 70 0 1 2 3 4 5
(a) Engineering strain, % (b) Engineering strain, %
1520 relaxation
1500 yp
True stress, MPa
1480 DSA
1460
1440 Fe-18%Mn-0.6%C-1.5%Al
1420 T: 25 C
Monotonic curve
1400 Relaxation test
1380
0.240 0.245 0.250 0.255 0.260
(c) True strain
1520 0.270
1500 yp
0.265
True stress, MPa
1480
True strain
0.260
1460 reloading
0.255
1440
0.250 band
1420 reloading
relaxation formation
1400 0.245
relaxation
1380 0.240
20 30 40 50 60 70 80 20 30 40 50 60 70 80
(d) Time,s (e) Time, s
Fig. 60. (a) Engineering stress-strain curves for Al-free and Al-alloyed Fe-18%Mn-0.6%C TWIP steel aged at room temperature under stress relaxation conditions. (b) Magnification
of the elasto-plastic transition region of (a), showing that both steels are susceptible to static strain aging. A yield drop phenomenon is seen on the stress-strain curves for both
steels. A distinct stress overshoot and a yield drop are visible on the curve for the Al-alloyed Fe-18%Mn-0.6%C TWIP steel. (c) Comparison of a monotonic tensile tes and a stress
relaxation test. A stress overshoot followed by a yield drop appears upon reloading after the stress relaxation test, and dynamic strain aging results in a small increment, DsDSA, in
the flow stress. (c) True stress vs. time plot for the reloading between two stress relaxation episodes showing the yield point resulting from aging during stress relaxation. (d)
Corresponding true strain vs. time plot showing evidence for localization at the start of the reloading. As the localized deformation band is formed outside of the strain gauge range,
the measured strain rate is initially lower than the imposed strain rate [216].
nanoscale deformation twins. Fe-22%Mn-0.6%C TWIP steel by recovery annealing for 3600 s at
Dini et al. [315] produced a sub-micron grained microstructure 500 C aimed at reducing the dislocation density without changing
in Fe-31%Mn-3%Al-3%Si TWIP steel by a partial recrystallization the nano-twinned structure induced by prior cold rolling. Their
annealing treatment in the temperature range of 575e625 C. Their recovery annealed TWIP steel had a yield strength of 1200 MPa, a
work provided a demonstration that an excellent tensile strength- tensile strength of approximately 1700 MPa and a uniform elon-
ductility balance can be achieved through grain refinement. The gation of 8%.
partially recrystallized 80%-cold rolled TWIP steel annealed at Haase et al. [268,317e319] used the evolution of the crystallo-
575 C had a yield strength of about 1 GPa and a total elongation graphic texture during recovery annealing of cold worked Fe-23%
close to 30%. Similarly, Bouaziz et al. [316] obtained a nano-twinned Mn-0.3%C-1.5%Al TWIP steel at 550 C and 630 C. Their work
was aimed at optimizing the thermomechanical processing of the cycles to fracture, i.e. in an improvement of fatigue life over that of
steel. Texture analysis revealed that the highest twin density was high gisf materials that deform by dislocation glide only. The effect
obtained after a 50% thickness reduction. They observed a slight of gisf comes to bearing in stage I of fatigue, during crack initiation
texture sharpening during recovery, which was indicated by an and growth. Crack growth in stage II is unaffected by gisf. Crack
increased intensity of the deformation texture component. This growth in stage I is transgranular, probably along twins which act
was followed by randomization of texture during primary recrys- as easy paths for crack propagation, similar to persistent slip bands.
tallization, as indicated by an increase of the volume fraction of Although fatigue testing involves very large accumulated strains
randomly oriented grains and a decrease of the intensity of the as compared to conventional tensile testing, the influence of
main texture components induced by cold rolling. The optimal deformation twinning on the high cycle fatigue (HCF) properties
annealing time was at the transition between the texture sharp- appears to be minor. Niendorf et al. [320] studied the low cycle
ening due to recovery and texture randomization during recrys- fatigue (LCF) properties of pre-strained Fe-22%Mn-0.52%C-0.25%V-
tallization. The recrystallized volume fraction after recovery 0.25%Si-0.20%Cr TWIP steel. The effect of pre-straining on the fa-
annealing was found to be less than 10%. In a separate study by the tigue life of TWIP steels is shown in Fig. 61. The fatigue limit defined
same research group the mechanical properties of the recovery as the stress amplitude corresponding to a fatigue life in excess of
annealed Fe-17%Mn-0.6%C-1.5%Al steel were studied in addition to 2 106 cycles is seen to be increased considerably by pre-straining
those of Fe-23%Mn-0.3%C-1.5%Al [319]. They reported that the high prior to the fatigue test. The cyclic stress hardening behaviour of a
yield strength retained after recovery annealing resulted in a sig- Fe-22%Mn-0.6%C-0.3%V reported by Hamada et al. is also illustrated
nificant increase of the energy absorption capacity of the steel and in Fig. 61 [321]. Similar to the commonly observed correlation be-
its superior crashworthiness as compared to the fully recrystallized tween the fatigue limit and the tensile strength of metallic mate-
state. They also demonstrated by means of cross-shaped cup rials [322], the ratio of the fatigue limit to the tensile strength of the
drawing experiments that a considerable ductility was regained TWIP steels investigated was typically in the range of 0.4e0.6.
after a recovery annealing treatment. Independently of the composition, the fatigue limit of these steels
was approximately 400 MPa [321]. The deformation twins formed
8. TWIP effect during cyclic deformation during pre-straining and the nucleation of new twins during cyclic
deformation provided a strength increase which is favorable in LCF
The fatigue performance of TWIP steels is generally considered conditions, as the extra hardening reduces fatigue damage. In the
to be excellent, and superior to the fatigue performance of absence of pre-strain, no cyclic deformation-induced twins were
austenitic stainless steels such as AISI 301 and 316 [320]. Three observed, while existing twins were found to grow. A drop in the
specific characteristics of TWIP steel, viz. a low magnitude of gisf, dislocation density below the level measured in the as-received
which inhibits cross slip, a pronounced glide planarity, and the material was observed in the fatigued one. This resulted in plastic
occurrence of deformation twinning, are known to influence the instability and cyclic softening. Hamada et al. also reported a pro-
fatigue properties in both high cycle and low cycle fatigue of fcc nounced Bauschinger effect during fatigue testing. This is tanta-
alloys in general. A low gisf generally results in a larger number of mount to a distinct tension-compression asymmetry and a build-
1000 1000
2% strain amplitude
900 900
30% pre-strain: UTS=1357MPa
YS=1152MPa
800 800
Fe-18%Mn-0.6%C 10% pre-strain: UTS=1129MPa
1% strain amplitude
YS=471MPa
Stress amplitude, MPa
Stress amplitude, MPa
700 700
600 As-received: UTS: 1004MPa 600

Fe-22%Mn-0.6%C 0.5% strain amplitude
500 AISI 316L 500

YS=471MPa
As-received (UTS: 866MPa) 0.3% strain amplitude
400 400
As-received (UTS: 530-680MPa)

300 300
200 200
Fe-22%Mn-0.6%C-0.3%V
100 100
103 104 105 106 107 1 10 102 103 104 105
(a) Fatigue cycles (b) Fatigue cycles
Fig. 61. (a) High cycle fatigue curves for TWIP steel, illustrating the positive effect of pre-strain on fatigue endurance. Data for AISI 316L and Fe-22%Mn-0.6%C TWIP steel is from
Hamada et al. [333], for Fe-18%Mn-0.6%C TWIP steel is from Kim et al. [326]. (b) Strain amplitude dependence of the stress amplitude versus number of cycles for a V-microalloyed
Fe-22%Mn-0.6%C-0.3%V TWIP steel, illustrating the cyclic hardening behavior of TWIP steel. Data taken from Hamada et al. [327]. The fatigue test conditions are R ¼ 1 (sym-
metrical tension-compression loading) and a frequency of 1 Hz.
up of a positive mean stress. straining can act as early crack nucleation sites under cyclic
Hamada et al. [323] studied the HCF properties, the fatigue loading. These conclusions are in accord with the above mentioned
strength, the crack initiation sites and the crack propagation paths principal observations by other authors indicating that no defor-
in three TWIP steels with gisf of 19, 23 and 26 mJ/m2. No deformation twins are formed under fatigue loading and that TWIP
mation twins were formed in the three steels during the cyclic steels have superior LCF properties. Since, in the spirit of the
loading. They concluded therefore that the TWIP effect had no ef- mentioned correlation, the HCF properties are controlled by the
fect on the HCF properties. The stress amplitude vs. the number of tensile strength, the HCF life of pre-strained TWIP steels is
cycles curves are independent of gisf. As mentioned above, the fa- improved with increasing amount of pre-straining.
tigue limit of the three TWIP steels, approximately 400 MPa, is The effect of cyclic pre-straining on the mechanical properties of
considerably higher than that of austenitic stainless steels. The ratio a Fe-22%Mn-0.6%C-0.3%V TWIP steel were investigated by Hamada
of the fatigue limit to tensile strength is, however, similar to that for et al. [327]. Their original aim was to use cyclic plastic deformation
austenitic stainless steels, i.e. in the range of 0.42e0.48. In the HCF as a way to increase the yield strength. The cyclic pre-straining
regime fatigue cracks were predominantly initiated at or near grain consisted in fully reversed strain-controlled cycling at different
boundaries due to the local elastic strain incompatibilities there strain amplitudes until the saturation stage was reached. By that,
and the intrinsic weakness of the grain boundaries in Mn-alloyed the yield strength was increased without much loss in elongation.
steels. Although the cracks are nucleated at grain boundaries, The cyclic pre-straining resulted in the formation of a dislocation
their propagation is mainly transgranular. The nature of the crack cell structure with very few mechanical twins. The dislocation
propagation path was determined as 61% transgranular, 18% inter- arrangement is mainly determined by the prevalent slip mode, viz.
granular, 11% along twins and 9% along slip bands. planar versus wavy dislocation glide. In the case of planar glide,
Grain refinement results in a significant improvement of the glide plane softening causes dislocations to follow the leading
fatigue strength. This was demonstrated by Karjalainen et al. [324] dislocation on its glide plane. This planar glide behavior has been
who showed that the fatigue limit of a Fe-16.4%Mn-0.29%C-1.54%Al related to a low stacking fault energy, gisf, short range ordering, and
TWIP steel rose from 400 MPa to 560 MPa upon grain size reduc- lattice friction effects. A characteristic dislocation pattern consists of
tion from 35 mm to 4.5 mm. The fatigue limit of TWIP steels was also regions containing a high density of parallel edge dislocations,
typically higher than the yield stress. Using atomic force micro- separated by low dislocation density regions in which screw dislo-
scopy, Hamada et al. [323] were able to demonstrate that surface cation segments glide. The frequent, uncorrelated cross-slip events
features of fatigued samples were not associated with deformation are the cause of wavy glide, and the typical dislocation arrangement
twins, but are rather extrusion-intrusion profiles due to bands of has a cell structure. Karjalainen et al. [324] report the formation of a
intense dislocation slip. This confirmed the already mentioned rough labyrinth or a vein-like dislocation structure in Fe-16.4%Mn-
absence of deformation twinning during cyclic loading reported by 0.29%C-1.54%Al TWIP steel tested in high cycle bending fatigue.
Niendorf et al. [320] and Hamada et al. [323]. However, the
observation that deformation twinning does not take place during 9. High strain rate behavior of TWIP steels
fatigue testing and the conclusion that deformation twinning does
not influence the fatigue properties of TWIP steels was questioned High strain rate properties are of considerable technical interest,
by Karjalainen et al. [324] who did observe 150e200 nm wide especially for automotive applications for passenger safety-related
deformation twins in fatigued Fe-16.4%Mn-0.29%C-1.54%Al TWIP parts where high impact energy absorption is at a premium. Fig. 62
steel with a 35 mm grain size by TEM. They noted that cracks were illustrates the results of axial crush tests carried out on various
initiated early in the fatigue life, and that the crack growth rate was advanced high strength steel grades. The TWIP steel evidently
low. These observations were supported by the results of Niendorf stands out as the material with a superior crashworthiness.
et al. [325] on the effect of twinning on the crack grow behavior of Using a split Hopkinson pressure bar, Ha et al. studied the
Fe-22%Mn-0.52%C-0.25%V-0.25%Si-0.20%Cr TWIP steel in minia- microstructure of Fe-22%Mn-0.4%C TWIP steel after high strain rate
ture compact tension specimens. According to Niendorf et al., deformation at 103 s1 [328], while Curtze and Kuokkala [127]
twinning was observed in the plastic zone, even though their employed the same method to investigate Fe-(25e28)%Mn-1.6%
density was low. They report a low crack growth rate, da/dN, in the Al-0.08%C TWIP steel. Both groups observed an appreciable in-
Paris-law regime: crease of tensile strength at high strain rates, e.g. to 2.265 GPa in the
study by Ha et al., and an increasing strain hardening up to a true
strain of approximately 0.19. Ha et al. report that the dynamic
da
ðmm=cycleÞ ¼ C$DK m ¼ 2$108 $DK 2:7 (82) loading leads to the formation of nanostructured austenite which
dN appears random at low magnification. At high magnification the
Here DK is the stress intensity, in MPa√m and C is a constant microstructure exhibits nano scale features, such as 20e30 nm
coefficient in the Paris law. The parameter values in Eq. (82) are sized grains which are twin-related to the neighboring grains in
given for R ¼ smin/smax ¼ 0.1, i.e. cyclic deformation in the tension their orientation. According to Ha et al. [328] the formation of twin-
range. matrix lamellae within the grains occurs only in the early stages of
The effect of pre-straining on the fatigue performance of Fe-18% the high strain rate deformation. The adiabatic heating of the
Mn-0.6%C-0.2%Si TWIP steel was studied by Kim et al. [326]. Pre- sample at larger strains has two consequences: an increase of gisf,
straining was effected by cold rolling with a thickness reduction which inhibits twin formation, and dynamic recovery. The grain
of 10% and 30%. This resulted in a considerable increase of tensile fragmentation resulting from interactions of twins with grain
strength and a decrease of the total elongation, from 78% in the un- boundaries and concurrent dynamic recovery is believed to
formed condition to 20% after a cold rolling to 30% thickness generate the nano-structured austenite observed.
reduction. Both strain-controlled LCF tests and stress-controlled
HCF tests were carried out. A significant reduction of the LCF life 10. Fracture of TWIP steels
with increasing pre-strain was found. The authors argued that
deformation twins generated during pre-straining reduce the Due to their low yield strength and large ductility, the failure
number of twins formed in the course of cyclic deformation. They mode of TWIP steel is usually ductile. The negative strain rate
also suggested that the twin boundaries generated during pre- sensitivity makes diffuse necking strain small, and fast fracture is
Twinning-induced plasticity steel
Transformation-induced plasticity steel
Dual phase steel
High strength low alloy steel
Interstitial-free steel
Bake-hardenable steel
Rephosphorized steel
0 0.1 0.2 0.3 0.4 0.5

