In-situ synchrotron radiation studies of the deformation

mechanisms in metastable austenitic TRIP/TWIP steels

C Ullrich1, S. Martin1, C Schimpf1, A Stark2, N Schell3 and D Rafaja1

1 Institute of Materials Science, TU Bergakademie Freiberg, Gustav-Zeuner-Str. 5,

09599 Freiberg, Germany 2 Institute of Materials Research, Helmholtz-Zentrum Geesthacht, Max-Planck-Str. 1,

21502 Geesthacht, Germany 3 Structural Research on New Materials, Helmholtz-Zentrum Geesthacht Outstation at

DESY, Hamburg, Germany

E-mail: Christiane.Ullrich@ww.tu-freiberg.de

Abstract. Many deformation mechanisms in metastable austenitic steels are governed by the

stacking fault energy. In this study, microstructure changes in Cr-Mn-Ni steels with different

Ni contents (3, 6 and 9 wt.%) and thus with different stacking fault energies (SFE = 7.5, 16.7

and 24.3 mJ/m²) are investigated using in-situ synchrotron X-ray diffraction under uniaxial

compression. The martensitic phase transformation is tracked by the evolution of the phase

composition. The changes in the defect densities (microstrain due to dislocations, stacking fault

probabilities) are monitored and correlated with the lattice strain in austenite. The onset of -

martensite formation was found to be coupled to an abundant stacking fault formation and to

their dense arrangement. The ´-martensite forms mainly in the steels with low SFE; it was not

found in the steel with SFE > 20 mJ/m². Increasing SFE reduces the stacking fault probability

and shifts the onset of the stacking fault formation to higher deformations. Concurrently,

higher SFE enhances the formation of dislocations and their slip activity. The interplay of the

microstructure features and defects and their effect on the stress-strain curve are discussed.

1. Introduction

High-alloy metastable austenitic steels, which show the TRIP (Transformation Induced Plasticity) or

TWIP (Twinning Induced Plasticity) effect, exhibit outstanding properties, especially with regard to

the strength, ductility and absorption of mechanical energy during plastic deformation [1-3]. Their

deformation behavior is largely controlled by the stacking fault energy (SFE) of austenite, which

depends on the chemical composition of the steel and on the deformation temperature [4-7]. For very

low SFE (< 20 mJ/m²), the dissociation of partial dislocations, the formation of stacking faults and

their bunching in deformation bands are observed. A high stacking fault density leads to a local

arrangement of stacking faults in the deformation bands. This arrangement resembles a hexagonal

crystal structure, which is identified as -martensite by diffraction methods [8-10]. At higher

deformations, ´-martensite forms – preferentially inside of the deformation bands or at their

intersections [9-11]. Increasing SFE postpones the stacking fault formation and the formation of ´-

martensite to higher deformation stages. For SFE above 20 mJ/m², the formation of -martensite is

gradually replaced by twinning. The twins are also produced via dissociation of partial dislocations

and subsequent accumulation of stacking faults [8, 12-15], but the stacking fault density is higher in

twinned austenite, where the stacking faults appear on every close-packed plane, than in -martensite

accommodating the stacking faults on every second close-packed plane. For SFE > 40 mJ/m², the

deformation is mostly accomplished through the slip of perfect dislocations, because the faulting and

the martensitic transformation are suppressed and postponed in this SFE range [6].

The aim of the present study is to analyze the microstructure evolution, the corresponding

deformation mechanisms and the interaction between different types of microstructure defects and the

martensitic phase transformation plus their effects on the mechanical behavior during deformation of

three metastable austenitic Cr-Mn-Ni-steels with varying compositions and SFEs.