Specific energy absorption at strain rate of 100/s, J/mm3
Fig. 62. A qualitative comparison of the crash performance properties of various high strength automotive steel grades obtained in axial crush tests.
observed once necking is initiated. Although not much work has fracture prior to necking. Fig. 63 shows the appearance of a fracture
been done in this area, fracture of TWIP steel has been given surface for Fe-18%Mn-0.6%C-1.5%Al TWIP steel after tensile strain-
considerable attention in connection to three phenomena (a) edge ing. The image shows equiaxed dimples associated with a small
fracture in hole expansion, (b) liquid metal-induced embrittlement particle, which plays a key role as a microvoid initiation site in the
and (c) the hydrogen-related delayed fracture. course of ductile fracture. SEM-EDS analysis results showed that
In the cited work by Hamada et al. [329] the fracture surface of the inclusions could also be AlN or TiN precipitates. The appearance
Fe-25%Mn and Fe-25%Mn-3%Al TWIP steel after room temperature of the fracture surface is identical to that observed by Hamada et al.
tensile deformation was analyzed. Despite the presence of strain- [329]. Premature ductile fracture of TWIP steels, with little or no
induced martensite, the appearance of the fracture surfaces of post-uniform necking, was originally attributed to large inclusions
these steels was consistent with a ductile fracture by microvoid or smaller precipitates. However, the prevalent current view is that
coalescence. Specifically, dimpled fracture surfaces with inclusions it is a negative strain rate sensitivity of the flow stress, found in
visible inside some of the dimples were observed. The SEM-EDS certain TWIP steels, which is a key factor determining their fracture
analysis revealed that these were nonmetallic inclusions such as behavior.
(Fe,Mn)O, Al2O3 and MnS, in agreement with the observation of
endogenous inclusions in Fe-Mn alloys by Tomota et al. [330], and 11. Fracture in hole expansion
in TWIP steel by Allain [71] and Gigacher et al. [331]. These in-
clusions are less deformable than austenite. They are therefore Cut edge stretchability is an important aspect of sheet form-
prone to fracture. Alternatively, strain gradients which develop at ability. Whereas the press forming performance of TWIP steels is
the matrix/inclusion interface can result in interfacial decohesion. excellent due to its high strain hardening capability, the lower
The coalescence of cavities formed by these processes can result in normal anisotropy, represented by the r-value (r0 ¼ 0.79,
r45 ¼ 1.13, r90 ¼ 1.28), and the negative strain rate sensitivity of the
flow stress of TWIP steels are detrimental to stretchability. Collec-
tively, these factors result in premature ductile fracture during
stretch-flanging (i.e. hole expansion) when the initial hole is made
using a technique that leads to a considerable prior deformation of
the hole edge, such as hole punching [300]. This behavior was
analyzed in detail using an advanced in situ method that permits
local measurement of the principal strains and the temperature.
The strain distribution analysis revealed that the hole edge defor-
mation in the deep drawing mode was similar to uniaxial tensile
deformation. Fig. 64 compares the hole expansion performance of
an interstitial free (IF) steel and a TWIP steel. Edge cracking was
ductile in both cases. Outspokenly ductile fracture at the hole edge
occurred after appreciable edge thinning and localized necking in
the case of the IF steel. By contrast, in the case of the TWIP steel a
single crack leading to catastrophic failure occurred. In both cases
1 m the fracture surface exhibited shallow dimples, which are indica-
tive of ductile fracture by micro-void coalescence.
Fig. 63. Fracture surface of Fe-18%Mn-0.6%C-1.5%Al TWIP steel after tensile straining.
The dimples seen in the micrograph are indicative of ductile fracture by micro-void
12. Hydrogen-delayed fracture
coalescence. Note an AlN single crystal precipitate at which a larger void is believed
to have nucleated. Hydrogen has been reported to reduce the stacking fault energy
(c) 500 m
(a)
500 m 50 m
(b) (d)
Fig. 64. Hole expansion tests for (a) IF steel and (b) TWIP steel. The arrows indicate the sites of crack initiation. (c) Edge crack appearance for IF steel, showing extensive thinning
prior to rupture. (d) Edge crack appearance for TWIP steel, showing the absence of diffuse necking prior to shear rupture. The dimpled appearance of the fracture surface provides
evidence for ductile fracture by micro-void coalescence.
of austenitic steels [332,333]. This effect promotes the 30%, making the crystal structure more deformable locally. The
deformation-induced ε-martensitic transformation [334,335] and mentioned effects of low gisf, i.e. limited cross-slip and enhanced
deformation twinning [336,337], resulting in a marked change in planar glide, are essential for obtaining an increase in strain hard-
stress-strain response. When captured by a dislocation in an fcc ening by reducing the dislocation recovery rate. It is still unclear
metal or alloy, a hydrogen atom affects the dislocation mobility by whether the effect of H is due to the presence of hydrogen atoms in
decreasing the activation barrier to dislocation motion. At tem- the stress field of a dislocation or through a decrease of the shear
peratures slightly below room temperature, H atoms can also form modulus they induce in fcc alloys. The modulus effect should
atmospheres around dislocations. Hydrogen has been reported to reduce the shear stress for dislocation nucleation and motion. H
enhance planar glide in fcc metals and alloys. Two reasons have also contributes to the nucleation of martensite by stabilizing the
been put forward to explain this effect: (a) H has an effect on the bcc and the hcp martensitic phases in fcc austenitic steel. Both the
character of a dislocation and (b) H reduces gisf. Due to these hydrostatic lattice distortion and the formation of H-H pairs by H
mechanisms, H stabilizes edge dislocation segments and inhibits charging have been reported to cause the stress-induced
cross-slip, thus promoting planar glide. The reduction of gisf caused martensitic transformation. The formation of H-H pairs has been
by H appears to be considerable, in the range of 20e40%. It is reported to stabilize the hcp lattice, because the hcp crystal struc-
related to a change in the local electronic structure of the stacking ture allows for the formation of H-H pairs by H atoms located in
fault in the presence of H. As a result of the H-Fe interaction via the opposite octahedral sites.
Fe 4s and H 1s orbitals, the Fe-Fe bond strength is reduced by about Fig. 65 illustrates the cracking of a deep drawn cup of Fe-22%
Crack
Longitudinal
Longitudinal
direction
direction
Hoop Hoop
direction direction
B
2 m 2 m
(a) (b) (c)
Fig. 65. (a) Large longitudinal crack on a deep drawn cup made from Fe-17%Mn-0.4%C-1.3%Al TWIP steel (grain size ¼ 15 mm; gisf ¼ 17 mJ/m2) H-charged for 32 h at room
temperature. (b) Intergranular crack. (c) Complex transgranular crack showing crack propagation (A) parallel to the twin/matrix interface and (B) across deformation twins. Note
that the cracks shown in (b) and (c) are two examples of the much finer cracks present on the outer side of the sample shown in (a). TEM analysis of samples taken from within the
cup wall reveal the presence of ε-martensite; magnetic measurements also showed the presence of 0.5e1.5 vol percent of a’-martensite.
Mn-0.6%C TWIP steel shortly after it broke spontaneously during a deformation-induced twinning and martensite transformation. The
hydrogen-charging test. Both intergranular and transgranular effect may be more subtle, involving the interaction between Al and
cracks are present on the outer surface in addition to a singular H atoms in, or close to, the stacking fault plane. H may increase the
major crack. The transgranular cracks propagate through defor- stacking fault width as it segregates on the stacking fault plane. This
mation twins and along twin/matrix interface boundaries. phenomenon, known as the Suzuki effect makes it easier to pull the
Although the H-delayed fracture of deep drawn cups is quite dra- partials apart, thus leading to a decrease of the twinning stress.
matic, the effect of H pre-charging on the tensile properties of TWIP It is now recognized that the delayed cracking phenomenon,
steels is generally insignificant, because the hydrogen diffusion in which occurs in air or during the immersion in water, shares many
the bulk of the material at room temperature is almost negligible. H similarities with stress corrosion cracking. The mechanisms favor-
diffusion in fcc g-Fe is considerably slower than in bcc Fe, by a factor able for delayed cracking include (a) the uptake of H in the material
of approximately 106 [338]. This low H diffusivity implies that the during steel production and processing, (b) the high tensile stresses
testing of the sensitivity of a TWIP steel to H-delayed fracture using at the edge of the deep drawn cup, and (c) crack-initiating features
H pre-charged tensile specimens results in an inhomogeneous such as the intersection of twins, strain-induced ε or a0 martensite,
fracture surface, with a brittle fracture close to the sample surface, and the presence of grain boundaries with an intrinsically low
within the region of solute hydrogen penetration, and a ductile cohesion strength.
fracture in the bulk of the sample. This is illustrated in Fig. 66, Possible influence of the H content on the fracture stress, frac-
which shows that only the near-surface regions of H-charged TWIP ture strain and time to fracture of smooth and notched samples of
steel experience a predominantly intergranular brittle fracture. The Fe-18%Mn-0.6%C-1.5%Al TWIP steel was tested by Soo et al. [340]. It
bulk remains very ductile. The surface of the grains in the inter- was found that the specimens were immune to H-induced cracking,
granular fracture zone also shows clear evidence of planar slip. This the fatigue properties being practically insensitive to the amount of
suggests that hydrogen does have an effect on the dislocations, but H with which they were pre-charged. Thermal desorption analysis
due to a limited penetration length, it does not embrittle the bulk of (TDA) of H introduced during cathodic charging revealed that most
the sample. of H is weakly bound inside the material and gradually diffuses out
The resistance to H-induced embrittlement can be enhanced by of the specimen with time. There are two kinds of hydrogen trap-
Al additions to TWIP steels. The concept was demonstrated for Fe- ping sites, having a low binding energy of 35 mJ/mol and a high one
(15e18)%Mn-1.5%Al-(0.15e0.60%)C TWIP steel, which proved to be of 62 mJ/mol. These trapping sites were identified as dislocations
resistant to H-delayed fracture, as first reported by Jung et al. [339]. (or grain boundaries), and twin boundaries, respectively [340].
It is usually considered that, through an increase of the SFE, the According to these authors, dislocations and grain boundaries are
addition of Al will result in a slowdown of the kinetics of considered to be shallow or “diffusible” H trap sites. Twin
10 m 10 m
(a) (b)
C
10 m 10 m
(c) (d)
2 m
(e)
Fig. 66. (a) Fracture surface and (b) sample surface of a Fe-18%Mn-0.6%C-1.5%Al TWIP steel. The appearance of the fracture surface indicates fully ductile fracture. It is entirely
covered with dimples, and no cracks are visible on the surface. (c) Fracture surface and (d) sample surface of the same steel pre-strained by 20% at 150 C, hydrogen-charged to
4.9 ppm of solute H, and tested to fracture at room temperature. The choice of the higher pre-straining temperature is to avoid formation of martensite. The fracture comprises a
brittle intergranular fracture zone near the surface (B), a fracture-mode transition zone (T), and a large area of ductile fracture. The sample surface exhibits a high density of wide
cracks (C) due to intergranular de-cohesion. (d) Magnification of the brittle fracture surface showing slip traces (S) at the surface of the grains in the zone where intergranular
fracture has occurred.
boundaries act as deep or “non-diffusible” H trap sites. The change in mechanical properties during in situ H-charging, typi-
diffusible H is considered to be responsible for H-delayed fracture, cally in a 3% NaCl solution containing 3 g/L NH4SCN at a current
whilst non-diffusing hydrogen atoms, strongly bound at deep traps, density in the range of 1e10 A/m2. In tensile tests at low strain rates
do not contribute to this phenomenon. (5 105 s1) they observed a clear reduction in ductility in Fe-22%
The role of strain-induced ε-martensite in H-delayed fracture Mn-0.6%C TWIP steel. They also observed that fracture occurred
was studied by Chun et al. [341]. They made the point that during stress relaxation tests. These effects are illustrated in Fig. 67.
ε-martensite in Fe-15%Mn-0.6%C-2%Cr TWIP steel offers additional The detailed fractography they conducted revealed that the crack
H trap sites, and that ε-martensite laths act both as crack initiation path was mostly intergranular. Transgranular fracture was associ-
sites and as a crack propagation path. H-delayed fracture of cold ated with crack propagation along twin boundaries. According to
deformed Fe-18%Mn-0.6%C-1.5%Al TWIP steel exhibited different Koyama et al. [352] cracks started at intersections between primary
features. As gisf in that steel was high enough, no ε-martensite was and secondary twins and propagated trans-granularly along twin/
formed during deformation. Although the cold deformation resul- matrix interfaces in a zigzag manner. The same authors reported
ted in a considerable increase of the tensile strength, the notch that hydrogen-assisted cracking in Fe-18%Mne1.2%C TWIP steel
fracture stress of H-charged samples was only very slightly initiated at both grain boundaries and deformation twins, with
reduced. This was attributed to increased density of deformation crack propagation occurring along both types of interfaces. Ac-
twins, which acted as strong hydrogen traps. The same research cording to these authors, the deformation twins assist the forma-
group also studied the influence of Al additions on the properties of tion of intergranular cracks and their propagation, by stress
notched tensile samples pre-charged with hydrogen [342,343]. The concentration at the tip of the twins [353].
addition of Al was found to be a very effective way to suppress the Van Tol et al. [168,169,354] suggested that the nucleation of the
susceptibility of TWIP steel to H-delayed cracking. They argued that delayed H-fracture cracks in the wall of deep drawn cups may be
a decrease of the solute H trapped at dislocations due to addition of due to the presence of a small volume fraction of strain-induced a’-
Al should also result in a decrease of the dislocation density. As the martensite formed at intersections of shear band or twins. Longi-
authors did not measure the dislocation density, this conjecture is tudinal cracks in the wall of H-delayed fractured deep drawn cups
yet to be proven. Park et al. [344] also observed a decrease of solute were mostly intergranular, but could become transgranular and
H at shallow trapping sites. They proposed an alternative mecha- propagate along twin boundaries. The longitudinal cracks propa-
nism for the suppression of H-delayed fracture of TWIP steel by Al gated mainly along the grain boundaries. Some cracks were also
addition by arguing that the improvement was due to the forma- found to propagate perpendicular to twin boundaries. The H-
tion of a double oxide layer at the sample surface. This (Fe0.8Mn0.2) induced delayed fracture of deep drawn TWIP steel was related to
O surface layer, together with an a-Al2O3 interface layer, was the magnitude of the residual hoop and longitudinal stresses.
considered to act as a diffusion barrier preventing the uptake of H The emerging picture of the delayed H-fracture thus relates this
during hydrogen charging. phenomenon to the effect of hydrogen on the planarity of dislo-
There also several studies on the influence of Cu and Si additions cation glide and the magnitude of the stacking fault energy. More
on the hydrogen embrittling of TWIP steels. According to Kwon subtle effects of pre-straining, mainly through deformation-
et al. [345] Cu additions (0e2 mass-%) to Fe-17%Mn-0.8%C TWIP induced twinning and formation of ε- or a’-martensite, and alloy-
steel improve its resistance to hydrogen embrittlement. A favorable ing with Al, Cu, and Si on the delayed H-fracture and hydrogen
effect of Si additions (0.01e2.95 mass-%) to Fe-18%Mn-0.6%C TWIP embrittlement are beginning to find rational explanations.
steel was explained in terms of the formation of a double oxide
layer on its surface [346]. This layer consisted of a surface layer, an 13. Liquid metal-induced embrittlement
oxide mixture of (Fe,Mn)O and (Fe,Mn)2SiO4, and a pure (Fe,Mn)2-
SiO4 interface layer. The diffusion barrier they formed reduced the Liquid metal induced embrittlement (LMIE) is an undesired
hydrogen uptake during H charging of the TWIP steel. The phenomenon that occurs due to transgranular or intergranular
(Fe,Mn)2SiO4 layer in the Si-added TWIP steel was, however, less decohesion of a metal or an alloy by rapid penetration of another
efficient than the a-Al2O3 layer on the Al-added steel in reducing liquid metal or alloy into its microstructure. The penetration is
the H uptake. The addition of 3% of Si also resulted in the strain- usually along grain boundaries in a polycrystalline material, or
induced formation of ε-martensite. According to Lee et al. [346], along sub-grain boundaries in a single crystal. When tensile
the binding energy of solute hydrogen to g/ε interfaces is approx- stresses are present, a liquid film appears to cause grain boundary
imately 22 kJ/mol. Based on H desorption spectra, Lee et al. claim decohesion, as the boundary within a solid is replaced by a liquid
that the enhanced embrittlement observed for the 3% Si-added film. This results in a brittle fracture due to the rapid intergranular
TWIP steel was due to the higher sensitivity of ε-martensite to H propagation of one or more cracks formed in the region where the
embrittlement. liquid metal has penetrated. The permeation by the liquid may be
Ryu et al. [347] looked into the effect of hydrogen on the char- preceded by rapid solid state diffusion of the embrittling atoms
acter of fracture. In their view, when solute H is present, inter- along grain boundaries. In this case it is not known whether the
granular and transgranular fracture of a TWIP steel is associated loss of grain boundary cohesion is due to the presence of the
with the accumulation of dislocations at grain boundaries and twin diffusing solutes at the grain boundary or the penetration of a
boundaries, respectively. In the presence of strain-induced liquid phase. When the sensitivity of a combination of materials to
ε-martensite formed in H-charged TWIP steel, ε-martensite plates LMIE is tested in laboratory conditions, i.e. by carrying out a tensile
fracture in a brittle manner and the crack path is transgranular. test in conditions in which the embrittling metal or alloy is in the
In general, most authors agree that the flow stress and the strain liquid state, a pronounced decrease in both the fracture strength
hardening behavior of tensile specimens are not affected by the ex and the amount of plastic deformation prior to fracture is
situ H pre-charging of recrystallization annealed TWIP steels. The observed. As TWIP steels are considered as prospective materials
tensile elongation is decreased, however [348,349]. Pre-straining for structural applications in the automotive industry, they need
prior to H-charging does not influence the flow stress and the to be able to be galvanized to protect them from perforation and
strain hardening of TWIP steels, but does have a minor effect on the cosmetic corrosion. Coating of a TWIP steel with Zn or Zn-alloy at
ductility. high temperatures may result in Zn liquid metal embrittlement
In contrast to other workers, Koyama et al. [350,351] observed a (Zn-LMIE), Fig. 68. The issue of Zn-LMIE of TWIP steels is mainly
Fe-18%Mn-0.6%C-1.5%Al Fe-18%Mn-0.6%C-1.5%Al Fe-22%Mn-0.6%C

0.0004 1100 1100
8% strain
4% strain
Hydrogen desorption rate, ppm/s
0% strain 1000 No H
1050
4.9ppm H
Engineering stress (MPa)

Strain rate: 10-5s-1
0.0003 Hydrogen: 12 ppm
900
1000
Stress, MPa
800 Fracture
0.0002 950
700
Strain rate: 10-2s-1
900 Hydrogen: 5.2 ppm
600
0.0001
20% 850
500
Pre-strain Hydrogen charging
0.0000 400 800

0 50 100 150 200 0 10 20 30 40 50 60 70 0 1 2 3 4 5 6 7 8 9 10
Temperature, ºC Strain, % Hydrogen charging time, hour
(a) (b) (c)
Fig. 67. (a) H-desorption rate curves for Fe-18%Mn0-0.6%C-1.5%Al TWIP steel for three different pre-strain levels: undeformed, 4% pre-strained and 8% pre-strained. The samples
were H-charged for 24 h [359]. (b) Influence of pre-straining on the mechanical properties of H-charged Fe-18%Mn0-0.6%C-1.5%Al TWIP steel [349]. (c) Stress relaxation during the
in-situ charging of Fe-22%Mn-0.6%C TWIP steel [352].
relevant to resistance spot welding, a key process in automotive region. The G-(Fe,Mn)3Zn10 phase was also observed at the tip of
body assembly. The sensitivity of TWIP steels to Zn-LMIE was the Zn penetration zone deep below the surface. Kang et al. [358]
al and co-workers [355e357]. Zn-LMIE of
studied in detail by Be proposed two mechanisms for the Zn-LMIE of Zn-coated TWIP
TWIP steels involves the presence of liquid Zn, Znliq, or a Zn-rich steel, which are consistent with the features of the Zn penetration
liquid alloy, combined with a high, uniformly distributed or observed by micro characterization: (a) the Krishtal-Gordon-An
locally concentrated, tensile stress. The Zn-LMIE of TWIP steels model [359,360] for LMIE crack formation initiated by rapid
was assessed by Kang et al. [358] by means of microstructural intergranular solid state diffusion of Zn, and (b) the Stoloff-
analysis of sub-critical LMIE cracks in Zn-coated TWIP steel Johnson-Westwood-Kamdar model [361,362] for LMIE crack for-
deformed in uniaxial tension at high temperature. Zn was found to mation due to intergranular liquid metal percolation.
penetrate into steel along the grain boundaries. Both Zn grain
boundary diffusion and Znliq percolation along grain boundaries 14. Hot ductility
was observed. Evidence of Znliq percolation was inferred from the
presence of the G-(Fe,Mn)3Zn10 phase along the Zn penetration Numerous production difficulties have been reported for
250
Engineering Stress, MPa
200 -Fe3Zn10
Uncoated TWIP Steel
150
Zn grain
100
Zn-coated TWIP Steel boundary
penetration
50
850ºC
Fe-0.44%C-17%Mn-1.4%Al
0
0.0 0.1 0.2 0.3 0.4
(a) Engineering Strain (b) (c)
Fig. 68. (a) Engineering stress-strain curves of Zn-coated TWIP steel tested at 850 C. (b) Fracture of a Zn-coated TWIP steel specimen at the end of a tensile test at 850 C. The LMIE
fracture is associated with a main crack with multiple secondary branches. (c) Electron-probe microanalysis of the Zn distribution in an area of Zn penetration. G-(Fe,Mn)3Zn10 is
present in the crack opening area of the Zn penetration. The Zn content at the tip of the Zn penetration zone is lower than the maximum solubility of Zn in g-(Fe,Mn) at the test
temperature [357].
avoid the limitations of standard electron backscattering diffraction

Mn concentration (wt.-%) or conventional X-ray diffraction methods.
It is now clear that the activation of the TWIP effect is governed
7.3 27.6 not solely by the value of the intrinsic stacking fault energy. Other
energy barriers, which describe the generalized stacking fault en-
ergy landscape, play an essential role as well. Progress in this area
will require a more systematic use of currently available ab initio
methods and the development of techniques to compute the
generalized stacking fault energy of multi-component alloys such
as TWIP steels at non-zero temperatures.
The role of carbon in TWIP steels remains unclear. Although the
room temperature diffusion data for C suggest that its influence
should be very limited, there are definite indications that solute C is
directly involved in many fundamental processes, from static and
10 dynamic strain aging to deformation twinning. It also has an effect
on the temperature and strain rate dependence of the mechanical
properties of TWIP steels. As with all point defect-related proper-
Fig. 69. Electron probe micro-analysis of the Mn distribution in a hot rolled TWIP steel
ties in metals and alloys, progress in this research area will require
illustrating the inhomogeneous Mn distribution resulting from the interdendritic
segregation during casting. the use of a range of advanced methods of microstructure analysis,
including internal friction measurements, and a much more sys-
tematic study of the low temperature dependence of the me-
casting, hot rolling, and cold rolling of high Mn steels. Mn loss oc- chanical properties of model TWIP steels in a wide strain rate
curs due to a high Mn vapour pressure reducing the Mn yield domain.
during steelmaking. The casting microstructure of TWIP steels is The fundamentals of TWIP steels and the developments high-
coarse with well-developed dendrites and coarse equiaxed grains. lighted in this review should help devising more refined micro-
This is due to the wide solidification range and the low thermal mechanical models which can predict the role of isotropic hard-
conductivity of TWIP steels. According to Scott et al. [363], the ening and kinematic hardening for complex deformation paths. The
solidification range of Fe-22%Mn-0.6%C is 120 C, from 1407 C to importance of dynamic strain aging and the manner in which
1287 C. The solidification range of Fe-18%Mn-0.6%C-1.5%Al is certain types of TWIP steels deform, characterized by a transition
110 C, i.e. from 1380 C to 1270 C. The hot ductility of austenitic from uniform to localized deformation, by propagating or station-
steels with a Mn content in the range of 9e23 mass-% and C content ary localized strain bands, highlight the complexity of the strain
in the range of 0.6-0.9 mass-% was studied by Bleck et al. [364]. hardening in TWIP steels. It encompasses an intricate interaction
They observed that a reduction of the Mn content resulted in an between dislocation glide and deformation twinning, and involves
improvement of the hot ductility. The steels investigated solidified solute effects, which have yet to be accounted for on a solid basis of
in a relatively wide temperature range. The fracture surfaces of Fe- physical metallurgy.
16%Mn-0.8%C TWIP steel tested at 1201 C and 1000 C show While the main thrust of the present review has been on the
dendrite branches, micro-shrinkage leading to the formation of strain hardening mechanisms and the role of deformation twinning
interdendritic micro-porosity and interdendritic segregation, cf. in the mechanical performance of TWIP steels, other aspects of this
Lan et al. [365]. This interdendritic segregation is associated with an important group of advanced steels relevant for their processing
inhomogeneous Mn distribution in the steel (Fig. 69). and properties under service conditions have been covered. It is
The grain boundary oxidation of continuously cast TWIP steel is hoped that the information compiled in the review and the critical
known to result in the formation of microscabs during hot rolling. assessment of the various contributions to this area will provide a
Direct strip casting was proposed as a possible solution to this starting help for a novice and also in-depth information for an
problem, as it avoids the formation of bending-unbending cracks expert in the burgeoning field of TWIP steels.
[366]. In rolling, edge cracks, poor surface quality and rolling in-
stabilities are typically reported [367,368].
Acknowledgements
15. Conclusions
The authors gratefully acknowledge the support and contribu-
tions form the following persons: Dr Gwan Keun Chin (POSLAB), Dr
The understanding of the mechanical properties of TWIP steels
Jin Kyung Kim (POSTECH), Professor Wolfgang Bleck (RWTH
has greatly improved in the recent years, yet some fundamental
Aachen), Professor Young-Kook Lee (Yonsei University), Dr Ste-
questions still remain unanswered. There is a general agreement on
phanie Sandlo € bes (Max-Planck-Institut für Eisenforschung, Düs-
the most likely strain hardening mechanisms operating in TWIP
seldorf), Professor Dierk Raabe (Max-Planck-Institut für
steel individually or collectively. These include deformation twin-
Eisenforschung, Düsseldorf), Professor Se Kyun Kwon (POSTECH),
ning, nano-twin hardening, buildup of back-stress, pronounced
and Mr Hojun Gwon (POSTECH). YE also acknowledges support
planar glide of dislocations, and dynamic strain aging. The relative
from the Russian Ministry of Education and Science under grant
magnitude of the individual contributions of these mechanisms is
#14.А12.31.0001 and the Increased Competitiveness Program of
still a matter of debate, however. In certain cases the quantitative
NUST « MISiS», grant #K2-2016-062.
evaluation of the contributions of the various mechanisms to strain
hardening remains uncertain. This is also the case for deformation
twinning. It is now well established that the volume fraction of On-line information on TWIP steel
deformation twins is relatively small, but no precise determination
of the evolution of the twin volume fraction by means of a direct On-line information on TWIP steel is made available by the
experimental method has been reported as yet. It is expected that German Sonderforschungsbereich SFB 761 Project “Stahl - ab initio;
the use of high resolution diffraction techniques will be required to Quantum Mechanics Guided Design of New Fe-based Materials”.
The site (http://abinitio.iehk.rwth-aachen.de/) has a wealth of in- (2014) 20e50.