2. Experimental

Three fine-grained metastable austenitic 16Cr-7Mn-xNi steels were studied. Their composition varied

in the Ni content, which amounted to 3, 6 and 9 wt.%. The SFE of these steels is known from in-situ

XRD bending tests to be 7.5, 16.7 and 24.3 mJ/m², respectively [16, 17]. The samples with 3 and 9

wt.% Ni were produced by powder metallurgy using field assisted sintering technique, whereas the

16Cr-7Mn-6Ni sample was taken from a hot-rolled bar. In-situ compression tests at room temperature

were done using a deformation dilatometer DIL 805A/D (TA Instruments, formerly Bähr), which is

installed at the beamline P07, PETRA III at DESY synchrotron radiation facility. The cylindrical

compression samples had a diameter of 4 mm, therefore the maximum stress applied with the 20 kN

maximum load device was 1590 MPa. The samples were compressed stepwise with a short holding (5

s) needed for recording of the diffraction patterns. The high-energy synchrotron radiation (100 keV,

= 0.1235 Å) allowed to conduct the diffraction experiments in transmission geometry. The patterns

were recorded using a 2D detector (PerkinElmer). In this setup, the angle between the diffraction

vector and the uniaxial load direction changes with the azimuthal angle (along the Debye rings)

between a half of the diffraction angle and 90°, thus the dependence of the line positions on the

azimuthal angle can be used for the analysis of the lattice deformation upon deformation. The electron

channeling contrast imaging (ECCI) and electron back-scatter diffraction (EBSD) on deformed

samples were done post mortem in a scanning electron microscope Zeiss LEO 1530 FEGSEM at 20

kV acceleration voltage using the HKL Channel 5 EBSD software and a Nordlys detector.

Figure 1. True stress-true strain curves

measured for the metastable austenitic steel

samples under study.

Figure 2. Diffraction patterns of sample 16Cr-

7Mn-6Ni after 21% compression obtained from the

integration of 5° sections in a single 2D diffraction

pattern. The angles in the legend refer to the

azimuthal position of the respective section with

respect to the load direction. Measured intensities

are plotted by dots, refined intensities by lines. The

peaks are labelled by diffraction indices.

3. Data analysis

The 2D diffraction images were integrated within azimuthal sections having the width of 5°, and

analyzed using the Rietveld refinement with the software MAUD [18]. The Rietveld analysis revealed

the phase fractions of fcc austenite (space group 𝐹𝑚3̅𝑚), hcp -martensite (𝑃63/𝑚𝑚𝑐) and bcc ´-martensite (𝐼𝑚3̅𝑚), and the macroscopic lattice deformation, the stacking fault probability and the microstrain in austenite. The austenite lattice parameters for the undeformed steels were determined to

be 3.5954(7) Å for 16Cr-7Mn-3Ni, 3.5930(6) Å for 16Cr-7Mn-6Ni and 3.5923(7) Å for 16Cr-7Mn-9Ni, respectively. The macroscopic lattice deformation was described using the Moment Pole Stress

Model [19], which includes the anisotropy of the elastic constants and allows therefore the description

of the anisotropic line shift due to strain. Elastic constants of austenite were taken from Refs. [20] and

[21]. For the description of the anisotropic line broadening, the Popa model [22] was applied. It was

assumed that the refined microstrain, which corresponds to the variation of lattice plane distances, is

caused by dislocations and thus proportional to the square root of the dislocation density [23, 24]. The

stacking fault probability was concluded from the line shift and from the line broadening according to

the Warren model [25]. Preferred orientation of crystallites during deformation was described by

orientation distribution functions, which were refined using the E-WIMV method that is incorporated

in the MAUD routine. For the description of the texture, a model developed by Wenk et al. [26]

assuming a fiber texture was employed.