[30] D. Kuhlmann-Wilsdorf, Theory of plastic deformation: properties of low
formation on TWIP steels.
energy dislocation structures, Mater. Sci. Eng. A 113 (1989) 1e41.
[31] M. Shiekhelsouk, V. Favier, K. Inal, M. Cherkaoui, Modelling the behaviour of
polycrystalline austenitic steel with twinning-induced plasticity effect, Int. J.
References Plast. 25 (2009) 105e133.
[32] S. Dancette, L. Delannay, K. Renard, M. Melchior, P. Jacques, Crystal plasticity
[1] O. Grassel, G. Frommeyer, C. Derder, H. Hofmann, Phase transformations and modeling of texture development and hardening in TWIP steels, Acta Mater
mechanical properties of Fe-Mn-Si-Al TRIP-steels, J. Phys. IV 7 (1997) 60 (2012) 2135e2145.
383e388. [33] O. Bouaziz, N. Guelton, Modelling of TWIP effect on work-hardening, Mater.
[2] O. Gr€ assel, G. Frommeyer, Effect of martensitic phase transformation and Sci. Eng. A 319 (2001) 246e249.
deformation twinning on mechanical properties of Fe-Mn-Si-AI steels, Mater. [34] S. Allain, J.-P. Chateau, O. Bouaziz, A physical model of the twinning-induced
Sci. Tech. 14 (1998) 1213e1217. plasticity effect in a high manganese austenitic steel, Mater. Sci. Eng. A 387
[3] O. Gra €ssel, L. Krüger, G. Frommeyer, L. Meyer, High strength Fe-Mn-(Al, Si) (2004) 143e147.
TRIP/TWIP steels developmentdpropertiesdapplication, Int. J. Plast. 16 [35] O. Bouaziz, S. Allain, C. Scott, Effect of grain and twin boundaries on the
(2000) 1391e1409. hardening mechanisms of twinning-induced plasticity steels, Scr. Mater 58
[4] O. Gra €ssel, G. Frommeyer, High-strength and ultra-ductile FeMn (Al, Si) TRIP/ (2008) 484e487.
TWIP light-weight steels for structural components in automotive engi- [36] J. Kim, Y. Estrin, H. Beladi, I. Timokhina, K.-G. Chin, S.-K. Kim, B.C. De Cooman,
neering, Stahl Eisen 122 (2002) 65e69. Constitutive modeling of the tensile behavior of Al-TWIP steel, Metall. Mater.
[5] G. Frommeyer, U. Brüx, P. Neumann, Supra-ductile and high-strength man- Trans. A 43 (2012) 479e490.
ganese-TRIP/TWIP steels for high energy absorption purposes, ISIJ Int. 43 [37] Y. Estrin, L. Kubin, Local strain hardening and nonuniformity of plastic
(2003) 438e446. deformation, Acta Metall. Mater 34 (1986) 2455e2464.
[6] O. Bouaziz, S. Allain, C. Scott, P. Cugy, D. Barbier, High manganese austenitic [38] G.I. Taylor, The mechanism of plastic deformation of crystals, Proc. R. Soc. A
twinning induced plasticity steels: a review of the microstructure properties 145 (1934) 362e387.
relationships, Curr. Opin. Solid. St. M. 15 (2011) 141e168. [39] R.L. Fullman, Measurement of approximately cylindrical particles in opaque
[7] B.C. De Cooman, O. Kwon, K.-G. Chin, State-of-the-knowledge on TWIP steel, samples, Trans. Metall. AIME 197 (1953) 1267e1268.
Mater Sci. Tech. 28 (2012) 513e527. [40] F. Roters, D. Raabe, G. Gottstein, Work hardening in heterogeneous alloys-a
[8] R. Neu, Performance and characterization of TWIP steels for automotive microstructural approach based on three internal state variables, Acta Mater
applications, Mater. Perform. Charact. 2 (2013) 244e284. 48 (2000) 4181e4189.
[9] K. Chung, K. Ahn, D.-H. Yoo, K.-H. Chung, M.-H. Seo, S.-H. Park, Formability of [41] D.R. Steinmetz, T. Ja €pel, B. Wietbrock, P. Eisenlohr, I. Gutierrez-Urrutia,
TWIP (twinning induced plasticity) automotive sheets, Int. J. Plast. 27 (2011) A. Saeed-Akbari, T. Hickel, F. Roters, D. Raabe, Revealing the strain-hardening
52e81. behavior of twinning-induced plasticity steels: theory, simulations, experi-
[10] B. Qin, Crystallography of TWIP Steel, MS Thesis, Pohang University of Sci- ments, Acta Mater 61 (2013) 494e510.
ence and Technology, 2007. [42] U.F. Kocks, A.S. Argon, M.F. Ashby, Thermodynamics and kinetics of slip, in:
[11] Y. Estrin, H. Mecking, A unified phenomenological description of work Progress in Materials Science, vol. 19, Pergamon Press, 1975.
hardening and creep based on one-parameter models, Acta Metall. Mater 32 [43] A.A. Saleh, E.V. Pereloma, B. Clausen, D.W. Brown, C.N. Tome , A.A. Gazder, On
(1984) 57e70. the evolution and modelling of lattice strains during the cyclic loading of
[12] H. Mecking, B. Nicklas, N. Zarubova, U. Kocks, A “universal” temperature TWIP steel, Acta Mater 61 (2013) 5247e5262.
scale for plastic flow, Acta. Metall. Mater 34 (1986) 527e535. [44] J.G. Sevillano, An alternative model for the strain hardening of FCC alloys that
[13] E. Nes, Modelling of work hardening and stress saturation in FCC metals, twin, validated for twinning-induced plasticity steel, Scr. Mater 60 (2009)
Prog. Mater. Sci. 41 (1997) 129e193. 336e339.
[14] U. Kocks, H. Mecking, Physics and phenomenology of strain hardening: the [45] I. Gutierrez-Urrutia, J. Del Valle, S. Zaefferer, D. Raabe, Study of internal
FCC case, Prog. Mater. Sci. 48 (2003) 171e273. stresses in a TWIP steel analyzing transient and permanent softening during
[15] S. Asgari, E. El-Danaf, S.R. Kalidindi, R.D. Doherty, Strain hardening regimes reverse shear tests, J. Mater. Sci. 45 (2010) 6604e6610.
and microstructural evolution during large strain compression of low [46] J.-L. Collet, Les Me canismes de De formation d'un Acier TWIP FeMnC: une
stacking fault energy fcc alloys that form deformation twins, Metall. Mater.
Etude par Diffraction des Rayons X, Institut National Polytechnique de
Trans. A 28 (1997) 1781e1795. Grenoble-INPG, 2009.
[16] S.R. Kalidindi, Modeling the strain hardening response of low SFE FCC alloys, [47] I. Karaman, H. Sehitoglu, Y.I. Chumlyakov, H.J. Maier, I. Kireeva, The effect of
Int. J. Plast. 14 (1998) 1265e1277. twinning and slip on the Bauschinger effect of Hadfield steel single crystals,
[17] L. Bracke, L. Kestens, J. Penning, Direct observation of the twinning mecha- Metall. Mater. Trans. A 32 (2001) 695e706.
nism in an austenitic Fe-Mn-C steel, Scr. Mater 61 (2009) 220e222. [48] L. Brown, D. Clarke, Work hardening due to internal stresses in composite
[18] J. Kim, Y. Estrin, B.C. De Cooman, Application of a dislocation density-based materials, Acta Metall. 23 (1975) 821e830.
constitutive model to Al-alloyed TWIP steel, Metall, Mater. Trans. A 44 [49] J.G. Sevillano, F. de Las Cuevas, Internal stresses and the mechanism of work
(2013) 4168e4182. hardening in twinning-induced plasticity steels, Scr. Mater 66 (2012)
[19] B. Mahato, S. Shee, T. Sahu, S.G. Chowdhury, P. Sahu, D. Porter, L. Karjalainen, 978e981.
An effective stacking fault energy viewpoint on the formation of extended [50] H. Mughrabi, Dislocation wall and cell structures and long-range internal
defects and their contribution to strain hardening in a Fe-Mn-Si-Al twinning- stresses in deformed metal crystals, Acta Metall. Mater 31 (1983)
induced plasticity steel, Acta Mater 86 (2015) 69e79. 1367e1379.
[20] K. Rahman, V. Vorontsov, D. Dye, The effect of grain size on the twin initi- [51] H. Mughrabi, Dislocation clustering and long-range internal stresses in
ation stress in a TWIP steel, Acta Mater 89 (2015) 247e257. monotonically and cyclically deformed metal crystals, Rev. Phys. Appl. 23
[21] D. Barbier, N. Gey, S. Allain, N. Bozzolo, M. Humbert, Analysis of the tensile (1988) 367e379.
behavior of a TWIP steel based on the texture and microstructure evolutions, [52] H. Idrissi, K. Renard, D. Schryvers, P. Jacques, On the relationship between
Mater. Sci. Eng. A 500 (2009) 196e206. the twin internal structure and the work-hardening rate of TWIP steels, Scr.
[22] T. Lebedkina, M. Lebyodkin, J.-P. Chateau, A. Jacques, S. Allain, On the Mater 63 (2010) 961e964.
mechanism of unstable plastic flow in an austenitic FeMnC TWIP steel, [53] O. Bouaziz, H. Zurob, B. Chehab, J. Embury, S. Allain, M. Huang, Effect of
Mater. Sci. Eng. A 519 (2009) 147e154. chemical composition on work hardening of FedMndC TWIP steels, Mater.
[23] K. Renard, P. Jacques, On the relationship between work hardening and Sci. Tech. 27 (2011) 707e709.
twinning rate in TWIP steels, Mater. Sci. Eng. A 542 (2012) 8e14. [54] J. Nakano, A thermo-mechanical correlation with driving forces for hcp
[24] D. Barbier, N. Gey, N. Bozzolo, S. Allain, M. Humbert, EBSD for analysing the martensite and twin formations in the Fe-Mn-C system exhibiting multi-
twinning microstructure in fine-grained TWIP steels and its influence on composition sets, Sci. Technol. Adv. Mater 14 (2013) 014207.
work hardening, J. Microsc. 235 (2009) 67e78. [55] Y. Dastur, W. Leslie, Mechanism of work hardening in Hadfield manganese
[25] I. Gutierrez-Urrutia, D. Raabe, Dislocation and twin substructure evolution steel, Metall. Mater. Trans. A 12 (1981) 749e759.
during strain hardening of an Fe-22wt.% Mn-0.6 wt.% C TWIP steel observed [56] P. Adler, G. Olson, W. Owen, Strain hardening of Hadfield manganese steel,
by electron channeling contrast imaging, Acta Mater 59 (2011) 6449e6462. Metall. Mater. Trans. A 17 (1986) 1725e1737.
[26] A. Saeed-Akbari, L. Mosecker, A. Schwedt, W. Bleck, Characterization and [57] V. Gerold, H. Karnthaler, On the origin of planar slip in fcc alloys, Acta Metall.
prediction of flow behavior in high-manganese twinning induced plasticity Mater 37 (1989) 2177e2183.
steels: Part I. Mechanism maps and work-hardening behavior, Metall. Mater. [58] J. Fisher, On the strength of solid solution alloys, Acta Metall. Mater 2 (1954)
Trans. A 43 (2012) 1688e1704. 9e10.
[27] I. Gutierrez-Urrutia, S. Zaefferer, D. Raabe, The effect of grain size and grain [59] Y. Brechet, Y. Estrin, Pseudo-portevin-le chatelier effect in ordered alloys,
orientation on deformation twinning in a Fe-22wt.% Mn-0.6 wt.% C TWIP Scr. Mater 35 (1996) 217e223.
steel, Mater. Sci. Eng. A 527 (2010) 3552e3560. [60] K.K. Choi, C.H. Seo, H.C. Lee, K.T. Park, N.J. Kim, in: Proceeding of the 9th
[28] A.A. Saleh, A.A. Gazder, A re-evaluation of “The micromechanics of twinning International Conference on Asia Steels, Busan, Korea, 2009, p. 203.
in a TWIP steel”, Mater. Sci. Eng. A 649 (2016) 184e189. [61] K.-T. Park, K.G. Jin, S.H. Han, S.W. Hwang, K. Choi, C.S. Lee, Stacking fault
[29] S. Zaefferer, N.-N. Elhami, Theory and application of electron channelling energy and plastic deformation of fully austenitic high manganese steels:
contrast imaging under controlled diffraction conditions, Acta Mater 75 effect of Al addition, Mater. Sci. Eng. A 527 (2010) 3651e3661.
[62] J.-E. Jin, Y.-K. Lee, Effects of Al on microstructure and tensile properties of C- (1976) 123e132.
bearing high Mn TWIP steel, Acta Mater 60 (2012) 1680e1688. [97] L. Remy, A. Pineau, Twinning and strain-induced FCC/ HCP transformation
[63] K. Rose, S. Gloverj, A study of strain-aging in austenite, Acta Metall. 14 (1966) in the Fe- Mn-Cr-C system, Mater. Sci. Eng. 28 (1977) 99e107.
1505e1516. [98] A.S.Hamada, Doctoral Thesis, University of Oulu, Linnanmaa, Finland, 2007.
[64] S.-J. Lee, J. Kim, S.N. Kane, B.C. De Cooman, On the origin of dynamic strain [99] S. Vercammen, B.C. De Cooman, N. Akdut, B. Blanpain, P. Wollants, Micro-
aging in twinning-induced plasticity steels, Acta Mater 59 (2011) structural evolution and crystallographic texture formation of cold rolled
6809e6819. austenitic Fe-30Mn-3Al-3Si TWIP-steel, Steel Res. Int. 74 (2003) 370e375.
[65] I. Jung, S.-J. Lee, B.C. De Cooman, Influence of Al on internal friction spectrum [100] B. Oh, S. Cho, Y. Kim, Y. Kim, W. Kim, S. Hong, Effect of aluminium on
of Fe-18Mn-0.6 C twinning-induced plasticity steel, Scr. Mater 66 (2012) deformation mode and mechanical properties of austenitic Fe-Mn-Cr-Al-C
729e732. alloys, Mater. Sci. Eng. A 197 (1995) 147e156.
[66] S. Allain, Universite de Lorraine, These d’Habilitation, INPL, Grenoble, 2012. [101] T.-H. Lee, E. Shin, C.-S. Oh, H.-Y. Ha, S.-J. Kim, Correlation between stacking
[67] J. von Appen, R. Dronskowski, Carbon-induced ordering in manganese-rich fault energy and deformation microstructure in high-interstitial-alloyed
austeniteda density-functional total-energy and chemical-bonding study, austenitic steels, Acta Mater 58 (2010) 3173e3186.
Steel Res. Int. 82 (2011) 101e107. [102] G. Olson, M. Cohen, A general mechanism of martensitic nucleation: Part I.
[68] M. Grujicic, W. Owen, A Chemical interstitial orderdbased model for ni- General concepts and the FCC/ HCP transformation, Metall. Trans. A 7
trogen strengthening of Fe-Ni-Cr austenite, Calphad 16 (1992) 291e299. (1976) 1897e1904.
[69] W. Owen, M. Grujicic, Strain aging of austenitic Hadfield manganese steel, [103] B.Hallstedt, Proceedings of the 2nd international Conference on high Mn
Acta Mater 47 (1998) 111e126. steel HMnS2014, W.Bleck and D.Raabe Editors, Aachen, Germany, August
[70] J. Gegner, Linearisierte Darstellung des Diffusionskoeffizienten von Koh- 31st-September 4th 2014, 113e116.
lenstoff in austenitischem Eisen, Mater. Werkst 36 (2005) 56e61. [104] S.-J. Lee, Y.-S. Jung, S.-I. Baik, Y.-W. Kim, M. Kang, W. Woo, Y.-K. Lee, The
[71] S. Allain, Institute National Polytechnique de Lorraine, Nancy, Doctoral effect of nitrogen on the stacking fault energy in Fe-15Mn-2Cr-0.6 C-xN
Thesis, 2002. twinning-induced plasticity steels, Scr. Mater 92 (2014) 23e26.
[72] S. Vercammen, Doctoral Thesis, Katholieke Universiteit Leuven, 2004. [105] I.-C. Jung, B.C. De Cooman, Temperature dependence of the flow stress of Fe-
[73] B. Huang, X. Wang, Y. Rong, L. Wang, L. Jin, Mechanical behavior and 18Mn-0.6 C-xAl twinning-induced plasticity steel, Acta Mater 61 (2013)
martensitic transformation of an Fe-Mn-Si-Al-Nb alloy, Mater. Sci. Eng. A 438 6724e6735.
(2006) 306e311. [106] X. Lu, Z. Qin, X. Tian, Y. Zhang, B. Ding, Z. Hu, Relations between the lattice
[74] H.Gholizadeh, Doctoral Thesis, 2013, Montan Universit€ at, Leoben, Austria. parameter and the stability of austenite against ε martensite for the Fe-Mn
[75] D. Jiang, E.A. Carter, Carbon dissolution and diffusion in ferrite and austenite based alloys, J. Mater. Sci. Technol. 19 (2003) 443e446.
from first principles, Phys. Rev. B 67 (2003) 214103. [107] M. Onink, C. Brakman, F. Tichelaar, E. Mittemeijer, S. Van der Zwaag, J. Root,
[76] Y. Tian, L. Zhao, S. Chen, A. Shibata, Z. Zhang, N. Tsuji, Significant contribution N. Konyer, The lattice parameters of austenite and ferrite in Fe-C alloys as
of stacking faults to the strain hardening behavior of Cu-15% Al alloy with functions of carbon concentration and temperature, Scr. Metall. Mater 29
different grain sizes, Sci. Rep. 5 (2015). (1993) 1011e1016.
[77] M. Meyers, M. Schneider, H. Jarmakani, B. Kad, B. Remington, D. Kalantar, [108] Y. Endoh, Y. Ishikawa, Antiferromagnetism of g iron manganes alloys, J. Phys.
J. McNaney, B. Cao, J. Wark, Deformation substructures and their transitions Soc. Jpn. 30 (1971) 1614e1627.
in laser shock-compressed copper-aluminum alloys, Metall. Mater. Trans. A [109] G. Inden, The role of magnetism in the calculation of phase diagrams, Phys.
39 (2008) 304e321. Bþ C 103 (1981) 82e100.
[78] H. Beladi, I. Timokhina, Y. Estrin, J. Kim, B. De Cooman, S. Kim, Orientation [110] M. Hillert, M. Jarl, A model for alloying in ferromagnetic metals, Calphad 2
dependence of twinning and strain hardening behaviour of a high manga- (1978) 227e238.
nese twinning induced plasticity steel with polycrystalline structure, Acta [111] S. Cotes, A.F. Guillermet, M. Sade, Fcc/Hcp martensitic transformation in the
Mater 59 (2011) 7787e7799. Fe-Mn system: Part II. Driving force and thermodynamics of the nucleation
[79] J. Bailey, P. Hirsch, The dislocation distribution, flow stress, and stored en- process, Metall. Mater. Trans. A 35 (2004) 83e91.
ergy in cold-worked polycrystalline silver, Philos. Mag. 5 (1960) 485e497. [112] P. Müllner, P.J. Ferreira, On the energy of terminated stacking faults, Philos.
[80] J.-H. Kang, T. Ingendahl, W. Bleck, A constitutive model for the tensile Mag. Lett. 73 (1996) 289e298.
behaviour of TWIP steels: composition and temperature dependencies, [113] P. Volosevich, V. Gridnev, Y. Petrov, Influence of Mn and the stacking fault
Mater. Des. 90 (2016) 340e349. energy of Fe-Mn alloys, Phys. Met. Metallogr. 42 (1976) 126e130.
[81] D. Kuhlmann-Wilsdorf, Advancing towards constitutive equations for the [114] S. Takaki, H. Nakatsu, Y. Tokunaga, Effects of austenite grain size on ε
metal industry via the LEDS theory, Metall. Mater. Trans. A 35 (2004) 369. martensitic transformation in Fe-15mass% Mn alloy, Mater. Trans. JIM 34
[82] L. Remy, The interaction between slip and twinning systems and the influ- (1993) 489e495.
ence of twinning on the mechanical behavior of fcc metals and alloys, Metall. [115] Y.-K. Lee, C. Choi, Driving force for g/ ε martensitic transformation and
Mater. Trans. A 12 (1981) 387e408. stacking fault energy of g in Fe-Mn binary system, Metall. Mater. Trans. A 31
[83] Z.-H. Jin, P. Gumbsch, K. Albe, E. Ma, K. Lu, H. Gleiter, H. Hahn, Interactions (2000) 355e360.
between non-screw lattice dislocations and coherent twin boundaries in [116] J.-H. Jun, C.-S. Choi, Variation of stacking fault energy with austenite grain
face-centered cubic metals, Acta Mater 56 (2008) 1126e1135. size and its effect on the M S temperature of g/ ε martensitic trans-
[84] T. Ezaz, M.D. Sangid, H. Sehitoglu, Energy barriers associated with slip-twin formation in Fe-Mn alloy, Mater. Sci. Eng. A 257 (1998) 353e356.
interactions, Philos. Mag. 91 (2011) 1464e1488. [117] A. Saeed-Akbari, J. Imlau, U. Prahl, W. Bleck, Derivation and variation in
[85] M.D. Sangid, T. Ezaz, H. Sehitoglu, I.M. Robertson, Energy of slip transmission composition-dependent stacking fault energy maps based on subregular
and nucleation at grain boundaries, Acta Mater 59 (2011) 283e296. solution model in high-manganese steels, Metall. Mater. Trans. A 40 (2009)
[86] H. Idrissi, D.Schryvers, Current microscopy contributions to advances in 3076e3090.
science and technology, Microscopy Book Series No. 5, vol. 1, Edited by [118] Y.-K. Lee, Relationship between austenite dislocation density introduced
A.Me ndez-Vilas, 1213e1224. during thermal cycling and M s temperature in an Fe-17 wt pct Mn alloy,
[87] J.P. Hirth, J. Lothe, Theory of Dislocations, second ed., John Willey & Sons, Metall. Mater. Trans. A 33 (2002) 1913e1917.
1982. [119] S. Allain, J.-P. Chateau, O. Bouaziz, S. Migot, N. Guelton, Correlations between
[88] S. Vercammen, B. Blanpain, B. De Cooman, P. Wollants, Cold rolling behav- the calculated stacking fault energy and the plasticity mechanisms in Fe-Mn-
iour of an austenitic Fe-30Mn-3Al-3Si TWIP-steel: the importance of C alloys, Mater. Sci. Eng. A 387 (2004) 158e162.
deformation twinning, Acta Mater 52 (2004) 2005e2012. [120] W.S. Wang, C.M. Wan, The influence of aluminium content to the stacking
[89] H. Idrissi, L. Ryelandt, M. Veron, D. Schryvers, P. Jacques, Is there a rela- fault energy in Fe-Mn-Al-C alloy system, J. Mater. Sci. 25 (1990) 1821e1823.
tionship between the stacking fault character and the activated mode of [121] D. Geissler, J. Freudenberger, A. Kauffmann, S. Martin, D. Rafaja, Assessment
plasticity of Fe-Mn-based austenitic steels? Scr. Mater 60 (2009) 941e944. of the thermodynamic dimension of the stacking fault energy, Philos. Mag.
[90] T. Tisone, The concentration and temperature dependence of the stacking 94 (2014) 2967e2979.
fault energy in face-centered cubic Co-Fe alloys, Acta Metall. 21 (1973) [122] W. Li, S. Lu, D. Kim, K. Kokko, S. Hertzman, S.K. Kwon, L. Vitos, First-principles
229e236. prediction of the deformation modes in austenitic Fe-Cr-Ni alloys, Appl.
[91] H. Idrissi, K. Renard, L. Ryelandt, D. Schryvers, P.J. Jacques, On the mechanism Phys. Lett. 108 (2016) 081903.
of twin formation in Fe-Mn-C TWIP steels, Acta Mater (2010) 2464e2476. [123] S.-J. Lee, Y.-K. Lee, A. Soon, The austenite/3 martensite interface: a first-
[92] I. Karaman, H. Sehitoglu, Y.I. Chumlyakov, H. Maier, I. Kireeva, Extrinsic principles investigation of the fcc Fe (111)/hcp Fe (0001) system, Appl.
stacking faults and twinning in Hadfield manganese steel single crystals, Scr. Surf. Sci. 258 (2012) 9977e9981.
Mater 44 (2001) 337e343. [124] J. Talonen, H. Ha €nninen, Formation of shear bands and strain-induced
[93] D.T. Pierce, J.A. Jime nez, J. Bentley, D. Raabe, C. Oskay, J. Wittig, The influence martensite during plastic deformation of metastable austenitic stainless
of manganese content on the stacking fault and austenite/ε-martensite steels, Acta Mater 55 (2007) 6108e6118.
interfacial energies in Fe-Mn-(Al-Si) steels investigated by experiment and [125] A. Dumay, J.-P. Chateau, S. Allain, S. Migot, O. Bouaziz, Influence of addition
theory, Acta Mater 68 (2014) 238e253. elements on the stacking-fault energy and mechanical properties of an
[94] J. Hirth, Thermodynamics of stacking faults, Metall. Trans. 1 (1970) 2367. austenitic Fe-Mn-C steel, Mater. Sci. Eng. A 483 (2008) 184e187.
[95] R. Reed-Hill, Deformation Twinning, in: TMS-AIME Conference, Am. Inst. [126] S. Curtze, V.-T. Kuokkala, A. Oikari, J. Talonen, H. Ha€nninen, Thermodynamic
Min. Metall. Pet. Eng., 1964. Gainesville Fla. modeling of the stacking fault energy of austenitic steels, Acta Mater 59
[96] L. Remy, A. Pineau, Twinning and strain-induced fcc/ hcp transformation (2011) 1068e1076.
on the mechanical properties of Co-Ni-Cr-Mo alloys, Mater. Sci. Eng. 26 [127] S. Curtze, V.-T. Kuokkala, Dependence of tensile deformation behavior of
TWIP steels on stacking fault energy, temperature and strain rate, Acta Mater made simple, Phys. Rev. Lett. 77 (1996) 3865.
58 (2010) 5129e5141. [161] M. Ekholm, I. Abrikosov, Structural and magnetic ground-state properties of
[128] J. Nakano, P.J. Jacques, Effects of the thermodynamic parameters of the hcp g-FeMn alloys from ab initio calculations, Phys. Rev. B 84 (2011) 104423.
phase on the stacking fault energy calculations in the Fe-Mn and Fe-Mn-C [162] A. Reyes-Huamantinco, P. Puschnig, C. Ambrosch-Draxl, O.E. Peil, A.V. Ruban,
systems, Calphad 34 (2010) 167e175. Stacking-fault energy and anti-Invar effect in Fe-Mn alloy from first princi-
[129] D. Djurovic, B. Hallstedt, J. Von Appen, R. Dronskowski, Thermodynamic ples, Phys. Rev. B 86 (2012) 060201.
assessment of the Mn-C system, Calphad 34 (2010) 279e285. [163] L. Vitos, P.A. Korzhavyi, B. Johansson, Evidence of large magnetostructural
[130] D. Djurovic, B. Hallstedt, J. von Appen, R. Dronskowski, Thermodynamic effects in austenitic stainless steels, Phys. Rev. Lett. 96 (2006) 117210.
assessment of the Fe-Mn-C system, Calphad 35 (2011) 479e491. [164] P. Denteneer, W. van Haeringen, Stacking-fault energies in semiconductors
[131] H.-S. Yang, J. Jang, H. Bhadeshia, D. Suh, Critical assessment: martensite-start from first-principles calculations, J. Phys. C. Solid State Phys. 20 (1987) L883.
temperature for the g/ ε transformation, Calphad 36 (2012) 16e22. [165] L. Vitos, J.-O. Nilsson, B. Johansson, Alloying effects on the stacking fault
[132] V. Schumann, Martensitische Umwandlung in austenitischen mangan-koh- energy in austenitic stainless steels from first-principles theory, Acta Mater
lenstoff-st€ahlen, Neue Hütte 17 (1972) 605e609. 54 (2006) 3821e3826.
[133] H. Van Swygenhoven, P. Derlet, A. Frøseth, Stacking fault energies and slip in [166] A. Dick, T. Hickel, J. Neugebauer, The effect of disorder on the concentration-
nanocrystalline metals, Nat. Mater 3 (2004) 399e403. dependence of stacking fault energies in Fe1-xMnx-a first principles study,
[134] V. Vitek, Intrinsic stacking faults in body-centred cubic crystals, Philos. Mag. Steel Res. Int. 80 (2009) 603e608.
18 (1968) 773e786. [167] P. Denteneer, J. Soler, Energetics of point and planar defects in aluminium
[135] V. Vitek, Multilayer stacking faults and twins on {211} planes in bcc metals, from first-principles calculations, Solid. State. Commun. 78 (1991) 857e861.
Scr. Metall. 4 (1970) 725e732. [168] R. Van Tol, J. Kim, L. Zhao, J. Sietsma, B. De Cooman, a0 -Martensite formation
[136] R.C. Pond, J. Hirth, A. Serra, D. Bacon, Atomic displacements accompanying in deep-drawn Mn-based TWIP steel, J. Mater. Sci. 47 (2012) 4845e4850.
deformation twinning: shears and shuffles, Mater. Res. Lett. 4 (2016) [169] R. Van Tol, Microstructural Evolution in Deformed Austenitic TWinning
185e190. Induced Plasticity Steels, TU Delft, Delft University of Technology, 2014.
[137] T. Gebhardt, D. Music, M. Ekholm, I.A. Abrikosov, L. Vitos, A. Dick, T. Hickel, [170] J. Kim, S.-J. Lee, B.C. De Cooman, Effect of Al on the stacking fault energy of
J. Neugebauer, J.M. Schneider, The influence of additions of Al and Si on the Fe-18Mn-0.6 C twinning-induced plasticity, Scr. Mater 65 (2011) 363e366.
lattice stability of fcc and hcp Fe-Mn random alloys, J. Phys.-Condens. Mat. 23 [171] J. Jeong, W. Woo, K. Oh, S. Kwon, Y. Koo, In situ neutron diffraction study of
(2011) 246003. the microstructure and tensile deformation behavior in Al-added high
[138] J.R. Rice, Dislocation nucleation from a crack tip: an analysis based on the manganese austenitic steels, Acta Mater 60 (2012) 2290e2299.
Peierls concept, J. Mech. Phys. Solids 40 (1992) 239e271. [172] K. Jeong, J.-E. Jin, Y.-S. Jung, S. Kang, Y.-K. Lee, The effects of Si on the me-
[139] E. Tadmor, S. Hai, A peierls criterion for the onset of deformation twinning at chanical twinning and strain hardening of Fe-18Mn-0.6 C twinning-induced
a crack tip, J. Mech. Phys. Solids 51 (2003) 765e793. plasticity steel, Acta Mater 61 (2013) 3399e3410.
[140] S. Ogata, J. Li, S. Yip, Energy landscape of deformation twinning in bcc and fcc [173] K. Oda, H. Fujimura, H. Ino, Local interactions in carbon-carbon and carbon-
metals, Phys. Rev. B 71 (2005) 224102. M (M: Al, Mn, Ni) atomic pairs in FCC gamma-iron, J. Phys.-Condens. Mat. 6
[141] Z. Jin, S. Dunham, H. Gleiter, H. Hahn, P. Gumbsch, A universal scaling of (1994) 679.
planar fault energy barriers in face-centered cubic metals, Scr. Mater 64 [174] H. Bhadeshia, Carbon-carbon interactions in iron, J. Mater. Sci. 39 (2004)
(2011) 605e608. 3949e3955.
[142] S. Kibey, J. Liu, D. Johnson, H. Sehitoglu, Predicting twinning stress in fcc [175] R. McLellan, C. Ko, The diffusion of carbon in austenite, Acta. Metall. Mater 36
metals: linking twin-energy pathways to twin nucleation, Acta Mater 55 (1988) 531e537.
(2007) 6843e6851. [176] S. Babu, H. Bhadeshia, Diffusion of carbon in substitutionally alloyed
[143] W. Li, S. Lu, Q.-M. Hu, S.K. Kwon, B. Johansson, L. Vitos, Generalized stacking austenite, J. Mater. Sci. Lett. 14 (1995) 314e316.
fault energies of alloys, J. Phys. -Condens. Mat 26 (2014) 265005. [177] K. Limmer, J.E. Medvedeva, D.C. Van Aken, N. Medvedeva, Ab initio simula-
[144] M. Jo, Y.M. Koo, B.-J. Lee, B. Johansson, L. Vitos, S.K. Kwon, Theory for plas- tion of alloying effect on stacking fault energy in fcc Fe, Comp. Mater. Sci. 99
ticity of face-centered cubic metals, P. Natl. Acad. Sci. 111 (2014) (2015) 253e255.
6560e6565. [178] D. Hepburn, D. Ferguson, S. Gardner, G. Ackland, First-principles study of
[145] R.J. Asaro, S. Suresh, Mechanistic models for the activation volume and rate helium, carbon, and nitrogen in austenite, dilute austenitic iron alloys, and
sensitivity in metals with nanocrystalline grains and nano-scale twins, Acta nickel, Phys. Rev. B 88 (2013) 024115.
Mater 53 (2005) 3369e3382. [179] T. Shun, C. Wan, J. Byrne, A study of work hardening in austenitic Fe- Mn- C
[146] T. Byun, On the stress dependence of partial dislocation separation and and Fe- Mn- Al- C alloys, Acta Metall. Mater 40 (1992) 3407e3412.
deformation microstructure in austenitic stainless steels, Acta Mater 51 [180] K. Haneda, Z. Zhou, A. Morrish, T. Majima, T. Miyahara, Low-temperature
(2003) 3063e3071. stable nanometer-size fcc-Fe particles with no magnetic ordering, Phys. Rev.
[147] N. Bernstein, E. Tadmor, Tight-binding calculations of stacking energies and B 46 (1992) 13832.
twinnability in fcc metals, Phys. Rev. B 69 (2004) 094116. [181] L. Del Bianco, C. Ballesteros, J. Rojo, A. Hernando, Magnetically ordered fcc
[148] E. Tadmor, N. Bernstein, A first-principles measure for the twinnability of structure at the relaxed grain boundaries of pure nanocrystalline Fe, Phys.
FCC metals, J. Mech. Phys. Solids 52 (2004) 2507e2519. Rev. Lett. 81 (1998) 4500.
[149] V. García-Sua rez, C. Newman, C.J. Lambert, J. Pruneda, J. Ferrer, First princi- [182] W.A. Macedo, F. Sirotti, G. Panaccione, A. Schatz, W. Keune, W.N. Rodrigues,
ples simulations of the magnetic and structural properties of Iron, Eur. Phys. G. Rossi, Magnetism of atomically thin fcc Fe overlayers on an expanded fcc
J. B 40 (2004) 371e377. lattice: Cu 84 Al 16(100), Phys. Rev. B 58 (1998) 11534.
[150] T. Gebhardt, D. Music, B. Hallstedt, M. Ekholm, I.A. Abrikosov, L. Vitos, [183] Y. Tsunoda, Y. Nishioka, R. Nicklow, Spin fluctuations in small g-Fe pre-
J. Schneider, Ab initio lattice stability of fcc and hcp Fe-Mn random alloys, cipitates, J. Magn. Magn. Mater 128 (1993) 133e137.
J. Phys.-Condens. Mat. 22 (2010) 295402. [184] T. Hickel, S. Sandlo €bes, R.K. Marceau, A. Dick, I. Bleskov, J. Neugebauer,
[151] N. Medvedeva, D. Van Aken, J.E. Medvedeva, Magnetism in bcc and fcc Fe D. Raabe, Impact of nanodiffusion on the stacking fault energy in high-
with carbon and manganese, J. Phys.-Condens. Mat. 22 (2010) 316002. strength steels, Acta Mater 75 (2014) 147e155.
[152] N. Medvedeva, D.C. Van Aken, J.E. Medvedeva, The effect of carbon distri- [185] K.R. Limmer, First-principles Investigations of Iron-based Alloys and Their
bution on the manganese magnetic moment in bcc Fe-Mn alloy, J. Phys.- Properties, Missouri University of Science and Technology, 2014.
Condens. Mat. 23 (2011) 326003. [186] W.S. Choi, B.C. De Cooman, S. Sandlo €bes, D. Raabe, Size and orientation ef-
[153] N.G. Kioussis, N.M. Ghoniem, Modeling of dislocation interaction with sol- fects in partial dislocation-mediated deformation of twinning-induced
utes, nano-precipitates and interfaces: a multiscale challenge, J. Comput. plasticity steel micro-pillars, Acta Mater 98 (2015) 391e404.
Theor. Nanosci. 7 (2010) 1317e1346. [187] K. Irvine, T. Gladman, F. Pickering, The strength of austenitic stainless steels,
[154] D. Boukhvalov, Y.N. Gornostyrev, M. Katsnelson, A. Lichtenstein, Magnetism J. Iron Steel. Inst. 207 (1969) 1017e1028.
and local distortions near carbon impurity in g-iron, Phys. Rev. Lett. 99 [188] P. Kusakin, A. Belyakov, D.A. Molodov, R. Kaibyshev, On the effect of chemical
(2007) 247205. composition on yield strength of TWIP steels, Mater. Sci. Eng. A 687 (2017)
[155] A. Abbasi, A. Dick, T. Hickel, J. Neugebauer, First-principles investigation of 82e84.