4. Results and discussion

The true stress-strain curves obtained from mechanical experiments (figure 1) show differently

pronounced strain hardening in the austenitic steels with different chemical compositions. The lowest

stress level was observed in the steel with 9 wt.% Ni, the highest stress level in the steels with less Ni

and thus with lower SFE. The stress-strain curve of the steel 16Cr-7Mn-3Ni has a distinct sigmoidal

shape, indicating the formation of a high amount of bcc ’-martensite. The steel 16Cr-7Mn-6Ni

possesses the highest yield strength, probably due to an incompletely recrystallized microstructure. As

the experiments were conducted in force-controlled mode, the maximum technical stress was equal for

all samples, but the deformations achieved at the respective stress level differ. The smallest sample

deformation was observed in sample 16Cr-7Mn-3Ni, the largest one in 16Cr-7Mn-9Ni. This strength

and hardening behavior can be expected, because a lower SFE and a larger extent of martensitic

transformation in steels with a lower Ni content increase the stress level and reduce the deformability

[1, 2]. Still, the mechanical behaviors of samples 16Cr-7Mn-3Ni and 16Cr-7Mn-6Ni differ less than

expected. The possible reasons for this similarity will be discussed in detail later, as they are also

reflected in the results of in situ synchrotron radiation diffraction.

Figure 3. Evolution of the phase fractions as obtained from the Rietveld refinement. a) all phases,

b) -martensite.

Figure 2 shows exemplarily the diffraction patterns that were extracted from the azimuthal sections

of the 2D scans recorded for the sample 16Cr-7Mn-6Ni at 21 % compression together with the

respective Rietveld fits. The evolution of the phase fractions under compressive load, which was

obtained from the Rietveld refinement, is summarized in figure 3. The initial microstructures of all

three steels were fully austenitic. In the steels with 3 and 6 wt.% Ni, the formation of hcp -martensite

starts very soon after the onset of the plastic deformation. The diffraction lines from -martensite were

detected already at 0.8 % compression for both compositions. The maximum amount of -martensite

(19 vol.%) was attained at ca. 17 % compression. Beyond this maximum, a slight reduction of the -

martensite fraction was observed for both steels, as -martensite transforms partially into ´-

martensite. The formation of the bcc ´-martensite started between 2 % and 3 % compression. The

maximum content of ´-martensite amounted 40 vol.% for 16Cr-7Mn-3Ni (at 21 % compression) and

32 vol.% for 16Cr-7Mn-6Ni (at 23 % compression). The 16Cr-7Mn-9Ni steel revealed the first -

martensite peaks at 5 % compression, the maximum fraction of 12 vol.% was reached at the maximum

compression of 33 %. In this sample, the bcc ´-martensite did not form in the investigated

deformation range, as the high Ni content stabilizes austenite, increases its SFE and lowers the driving

force for the (’) martensitic transformation.

The effect of the Ni content on the SFE can also be seen on different stacking fault probabilities

(SFP) in individual samples (figure 4). In sample 16Cr-7Mn-3Ni, an enormous increase of SFP was

observed beyond 0.7 % compression. This means that the plastic deformation of this steel is

dominated by the stacking fault formation. For 16Cr-7Mn-6Ni, a similar course is observed, with a

comparable, i.e., early start of the SFP increase. The final values of SFP are even higher for 16Cr-

7Mn-6Ni than for 16Cr-7Mn-3Ni. This behavior is probably again due to the incompletely

recrystallized initial microstructure, which provides more nucleation sites for stacking fault formation

and also creates higher levels of internal stress, which is necessary to dissociate partial dislocations

[16, 17]. Additionally, a higher SFP was observed in 16Cr-7Mn-6Ni than in 16Cr-7Mn-3Ni already in

the non-deformed state. In steel 16Cr-7Mn-9Ni, almost no stacking faults are present prior to the 3 %

compression. The deformation in this stage runs exclusively via slip of perfect dislocations. The

critical stress for wide dissociation of partial dislocations is reached first at higher deformations (>

3%), after a certain amount of work hardening due to the dislocations is achieved. As the microstrain

in sample 16Cr-7Mn-9Ni is relatively low (figure 5), these dislocations form probably dislocation

structures with partially compensated strain fields [23]. Although the plastic deformation of low-SFE

alloys is dominated by stacking faults, at least initial plastic deformation by perfect dislocation slip is

generally necessary for the initiation of the stacking fault formation.