the effect of carbon on the stacking fault energy of Fe-C alloys, Acta Mater 59 [189] J.-H. Kang, T. Ingendahl, J. von Appen, R. Dronskowski, W. Bleck, Impact of
(2011) 3041e3048. short-range ordering on yield strength of high manganese austenitic steels,
[156] S. Kibey, J. Liu, M. Curtis, D. Johnson, H. Sehitoglu, Effect of nitrogen on Mater. Sci. Eng. A 614 (2014) 122e128.
generalized stacking fault energy and stacking fault widths in high nitrogen [190] D. Parris, R.B. McLellan, The diffusivity of carbon in austenite, Acta Metall. 24
steels, Acta Mater 54 (2006) 2991e3001. (1976) 523e528.
[157] N. Medvedeva, M. Park, D.C. Van Aken, J.E. Medvedeva, First-principles study [191] R.B. McLellan, Cell models for interstitial solid solutions, Acta Metall. 30
of Mn, Al and C distribution and their effect on stacking fault energies in fcc (1982) 317e322.
Fe, J. Alloy. Compd. 582 (2014) 475e482. [192] S.-J. Lee, D.K. Matlock, C.J. Van Tyne, Carbon diffusivity in multi-component
[158] L. Vitos, Computational Quantum Mechanics for Materials Engineers: the austenite, Scr. Mater 64 (2011) 805e808.
EMTO Method and Applications, Springer Science & Business Media, 2007. [193] B.C. De Cooman, K.-G. Chin, J. Kim, High Mn TWIP Steels for Automotive
[159] M. Petersilka, U. Gossmann, E. Gross, Time-dependent Optimized Effective Applications, New Trends and Developments in Automotive System Engi-
Potential in the Linear Response Regime, Electronic Density Functional neering, InTech, 2011.
Theory, Springer, 1998, pp. 177e197. [194] M. Kang, W. Woo, Y.-K. Lee, B.-S. Seong, Neutron diffraction analysis of
[160] J.P. Perdew, K. Burke, M. Ernzerhof, Generalized gradient approximation stacking fault energy in Fe-18Mn-2Al-0.6 C twinning-induced plasticity
steels, Mater. Lett. 76 (2012) 93e95. [228] H.-W. Yen, M. Huang, C. Scott, J.-R. Yang, Interactions between deformation-
[195] T.S. Shun, C. Wan, J. Byrne, Serrated flow in austenitic Fe-Mn-C and Fe-Mn- induced defects and carbides in a vanadium-containing TWIP steel, Scr.
Al-C alloys, Scr. Metall. Mater 25 (1991) 1769e1774. Mater 66 (2012) 1018e1023.
[196] X. Tian, Y. Zhang, Effect of Si content on the stacking fault energy in g-Fe- [229] R.Z. Valiev, Y. Estrin, Z. Horita, T.G. Langdon, M.J. Zechetbauer, Y.T. Zhu,
Mn-Si-C alloys: Part I. X-ray diffraction line profile analysis, Mater. Sci. Eng. A Producing bulk ultrafine-grained materials by severe plastic deformation,
516 (2009) 73e77. JOM 58 (2006) 33e39.
[197] B. Huang, X. Wang, L. Wang, Y. Rong, Effect of nitrogen on stacking fault [230] Y. Estrin, A. Vinogradov, Extreme grain refinement by severe plastic defor-
formation probability and mechanical properties of twinning-induced plas- mation: a wealth of challenging science, Acta Mater 61 (2013) 782e817.
ticity steels, Metall. Mater. Trans. A 39 (2008) 717e724. [231] R. Ueji, N. Tsuchida, H. Fujii, D. Kondo, K. Kunishige, Effect of grain size on
[198] B. Jiang, X. Qi, S. Yang, W. Zhou, T. Hsu, Effect of stacking fault probability on tensile properties of TWIP steel, J. Jpn. I. Met. 71 (2007) 815e821.
g-ε martensitic transformation and shape memory effect in Fe-Mn-Si based [232] R. Ueji, N. Tsuchida, D. Terada, N. Tsuji, Y. Tanaka, A. Takemura, K. Kunishige,
alloys, Acta Mater 46 (1998) 501e510. Tensile properties and twinning behavior of high manganese austenitic steel
[199] S. Lua, W. Lib, S.K. Kwon, K. Kokkoa, Q.M. Hud, S. Hertzman, L. Vitos, Scripta with fine-grained structure, Scr. Mater 59 (2008) 963e966.
Materialia, 2015. [233] I. Gutierrez-Urrutia, S. Zaefferer, D. Raabe, The effect of grain size and grain
[200] S. Lee, J. Kim, S.-J. Lee, B.C. De Cooman, Effect of Cu addition on the me- orientation on deformation twinning in a Fe-22 wt.% Mn-0.6 wt.% C TWIP
chanical behavior of austenitic twinning-induced plasticity steel, Scr. Mater steel, Mater. Sci. Eng. A 66 (2010) 3552e3560.
65 (2011) 1073e1076. [234] I. Timokhina, A. Medvedev, R. Lapovok, Severe plastic deformation of a TWIP
[201] M. Müller, An antiferromagnetic temperature-compensating elastic Elinvar- steel, Mater. Sci. Eng. A 593 (2014) 163e169.
alloy on the basis of Fe-Mn, J. Magn. Magn. Mater 78 (1989) 337e346. [235] G. Fontaine, Doctoral Thesis, Orsay, France, 1968.
[202] G. Hausch, Elastic and magnetoelastic effects in invar alloys, J. Magn. Magn. [236] J.W. Christian, S. Mahajan, Deformation twinning, Prog. Mater. Sci. 39 (1995)
Mater 10 (1979) 163e169. 1e157.
[203] S. Reeh, D. Music, T. Gebhardt, M. Kasprzak, T. Ja €pel, S. Zaefferer, D. Raabe, [237] M. Niewczas, Dislocations and twinning in face centred cubic crystals, Disloc.
S. Richter, A. Schwedt, J. Mayer, Elastic properties of face-centred cubic Fe- Solids 13 (2007) 263e364.
Mn-C studied by nanoindentation and ab initio calculations, Acta Mater 60 [238] X. Liao, S. Srinivasan, Y. Zhao, M. Baskes, Y. Zhu, F. Zhou, E. Lavernia, H. Xu,
(2012) 6025e6032. Formation mechanism of wide stacking faults in nanocrystalline Al, Appl.
[204] H. Yang, V. Doquet, Z. Zhang, Micro-scale measurements of plastic strain Phys. Lett. 84 (2004) 3564e3566.
field, and local contributions of slip and twinning in TWIP steels during in [239] J. Narayan, Y. Zhu, Self-thickening, Cross-slip deformation twinning model,
situ tensile tests, Mater. Sci. Eng. A 672 (2016) 7e14. Appl. Phys. Lett. 92 (2008) 151908.
[205] T. Gebhardt, D. Music, D. Kossmann, M. Ekholm, I.A. Abrikosov, L. Vitos, [240] Y. Zhu, J. Narayan, J. Hirth, S. Mahajan, X. Wu, X. Liao, Formation of single and
J.M. Schneider, Elastic properties of fcc Fe-Mn-X (X¼ Al, Si) alloys studied by multiple deformation twins in nanocrystalline fcc metals, Acta Mater 57
theory and experiment, Acta Mater 59 (2011) 3145e3155. (2009) 3763e3770.
[206] D. Pierce, K. Nowag, A. Montagne, J. Jime nez, J. Wittig, R. Ghisleni, Single [241] R.J. McCabe, I.J. Beyerlein, J.S. Carpenter, N.A. Mara, The critical role of grain
crystal elastic constants of high-manganese transformation-and twinning- orientation and applied stress in nanoscale twinning, Nat, Commun 5 (2014).
induced plasticity steels determined by a new method utilizing nano- [242] J. Cohen, J. Weertman, A dislocation model for twinning in fcc metals, Acta
indentation, Mater. Sci. Eng. A 578 (2013) 134e139. Metall. 11 (1963) 996e998.
[207] M. van Schilfgaarde, I. Abrikosov, B. Johansson, Origin of the Invar effect in [243] H. Fujita, T. Mori, A formation mechanism of mechanical twins in FCC Metals,
iron-nickel alloys, Nature 400 (1999) 46e49. Scr. Mater 9 (1975) 631e636.
[208] S. Pugh, XCII, Relations between the elastic moduli and the plastic properties [244] S. Mahajan, G. Chin, Comments on deformation twinning in silver-and
of polycrystalline pure metals, the London, Edinburgh, and Dublin, Philos. copper-alloy crystals, Scr. Metall. 9 (1975) 815e817.
Mag. J. Sci. 45 (1954) 823e843. [245] S. Miura, J.-I. Takamura, N. Narita, Orientation dependence of flow stress for
[209] D. Pettifor, Theoretical predictions of structure and related properties of twinning in silver crystals, Trans. Jpn. Inst. Met. 9 (1968) 555e561.
intermetallics, Mater. Sci. Tech. 8 (1992) 345e349. [246] S. Copley, B. Kear, The dependence of the width of a dissociated dislocation
[210] I.C.Jung, Doctoral Thesis, Pohang University of Science and Technology, 2013, on dislocation velocity, Acta Metall. 16 (1968) 227e231.
South Korea. [247] J. Hirth, On the pole mechanism in FCC metals, Deform. Twinning 25 (1964)
[211] S. Wu, H. Yen, M. Huang, A. Ngan, Deformation twinning in submicron and 112e115.
micron pillars of twinning-induced plasticity steel, Scr. Mater 67 (2012) [248] J. Venables, On dislocation pole models for twinning, Philos. Mag. 30 (1974)
641e644. 1165e1169.
[212] Q. Yu, Z.-W. Shan, J. Li, X. Huang, L. Xiao, J. Sun, E. Ma, Strong crystal size [249] J. Venables, The nucleation and propagation of deformation twins, J. Phys.
effect on deformation twinning, Nature 463 (2010) 335e338. Chem. Solids 25 (1964) 693e700.
[213] S. Allain, O. Bouaziz, J. Chateau, Thermally activated dislocation dynamics in [250] M. Niewczas, G. Saada, Twinning nucleation in Cu-8 at.% Al single crystals,
austenitic FeMnC steels at low homologous temperature, Scr. Mater 62 Philos. Mag. A 82 (2002) 167e191.
(2010) 500e503. [251] J. Liu, X. Liu, W. Liu, Y. Zeng, K. Shu, Transmission electron microscopy
[214] J.N. Wang, A new modification of the formulation of Peierls stress, Acta observation of a deformation twin in TWIP steel by an ex situ tensile test,
Mater 44 (1996) 1541e1546. Philos. Mag. 91 (2011) 4033e4044.
[215] J. Wang, Prediction of Peierls stresses for different crystals, Mater. Sci. Eng. A [252] I. Karaman, H. Sehitoglu, K. Gall, Y.I. Chumlyakov, H. Maier, Deformation of
206 (1996) 259e269. single crystal Hadfield steel by twinning and slip, Acta Mater 48 (2000)
[216] O. Majidi, B.C. De Cooman, F. Barlat, M.-G. Lee, Y.P. Korkolis, Thermo- 1345e1359.
mechanical response of a TWIP steel during monotonic and non-monotonic
[253] S. Martin, C. Ullrich, D. Simek, U. Martin, D. Rafaja, Stacking fault model of∊-
uniaxial loading, Mater. Sci. Eng. A 674 (2016) 276e285. martensite and its DIFFaX implementation, J. Appl. Crystallogr. 44 (2011)
[217] G. Dini, A. Najafizadeh, R. Ueji, S. Monir-Vaghefi, Tensile deformation 779e787.
behavior of high manganese austenitic steel: the role of grain size, Mater, [254] I. Beyerlein, C. Tome , A probabilistic twin nucleation model for HCP poly-
Des 31 (2010) 3395e3402. crystalline metals, Proc. R. Soc. A 466 (2010) 2517e2544.
[218] C. Scott, B. Remy, J.-L. Collet, A. Cael, C. Bao, F. Danoix, B. Malard, C. Curfs, [255] A. Vinogradov, E. Vasilev, D. Merson, Y. Estrin, A phenomenological model of
Precipitation strengthening in high manganese austenitic TWIP steels, Int. J. twinning kinetics, Adv. Eng. Mater 19 (2017) 1e10.
Mater. Res. 102 (2011) 538e549. [256] N. Narita, J. Takamura, Deformation twinning in silver-and copper-alloy
[219] S.Lee, B.C.De Cooman, unpublished research results. crystals, Philos. Mag. 29 (1974) 1001e1028.
[220] Y. Shen, N. Jia, R. Misra, L. Zuo, Softening behavior by excessive twinning and [257] H. Suzuki, C. Barrett, Deformation twinning in silver-gold alloys, Acta Metall.
adiabatic heating at high strain rate in a Fe-20Mn-0.6 C TWIP steel, Acta 6 (1958) 156e165.
Mater 103 (2016) 229e242. [258] M. Meyers, O. Vo €hringer, V. Lubarda, The onset of twinning in metals: a
[221] H. Gwon, MS Thesis, Pohang University of Science and Technology, South constitutive description, Acta Mater 49 (2001) 4025e4039.
Korea, 2017. [259] K. Phiu-on, W. Bleck, Deformation Mechanisms and Mechanical Properties of
[222] J.-H. Kang, S. Duan, S.-J. Kim, W. Bleck, Grain boundary strengthening in high Hot Rolled Fe-mn-c-(Al)-(Si) Austenitic Steels, Lehrstuhl und Institut für
Mn austenitic steels, Metall. Mater. Trans. A 47 (2016) 1918e1921. Eisenhüttenkunde, 2008.
[223] J. Li, Y. Chou, The role of dislocations in the flow stress grain size relation- [260] A.A. Mohammed, E.A. El-Danaf, A.-K.A. Radwan, Equivalent twinning criteria
ships, Metall. Mater. Trans. 1 (1970) 1145. for FCC alloys under uniaxial tension at high temperatures, Mater. Sci. Eng. A
[224] F. De Las Cuevas, M. Reis, A. Ferraiuolo, G. Pratolongo, L.P. Karjalainen, 457 (2007) 373e379.
J. Alkorta, J. Gil Sevillano, Hall-Petch relationship of a TWIP steel, Key Eng. [261] S.A. Kibey, L.-L. Wang, J. Liu, H. Johnson, H. Sehitoglu, D.D. Johnson, Quan-
Mat. Trans. Tech. Publ. (2010) 147e152. titative prediction of twinning stress in fcc alloys: application to Cu-Al, Phys.
[225] S.-H. Wang, Z.-Y. Liu, G.-F. Wang, Influence of grain size on TWIP effect in a Rev. B 79 (2009) 214202.
TWIP steel, Acta Mater. Sin. 45 (2009) 1083. [262] G. Dini, R. Ueji, A. Najafizadeh, S. Monir-Vaghefi, Flow stress analysis of TWIP
[226] T. Lee, M. Koyama, K. Tsuzaki, Y.-H. Lee, C.S. Lee, Tensile deformation steel via the XRD measurement of dislocation density, Mater. Sci. Eng. A 527
behavior of Fe-Mn-C TWIP steel with ultrafine elongated grain structure, (2010) 2759e2763.
Mater. Lett. 75 (2012) 169e171. [263] K. Rahman, N. Jones, D. Dye, Micromechanics of twinning in a TWIP steel,
[227] J. Chateau, A. Dumay, S. Allain, A. Jacques, Precipitation hardening of a FeMnC Mater. Sci. Eng. A 635 (2015) 133e142.
TWIP steel by vanadium carbides, J. Phys. Conf. Ser. IOP Publ. (2010) 012023. [264] I. Karaman, H. Sehitoglu, A. Beaudoin, Y.I. Chumlyakov, H. Maier, C. Tome,
Modeling the deformation behavior of Hadfield steel single and polycrystals TWIP steels, Scr. Mater 66 (2012) 1007e1011.
due to twinning and slip, Acta Mater 48 (2000) 2031e2047. [296] L. Barrales-Mora, Y. Lü, D. Molodov, Experimental determination and
[265] D. Barbier, Doctoral Thesis, Universite Paul Verlaine de Metz, France. simulation of annealing textures in cold rolled TWIP and TRIP steels, Steel
[266] J. Kim, Doctoral Thesis, Pohang University of Science and Technology, South Res. Int. 82 (2011) 119e126.
Korea. [297] Y. Estrin, L. Kubin, Spatial Coupling and Propagative Plastic Instabilities,
[267] A. Soulami, K.S. Choi, Y. Shen, W.N. Liu, X. Sun, M.A. Khaleel, On deformation Continuum Models for Materials with Microstructure, 1995, pp. 395e450.
twinning in a 17.5% Mn-TWIP steel: a physically based phenomenological [298] M. Dao, L. Lu, Y. Shen, S. Suresh, Strength, strain-rate sensitivity and ductility
model, Mater. Sci. Eng. A 528 (2011) 1402e1408. of copper with nanoscale twins, Acta Mater 54 (2006) 5421e5432.
[268] C. Haase, L.A. Barrales-Mora, F. Roters, D.A. Molodov, G. Gottstein, Applying [299] L. Chen, J.K. Kim, S.K. Kim, G.S. Kim, K.G. Chin, B. De Cooman, Stretch-flan-
the texture analysis for optimizing thermomechanical treatment of high geability of high Mn TWIP steel, Steel Res. Int. 81 (2010) 552e568.
manganese twinning-induced plasticity steel, Acta Mater 80 (2014) [300] L. Xu, L. Chen, B. De Cooman, D. Steglich, F. Barlat, Hole expansion of
327e340. advanced high strenth steel sheet sample, Inter. J. Mater. Form. 3 (2010)
[269] J.C. Fisher, E.W. Hart, R.H. Pry, Phys. Rev. 79 (1950) 722. 247e250.
[270] V. Shterner, A. Molotnikov, I. Timokhina, Y. Estrin, H. Beladi, A constitutive [301] J.-K. Kim, L. Chen, H.-S. Kim, S.-K. Kim, Y. Estrin, B. De Cooman, On the tensile
model of the deformation behaviour of twinning induced plasticity (TWIP) behavior of high-manganese twinning-induced plasticity steel, Metall.
steel at different temperatures, Mater. Sci. Eng. A 613 (2014) 224e231. Mater. Trans. A 40 (2009) 3147e3158.
[271] G. Piercy, R. Cahn, A. Cottrell, A study of primary and conjugate slip in [302] L. Qian, P. Guo, J. Meng, F. Zhang, Unusual grain-size and strain-rate effects
crystals of alpha-brass, Acta Metall. 3 (1955) 331e338. on the serrated flow in FeMnC twin-induced plasticity steels, J. Mater. Sci. 48
[272] T. Leffers, Deformation rate dependence of rolling texture in brass containing (2013) 1669e1674.
5% zinc, Scr. Metall. 2 (1968) 447e452. [303] A. Saeed-Akbari, A.K. Mishra, J. Mayer, W. Bleck, Metall. Mater. Trans. A 43
[273] T. Leffers, R. Ray, The brass-type texture and its deviation from the copper- (2012) 1705e1723.
type texture, Prog. Mater. Sci. 54 (2009) 351e396. [304] A. B€aumer, W. Bleck, Proceeding of IDDRG 2007, M.Tisza Editor, 2007, 47.
[274] S. Sato, E.-P. Kwon, M. Imafuku, K. Wagatsuma, S. Suzuki, Microstructural [305] K. Renard, S. Ryelandt, P. Jacques, Characterisation of the Portevin-Le
characterization of high-manganese austenitic steels with different stacking Châtelier effect affecting an austenitic TWIP steel based on digital image
fault energies, Mater. Charact. 62 (2011) 781e788. correlation, Mater. Sci. Eng. A 527 (2010) 2969e2977.
[275] I. Gutierrez-Urrutia, D. Raabe, Study of deformation twinning and planar slip [306] L. Chen, H.-S. Kim, S.-K. Kim, B. De Cooman, Localized deformation due to
in a TWIP steel by electron channeling contrast imaging in a SEM, Mater. Sci. Portevin-LeChatelier effect in 18Mn-0.6 C TWIP austenitic steel, ISIJ Int. 47
Forum (2012) 523e529. (2007) 1804e1812.
[276] D. Geissler, J. Freudenberger, A. Kauffmann, M. Krautz, H. Klauss, A. Voss, [307] H. Karabulut, S. Gündüz, Effect of vanadium content on dynamic strain aging
J. Eickemeyer, L. Schultz, Appearance of dislocation-mediated and twinning- in microalloyed medium carbon steel, Mater. Des. 25 (2004) 521e527.
induced plasticity in an engineering-grade FeMnNiCr alloy, Acta Mater. 59 [308] L. Kubin, Y. Estrin, The critical conditions for jerky flow. Discussion and
(2011) 7711e7723. application to CuMn solid solutions, Phys. Status Solidi B 172 (1992)
[277] E.I. Kuprekova, Y.I. Chumlyakov, I.P. Chernov, Dependence of critical cleavage 173e185.
stresses as a function of orientation and temperature in single crystals of Fe- [309] S.D. Andrews, H. Sehitoglu, I. Karaman, Constriction energy in the presence
18% Cr-14% Ni-2% Mo austenitic stainless steel containing hydrogen, Met. Sci. of a solute field, J. Appl. Phys. 87 (2000) 2194e2203.
Heat. Treat. 50 (2008) 282e288. [310] M. Koyama, T. Sawaguchi, K. Tsuzaki, Deformation twinning behavior of
[278] P. Yang, Q. Xie, L. Meng, H. Ding, Z. Tang, Dependence of deformation twinning-induced plasticity steels with different carbon Concentrations-Part
twinning on grain orientation in a high manganese steel, Scr. Mater 55 2: proposal of dynamic-strain-aging-assisted deformation twinning, ISIJ Int.
(2006) 629e631. 55 (2015) 1754e1761.
[279] A. Prakash, T. Hochrainer, E. Reisacher, H. Riedel, Twinning models in self- [311] A. Van den Beukel, Theory of the effect of dynamic strain aging on me-
consistent texture simulations of TWIP steels, Steel Res. Int. 79 (2008) chanical properties, Phys. Status Solidi A 30 (1975) 197e206.
645e652. [312] S. Allain, O. Bouaziz, T. Lebedkina, M. Lebyodkin, Relationship between
[280] B.C. De Cooman, J. Kim, S. Lee, Heterogeneous deformation in twinning- relaxation mechanisms and strain aging in an austenitic FeMnC steel, Scr.
induced plasticity steel, Scr. Mater 66 (2012) 986e991. Mater 64 (2011) 741e744.
[281] J. Hirsch, K. Lücke, Overview no. 76: mechanism of deformation and devel- [313] N. Tsuji, Y. Ito, Y. Saito, Y. Minamino, Strength and ductility of ultrafine
opment of rolling textures in polycrystalline fcc metalsdI. Description of grained aluminum and iron produced by ARB and annealing, Scr. Mater 47
rolling texture development in homogeneous CuZn alloys, Acta Metall. 36 (2002) 893e899.
(1988) 2863e2882. [314] S. Kang, Y.-S. Jung, J.-H. Jun, Y.-K. Lee, Effects of recrystallization annealing
[282] J. Hirsch, K. Lücke, M. Hatherly, Overview No. 76: mechanism of deformation temperature on carbide precipitation, microstructure, and mechanical
and development of rolling textures in polycrystalline f.c.c. MetalsdIII. The properties in Fe-18Mn-0.6 C-1.5 Al TWIP steel, Mater. Sci. Eng. A 527 (2010)
influence of slip inhomogeneities and twinning, Acta Metall. 36 (1988) 745e751.
2905e2927. [315] G. Dini, A. Najafizadeh, R. Ueji, S. Monir-Vaghefi, Improved tensile properties
[283] R. Smallman, D. Green, The dependence of rolling texture on stacking fault of partially recrystallized submicron grained TWIP steel, Mater. Lett. 64
energy, Acta Metall. 12 (1964) 145e154. (2010) 15e18.
[284] I. Dillamore, The stacking fault energy dependence of the mechanisms of [316] O. Bouaziz, C. Scott, G. Petitgand, Nanostructured steel with high work-
deformation in fcc metals, Metall. Trans. 1 (1970) 2463e2470. hardening by the exploitation of the thermal stability of mechanically
[285] K. Morii, H. Mecking, Y. Nakayama, Development of shear bands in fcc single induced twins, Scr. Mater 60 (2009) 714e716.
crystals, Acta Metall. 33 (1985) 379e386. [317] C. Haase, L.A. Barrales-Mora, D.A. Molodov, G. Gottstein, Tailoring the me-
[286] C. Donadille, R. Valle, P. Dervin, R. Penelle, Development of texture and chanical properties of a twinning-induced plasticity steel by retention of
microstructure during cold-rolling and annealing of FCC alloys: example of deformation twins during heat treatment, Metall. Mater. Trans. A 44 (2013)
an austenitic stainless steel, Acta Metall. 37 (1989) 1547e1571. 4445e4449.
[287] F. Roters, P. Eisenlohr, L. Hantcherli, D.D. Tjahjanto, T.R. Bieler, D. Raabe, [318] C. Haase, L.A. Barrales-Mora, D.A. Molodov, G. Gottstein, Application of
Overview of constitutive laws, kinematics, homogenization and multiscale texture analysis for optimizing thermo-mechanical treatment of a high Mn
methods in crystal plasticity finite-element modeling: theory, experiments, TWIP steel, Adv. Mat. Res. (2014) 213e218.
applications, Acta Mater 58 (2010) 1152e1211. [319] C. Haase, T. Ingendahl, O. Güvenç, M. Bambach, W. Bleck, D.A. Molodov,
[288] D. Barbier, V. Favier, B. Bolle, Modeling the deformation textures and L.A. Barrales-Mora, On the applicability of recovery-annealed twinning-
microstructural evolutions of a Fe-Mn-C TWIP steel during tensile and shear induced plasticity steels: potential and limitations, Mater. Sci. Eng. A 649
testing, Mater. Sci. Eng. A 540 (2012) 212e225. (2016) 74e84.
[289] T. Leffers, Computer simulation of the plastic deformation in face-centred [320] T. Niendorf, C. Lotze, D. Canadinc, A. Frehn, H. Maier, The role of monotonic
cubic polycrystals and the rolling texture derived, Phys. Status Solidi B 25 pre-deformation on the fatigue performance of a high-manganese austenitic
(1968) 337e344. TWIP steel, Mater. Sci. Eng. A 499 (2009) 518e524.
[290] I. Dillamore, H. Katoh, A comparison of the observed and predicted defor- [321] A. Hamada, L. Karjalainen, A. Ferraiuolo, J.G. Sevillano, F. De Las Cuevas,
mation textures in cubic metals, Met. Sci. 8 (1974) 21e27. G. Pratolongo, M. Reis, Fatigue behavior of four high-Mn twinning induced
[291] Y. Lü, D.A. Molodov, G. Gottstein, Recrystallization kinetics and microstruc- plasticity effect steels, Metall. Mater. Trans. A 41 (2010) 1102e1108.
ture evolution during annealing of a cold-rolled Fe-Mn-C alloy, Acta Mater [322] Y. Estrin, A. Vinogradov, Fatigue behaviour of light alloys with ultrafine grain
59 (2011) 3229e3243. structure produced by severe plastic deformation: an overview, Int. J. Fatigue
[292] Y. Lü, B. Hutchinson, D.A. Molodov, G. Gottstein, Effect of deformation and 32 (2010) 898e907.
annealing on the formation and reversion of ε-martensite in an Fe-Mn-C [323] A.S. Hamada, L.P. Karjalainen, J. Puustinen, Fatigue behavior of high-Mn
alloy, Acta Mater 58 (2010) 3079e3090. TWIP steels, Mater. Sci. Eng. A 517 (2009) 68e77.
[293] R. Ray, J.J. Jonas, R. Hook, Cold rolling and annealing textures in low carbon [324] L.P. Karjalainen, A. Hamada, R.D.K. Misra, D.A. Porter, Some aspects of the
and extra low carbon steels, Int. Mater. Rev. 39 (1994) 129e172. cyclic behavior of twinning-induced plasticity steels, Scr. Mater 66 (2012)
[294] L. Bracke, K. Verbeken, L. Kestens, J. Penning, Microstructure and texture 1034e1039.
evolution during cold rolling and annealing of a high Mn TWIP steel, Acta [325] T. Niendorf, F. Rubitschek, H. Maier, J. Niendorf, H. Richard, A. Frehn, Fatigue
Mater 57 (2009) 1512e1524. crack growth-Microstructure relationships in a high-manganese austenitic
[295] L. Bracke, K. Verbeken, L.A. Kestens, Texture generation and implications in TWIP steel, Mater. Sci. Eng. A 527 (2010) 2412e2417.
[326] Y.W. Kim, G. Kim, S.-G. Hong, C.S. Lee, Energy-based approach to predict the embrittlement of Fe-18Mn-0.6 C-xSi twinning-induced plasticity steels, Acta
fatigue life behavior of pre-strained Fe-18Mn TWIP steel, Mater. Sci. Eng. A Mater 103 (2016) 264e272.
528 (2011) 4696e4702. [347] J.H. Ryu, S.K. Kim, C.S. Lee, D.W. Suh, H.K.D.H. Bhadeshia, Effect of aluminium
[327] A. Hamada, A. J€ arvenpa€a
€, M. Honkanen, M. Jaskari, D. Porter, L. Karjalainen, on hydrogen-induced fracture behaviour in austenitic Fe-Mn-C steel, Proc. R.
Effects of cyclic pre-straining on mechanical properties of an austenitic Soc. A 469 (2015) 20120458.
microalloyed high-Mn twinning-induced plasticity steel, Procedia Eng. 74 [348] J. Ronevich, S. Kim, J. Speer, D. Matlock, Hydrogen effects on cathodically
(2014) 47e52. charged twinning-induced plasticity steel, Scr. Mater 66 (2012) 956e959.
[328] Y. Ha, H. Kim, K.H. Kwon, S.-G. Lee, S. Lee, N.J. Kim, Microstructural evolution [349] L. Chen, S.-J. Lee, B.C. De Cooman, Mechanical properties of H-charged Fe-
in Fe-22Mn-0.4 C twinning-induced plasticity steel during high strain rate 18Mn-1.5 Al-0.6 C TWIP steel, ISIJ Int. 52 (2012) 1670e1677.
deformation, Metall. Mater. Trans. A 46 (2015) 545e548. [350] M. Koyama, E. Akiyama, K. Tsuzaki, Effect of hydrogen content on the
[329] A. Hamada, L. Karjalainen, M. Somani, The influence of aluminum on hot embrittlement in a Fe-Mn-C twinning-induced plasticity steel, Corros. Sci. 59
deformation behavior and tensile properties of high-Mn TWIP steels, Mater. (2012) 277e281.
Sci. Eng. A 467 (2007) 114e124. [351] M. Koyama, E. Akiyama, T. Sawaguchi, D. Raabe, K. Tsuzaki, Hydrogen-
[330] Y. Tomota, M. Strum, J. Morris, The relationship between toughness and induced cracking at grain and twin boundaries in an Fe-Mn-C austenitic
microstructure in Fe-high Mn binary alloys, Metall. Trans. A 18 (1991) steel, Scr. Mater 66 (2012) 459e462.
1073e1081. [352] M. Koyama, E. Akiyama, K. Tsuzaki, Hydrogen-induced delayed fracture of a
[331] G. Gigacher, W. Krieger, P.R. Scheller, C. Thomser, Non-metallic inclusions in Fe-22Mn-0.6C steel pre-strained at different strain rates, Scr. Mater 66
high-manganese-alloy steels, Steel Res. Int. 76 (2005) 644e649. (2012) 947e950.
[332] M. Whiteman, A. Troiano, The influence of hydrogen on the stacking fault [353] M. Koyama, E. Akiyama, K. Tsuzaki, D. Raabe, Hydrogen-assisted failure in a
energy of an austenitic stainless steel, Phys. Status Solidi B 7 (1964). twinning-induced plasticity steel studied under in situ hydrogen charging by
[333] A.E. Pontini, J.D. Hermida, X-ray diffraction measurement of the stacking electron channeling contrast imaging, Acta Mater 61 (2013) 4607e4618.
fault energy reduction induced by hydrogen in an AISI 304 steel, Scr. Mater [354] R. Van Tol, L. Zhao, H. Schut, J. Sietsma, Experimental investigation of
37 (1997) 1831e1837. structural defects in deep-drawn austenitic Mn-based TWIP steel, Mater. Sci.
[334] P. Rozenak, D. Eliezer, Phase changes related to hydrogen-induced cracking Tech. 28 (2012) 348e353.
in austenitic stainless steel, Acta Metall. 35 (1987) 2329e2340. [355] C. Beal, X. Kleber, D. Fabregue, M. Bouzekri, Embrittlement of a zinc coated
[335] N. Narita, C. Altstetter, H. Birnbaum, Hydrogen-related phase trans- high manganese TWIP steel, Mater. Sci. Eng. A 543 (2012) 76e83.
formations in austenitic stainless steels, Metall. Mater. Trans. A 13 (1982) [356] D. Fabrege, C. Be al, X. Kleber, M. Bouzekri, Proc. 1st Int. Conf. on High Mn
1355e1365. Steels, Seoul, 2011, B-44.
[336] J. Rigsbee, R. Benson, A TEM investigation of hydrogen-induced deformation [357] C. Beal, Mechanical Behaviour of a New Automotive High Manganese TWIP
twinning and associated martensitic phases in 304-type stainless steel, Steel in the Presence of Liquid Zinc, INSA, Lyon, 2011.
J. Mater. Sci. 12 (1977) 406e409. [358] H. Kang, L. Cho, C. Lee, B.C. De Cooman, Zn penetration in liquid metal
[337] E. Astafurova, G. Zakharova, H. Maier, Hydrogen-induced twinning in< 001> embrittled TWIP steel, Metall. Mater. Trans. A 47 (2016) 2885e2905.
Hadfield steel single crystals, Scr. Mater 63 (2010) 1189e1192. [359] M. Krishtal, the formation of dislocations in metals on diffusion of surface-
[338] B.J. Lee, J.W. Jang, A modified embedded-atom method interatomic potential active substances in connection with the effect of adsorption embrittle-
for the Fe-H system, Acta Mater 55 (2007) 6779e6788. ment, Sov. Phys. Dokl. (1970) 614.
[339] J. Jung, O. Lee, Y. Park, D. Kim, K. Jin, S. Kim, K. Song, Microstructure and [360] P. Gordon, H.H. An, The mechanisms of crack initiation and crack propaga-
mechanical properties of high Mn TWIP steels, Korean J. Met. Mater 46 tion in metal-induced embrittlement of metals, Metall. Trans. A 13 (1982)
(2008) 627e633. 457e472.
[340] K.H. So, J.S. Kim, Y.S. Chun, K.-T. Park, Y.-K. Lee, C.S. Lee, Hydrogen delayed [361] N. Stoloff, T. Johnston, Crack propagation in a liquid metal environment, Acta
fracture properties and internal hydrogen behavior of a Fe-18Mn-1.5 Al-0.6 Metall. 11 (1963) 251e256.
C TWIP steel, ISIJ Int. 49 (2009) 1952e1959. [362] A. Westwood, M. Kamdar, Concerning liquid metal embrittlement, particu-
[341] Y.S. Chun, J.S. Kim, K.T. Park, Y.K. Lee, C.S. Lee, Role of ε martensite in tensile larly of zinc monocrystals by mercury, Philos. Mag. 8 (1963) 787e804.
properties and hydrogen degradation of high-Mn steels, Mater. Sci. Eng. A [363] C. Scott, S. Allain, M. Faral, N. Guelton, The development of a new Fe-Mn-C
533 (2012) 87e95. austenitic steel for automotive applications, Rev. Metall. 103 (2006)
[342] Y.S. Chun, J. Lee, C.M. Bae, K.-T. Park, C.S. Lee, Caliber-rolled TWIP steel for 293e302.
high-strength wire rods with enhanced hydrogen-delayed fracture resis- [364] W. Bleck, K. Phiu-on, C. Heering, G. Hirt, Hot workability of as-cast high
tance, Scr. Mater 67 (2012) 681e684. manganese-high carbon steels, Steel Res. Int 78 (2007) 536e545.
[343] Y.S. Chun, K.-T. Park, C.S. Lee, Delayed static failure of twinning-induced [365] P. Lan, J. Zhang, Thermophysical properties and solidification defects of Fe-
plasticity steels, Scr. Mater 66 (2012) 960e965. 22Mn-0.7C TWIP Steel, Steel Res. Int. 87 (2016) 250e261.
[344] I.-J. Park, K.-H. Jeong, J.-G. Jung, C.S. Lee, Y.-K. Lee, The mechanism of [366] K.H. Spitzer, F. Rüppel, R. Vis
corov
a, R. Scholz, J. Kroos, V. Flaxa, Direct strip
enhanced resistance to the hydrogen delayed fracture in Al-added Fe-18Mn- casting (DSC)-an option for the production of new steel grades, Steel Res. Int.
0.6 C twinning-induced plasticity steels, Int. J. Hydrogen Energy 37 (2012) 74 (2003) 724e731.
9925e9932. [367] H. Hoffmann, B. Engl, M. Menne, T. Heller, W. Zimmermann, Highly stable,
[345] Y.J. Kwon, T. Lee, J. Lee, Y.S. Chun, C.S. Lee, Role of Cu on hydrogen embrit- strips or steel sheets cold-formed, method for the production of steel strips
tlement behavior in Fe-Mn-C-Cu TWIP steel, Int. J. Hydrogen Energy 40 and uses of said steel, US Patent US 20030145911 A1 (2003).
(2015) 7409e7419. [368] T.W. Kim, Y.G. Kim, S.H. Park, Process for manufacturing high manganese hot
[346] S.-M. Lee, I.-J. Park, J.-G. Jung, Y.-K. Lee, The effect of Si on hydrogen rolled steel sheet without any crack, US Patent US 5647922 A (1997).