Figure 4. Stacking fault probability for isolated

intrinsic stacking faults in austenite obtained by

using the Warren model [25].

Figure 5. Evolution of the squared microstrain

〈ε1002 〉 in austenite.

The squared microstrain plotted in figure 5 is as a measure of the dislocation density [23].

However, as already mentioned above, the microstrain values are possibly affected by the correlations

in the dislocation positions. For the steel with the lowest SFE (16Cr-7Mn-3Ni), an initial increase in

squared microstrain up to ca. 7 % compression is observed. Afterwards the increase is lowered, and

the microstrain almost stagnates between 12 and 16 % compression. This indicates that the dislocation

slip is a less important mechanism of the plastic deformation in this compression range, as the

formation of stacking faults is dominant (cf. figure 4). For 16Cr-7Mn-6Ni, the microstrain increases

during the whole deformation process. High initial microstrain indicates again that the sample was not

completely recrystallized, as it had a high dislocation density in the non-deformed state. The high

dislocation density in the starting state may also account for the early onset of the martensitic

transformation that occurs at a similar deformation like in the steel 16Cr-7Mn-3Ni (figure 3) despite

its lower SFE. Nevertheless, the microstrain in 16Cr-7Mn-6Ni attains a clearly higher level than in

sample 16Cr-7Mn-3Ni, and shows a constant increase, what indicates an enhanced dislocation slip

activity. The steel containing 9 wt.% Ni has nearly the same initial microstrain as steel 16Cr-7Mn-3Ni,

and shows a permanent increase of the microstrain. At highest deformations, the maximum

microstrain in sample 16Cr-7Mn-9Ni is comparable with the maximum microstrain in sample 16Cr-

7Mn-6Ni. High SFE in sample 16Cr-7Mn-9Ni retards the stacking fault formation and promotes the

dislocation slip activity of undissociated perfect dislocations instead. New dislocations are formed

during the whole deformation process, even in the deformation range, in which the stacking faults

form. By comparing the microstrain evolution in these three steels, it is obvious that with decreasing

SFE, the portion of dislocation slip for plastic deformation decreases, because other deformation

mechanisms such as stacking faults and phase transformation excel them.

Other outputs of the in situ diffraction measurement under deformation were the lattice stress in

austenite and its dependence on the macroscopic deformation (figure 6a) as a complement to the

stress-strain curve from figure 1. The lattice stress was obtained from the anisotropic shift of the

diffraction lines measured at different angles between the diffraction vector and the compression

direction [26, 27]. In the elastic range (approx. below 0.5 % compression), the mechanical stress and

the lattice stress in austenite agree quite well. After the plastic deformation starts, the mechanical

stress becomes higher than the lattice stress in austenite. Whereas the mechanical stress is a sum of

stresses needed for both, elastic and plastic deformations of all phases in the sample, the lattice stress

comprises only the elastic component in the austenite. Thus, the difference between both stresses

(figure 6b) is a measure of the driving force for the plastic deformation. The stress difference increases

Figure 6. (a) Lattice stress in

austenite (in the compression

direction) obtained from the shift of

the diffraction lines and (b) the

difference between mechanical true

stress (cf. fig. 1) and the lattice

stress.

during the compression, which confirms the expected strain hardening of the samples due to the

increasing density of microstructure defects, intense interaction between their strain fields and

deformation induced phase transformations.

The extents of individual hardening mechanisms follow from the comparison of the stress

difference (figure 6b) with the phase fractions (figure 3), stacking fault probability (figure 4) and

squared microstrain (figure 5). The steep increase of the stress difference that was observed in samples

16Cr-7Mn-3Ni and 16Cr-7Mn-6Ni at low compressions is mainly caused by the formation of stacking

faults and -martensite. As -martensite originates from faulted austenite with dense packed stacking

faults [8], the formation of stacking faults and -martensite are coupled phenomena that have a similar

effect on the stress-strain curve. The main contribution to the increase of the stress difference and thus

to the strain hardening in sample 16Cr-7Mn-3Ni at the deformations > 9 % comes from the formation

of ´-martensite. The formation of dislocations plays a minor role. This model is also applicable for

sample 16Cr-7Mn-9Ni, in which the increase of the stress difference is delayed in the same manner as

the formation of the stacking faults and -martensite.

In steel 16Cr-7Mn-9Ni, the stacking fault probabilities are much lower than in steels with lower Ni

contents, despite the low yield strength. As the plasticity of the steel with 9 wt.% Ni can be explained

neither by high dislocation density nor by formation of random stacking faults nor by the -martensite

formation, the only possible deformation mechanisms are the absence of ’-martensite and the

presence of twins. The presence of twins affects both, the broadening and asymmetry of X-ray

diffraction lines, but the X-ray diffraction is generally less sensitive to the presence of twins than to

the presence of stacking faults and dislocations. Therefore, the existence of twins in deformed samples

was examined by ECCI and EBSD in SEM.

Representative ECCI microstructure images are shown in Figure 7. For the 16Cr-7Mn-3Ni,

deformation bands generally occur on several slip planes, providing nucleation sites for the ´-

a) 16Cr-7Mn-3Ni

b) 16Cr-7Mn-6Ni

c) 16Cr-7Mn-9Ni

Figure 7. ECCI micrographs of the samples after

deformation. a) Deformation bands (marked by

straight lines), stacking faults (arrows) and ´-

martensite (circles) in 16Cr-7Mn-3Ni. b) Very

high defect density with dislocations (blue

arrows), dense deformation bands (lines) and ´-

martensite (circles) in 16Cr-7Mn-6Ni. c)

Dislocation structures (blue arrows) and

deformation bands (lines) identified as -

martensite and twins in 16Cr-7Mn-9Ni.

martensite formation. Additionally, nucleation sites can also be found inside deformation bands

consisting of -martensite. In the remaining austenite matrix, a high concentration of stacking faults

can be detected, but the dislocation slip and accordingly the formation of dislocation structures is less

noticeable. The microstructure of the steel 16Cr-7Mn-6Ni also contains a high stacking fault density in

austenite, many deformation bands on several slip systems and nuclei of ´-martensite. Additionally, the pronounced contrasts from dislocations evidence a highly partitioned microstructure and verify the

high dislocation density found by the XRD analysis. The ECCI image of the steel 16Cr-7Mn-9Ni

clearly reveals the dominant role of the dislocation slip in the deformation of this sample. Some areas

are free of typical deformation bands, while tangled dislocations and dislocation walls are visible. The

EBSD measurements revealed considerable misorientations inside of grains, which can be interpreted

as local lattice rotations due to presence of dislocation structures. The interaction between deformation

bands and dislocations results in a curvature of the deformation bands. Furthermore, the results of

EBSD demonstrate that the deformation bands consist partly of twins and partly of -martensite. The

´-martensite was not found using EBSD, which is also in a good agreement with the result of the XRD phase analysis.

5. Conclusions

The microstructure evolution in metastable austenitic steels with different stacking fault energies was

studied using in-situ synchrotron XRD under compressive load. The formation of dislocations,

stacking faults and twins, and the martensitic transformation of austenite to -martensite and ´-

martensite were correlated with the stress-strain curves obtained from the mechanical experiments and

with the elastic lattice deformation obtained from the diffraction experiments. In all steels under study,

the plastic deformation starts with the production of perfect dislocations. The significance of

individual deformation mechanisms at higher deformations depends strongly on the stacking fault

energy, which controls the onset of the rapid dissociation of partial dislocations and the stacking fault

formation. Especially in steels with higher stacking fault energies, some critical stress must be reached

to be able to activate the partial dislocation dissociation and the stacking fault formation. The critical

stress can be produced by interacting microstructure defects, in particular by interaction of perfect

dislocations. With increasing Ni content and stacking fault energy, the stacking fault formation and the

martensitic phase transformations are retarded, the formation of ´-martensite is suppressed, and the importance of the dislocation slip increases. A higher initial defect density, as present in sample 16Cr-