Twinning-Induced Plasticity (TWIP) Steels

Uploaded by

Copyright:

Available Formats

Twinning-Induced Plasticity (TWIP) Steels

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Twinning-Induced Plasticity (TWIP) Steels

Uploaded by

Copyright:

Available Formats

Acta Materialia 142 (2018) 283e362

Contents lists available at ScienceDirect

By invitation only: overview article

Twinning-induced plasticity (TWIP) steels

3.3. Effect of alloying elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312

1. Introduction principle calculations. The experimental analysis of the properties

Strain hardening, MPa

Strain at onset of twinning

800 -0.3 (111)[011] spacing

Local viscoplastic Local viscoplastic Local viscoplastic Local viscoplastic

(a) (b) (c)

Strain hardening (MPa)

True stress (MPa)

Fe-18%Mn-0.5%C-1.5%Al 800 Fe-22%Mn-0.6%C

E=180GPa XRD data

(a) True strain (b) True strain

ð1 FT ðεÞÞsi;M þ FT ðεÞsi;T ¼ 0 (16)

Engineering stress, MPa

Engineering strain, % Temperature, ºC Temperature, ºC

3500 z: 0.116, TOI:0.33

3000 z: 0.008, TOI:0.03

0.0020 Fe-18%Mn-0.6%C 0.0020 Fe-18%Mn-0.6%C-1.5%Al

0.0015 Internal Friction, Q-1 0.0015

0 100 200 300 400 500 600

between dislocations and deformation twin boundaries in TWIP

3. The energy of planar faults in a TWIP steel

3.1. Intrinsic vs. extrinsic stacking faults

There are relatively few publications which have addressed the

(a) (b) (c)

CTB (K1 /conjugate plane)

Primary slip (K2) plane

No slip transmission Twin Matrix

3.2. Stacking fault energy computations

Classical texts on the mechanical behavior of fcc metals and

Twinning Dislocation cells

Twinning Dislocation cells

0.3610 0.3660 -Fe-xMn

Olson-Cohen Spherical [102] g/ε 75n 2 fcc 2 !2

(1.6e4.1)%Al-0.08%C TWIP steel. Nakano and Jacques [128] pro-

15 mJ/m2 isf : 13 mJ/m

Allain et al. [71] Fe-Mn-C Not Included Not Included 9 SGTE

G 0 determination of these energy barriers is not possible and ab initio

between speciﬁc shear displacements of a rigid lattice and the

30 outcomes of these calculations, which will be discussed in the next

It should be noted that conceptually, twinning should be seen as

utf utf utf

mation, are equally likely to occur. As the universal relation vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

Screw orientation Edge orientation

1=-30º 2=+30º =0º 1=0º 2=+60º =+30º 1=+60º 2=+120º =+90º

Twinnability Parameter Parameter Value Reference

dut=us ¼ gutf gusf dut/us ¼ 20 mJ/m2 > 0 Tadmor-Hai [139]

Asaro-Suresh parameter TAS.

usf utf utf

Lattice shear Lattice shear

2500 isf max -Fe 2500 Fe24C supercell

1000 diffused 1000 diffused

3.3.2. Manganese faults during dislocation glide [193].

(a) Temperature, ºC (b) Temperature, ºC

GVoigt þ GReuss C þ 2C12

tT $b112 gutf (70)

5 4 5.3. Twinning kinetics and twinning saturation