7Mn-6Ni, brings about higher defect densities (stacking faults and dislocations) and an earlier onset of

stacking fault formation and phase transformation during deformation, mainly due to a higher level of

lattice stress in the austenite.

Acknowledgements

This work was funded by the German Research Foundation (DFG) as part of a research project within

the Collaborative Research Centre SFB 799. We would like to thank Dr. S. Decker for sintering

samples by field assisted sintering technology and Mr. R. Prang for the SEM sample preparation.

References

[1] Jahn A, Kovalev A, Weiß A, Wolf S, Krüger L and Scheller P R 2011 steel res.int. 82 39 [2] Krüger L, Wolf S, Martin U, Martin S, Scheller P R, Jahn A and Weiß A 2010 15th

International Conference on the Strength of Materials (ICSMA-15) J. Phys., 240 1

[3] Frommeyer G, Brüx U and Neumann P 2003 ISIJ Int. 43 438 [4] Rémy L 1977 Acta metal. 25 173 [5] Rémy L 1977 Metall. Trans. A 8A 253 [6] Martin S, Wolf S, Martin U, Krüger L and Rafaja D 2016 Metall. Mater. Trans. A A47 49 [7] Martin S, Fabrichnaya O and Rafaja D 2015 Mater. Lett. 159 484

[8] Martin S, Ullrich C, Simek D, Martin U and Rafaja D 2011 J. Appl. Cryst. 44 779 [9] Martin S, Ullrich C and Rafaja D 2015 Mater. Today: Proceedings 2S 643 [10] Bracke L, Kestens L and Penning J 2007 Scripta Mater. 57 385 [11] Olson G B and Cohen M 1972 J. Less Common Metals 28 107 [12] Ullrich C, Eckner R, Krüger L, Martin S, Klemm V and Rafaja D 2016 Mater. Sci. Eng. A 649

390

[13] Shen Y, Li X, Sun Y, Wang L and Zuo L 2012 Mater. Sci. Eng. A 552 514 [14] Hamada A S, Karjalainen L P, Misra R D K and Talonen J 2013 Mater. Sci. Eng. A 559 336 [15] Bracke L, Kestens L and Penning J 2009 Scripta Mater. 61 220 [16] Rafaja D, Krbetschek C, Borisova D, Schreiber G and Klemm V 2013 Thin Solid Films 530 105 [17] Rafaja D, Krbetschek C, Ullrich C and Martin S 2014 J. Appl. Cryst. 47 936 [18] Lutterotti L, Matthies S and Wenk H R 1999 IUCr Commission on Powder

DiffractionNewsletter 21 14

[19] Matthies S and Vinel G W 1982 phys. stat. solidi 112 111 [20] Pierce D T, Nowag K, Montagne A, Jiménez J A, Wittig J E and Ghisleni R 2013 Mater. Sci.

Eng. A 578 134

[21] Ledbetter H M 1984 phys. stat. solidi A 86 89 [22] Popa N C 1998 J. Appl. Cryst. 31 176 [23] Wilkens M 1970 phys. stat. solidi A 2 359 [24] Ungár T and Borbély A 1996 Appl. Phys. Lett. 69 3173 [25] Warren B E 1969 X-Ray Diffraction New York: Dover [26] Wenk H R and Grigull S 2003 J. Appl. Cryst. 36 1040 [27] Merkel S, Liermann H P, Miyagi L and Wenk H R 2013 Acta mater.61 5144