15. Scanning Electron Microscopy and Complementary In Situ Characterization Techniques for Characterization of Deformation and Damage Processes

With the possibility to perform in situ deformation tests in the SEM it was just a short step to combine these deformation experiments with digital image correlation. The advantage of deformation experiments within the SEM is the determination of deformation fields with a high lateral resolution and the combination with analytical characterization methods such as EBSD. In order to achieve the highest possible resolution of DIC, it is necessary to generate high-resolution micrographs. Therefore, it is essential to set optimal beam conditions to minimize image distortion, drift, charging and other artifacts [29]. Probably one of the most challenging tasks for the application of DIC on SEM micrographs is a good contrast pattern at the specimen surface without influencing the image conditions. Moreover, the size of the individual features of the pattern should have dimensions well below the features, which will be studied by DIC. In multiphase microstructures such as dual phase steels, the inherent microstructure can already provide a suitable contrast [30]. The only prerequisite in this case would be small size and a fine distribution of the phase components. Short-term etching and the different etching attack on the individual phases generate a topography on the surface which leads to a sufficient contrast for SEM micrographs using secondary electrons. The resolution of this method is determined by the grain size of the phase components. Single phase materials can also be contrasted by etching of the surface where the resulting etch pits are used for the correlation [31]. Another method to contrast surface structures are particle deposition methods using e.g. SiO₂ particles (colloidal silica). The advantage of this method lies in the large-area contrast of the surface, but particle agglomerations can be challenging. In order to achieve a high resolution of the DIC it is necessary to use as small particles as possible for the contrast. The smaller the particles are the higher the resolution of the DIC can be. Yan et al. [26] investigated the damage behavior of a dual phase (DP) steel under tensile load and used for describing the quality/resolution of DIC measurement the product of subset size in nm and subset size in px. In their work they achieved a subset size of 100 nm and 17 px, respectively, by deposition of colloidal silica. Gioacchino and Fonseca [32] investigated the deformation behavior of the steel AISI 304L under uniaxial load. The contrast patterns were achieved by the deposition of gold nanoparticles which result in a subset size of 216 × 216 nm² and 6 × 6 px, respectively. Due to the small subset size, occurring deformation bands with sub-µm resolution were recorded. Wang et al. [33] combined EBSD and DIC to investigate the deformation and damage behavior of a low-alloy Q&P steel showing initiation of micro-cracks in regions of tempered martensite where strain was accommodated. Similar observations were made by Ghadbeigi et al. [34] on a commercial dual phase steel (DP 600) observing a heterogeneous strain distribution and localization within large ferritic grains. Na et al. [35] studied the behavior of a low carbon martensitic steel under tensile load. Using complementary EBSD measurements it was shown that the shear distribution in martensite correlates with the sub-cell formation and contributes, thus, to grain refinement. Weidner et al. [31] studied the deformation behavior of a high-alloy CrMnNi cast TRIP steel under tensile load. A short time etching was used for contrasting the surface which also resulted in a sub-µm resolution.

Infrared thermography is an optical contactless temperature measurement method. Each body with a temperature above absolute zero emits electromagnetic radiation with a wavelength distribution in the range of 0.8 µm up to 14 µm depending on the temperature, which is described by Planck’s radiation law [36]. The surface temperature of the body can be determined, therefore, by the detection of this infrared radiation. All plastic deformation processes are associated with the emission of energy and lead, thus, directly to an increase in temperature [37‐39]. Therefore, IR thermography is suitable as a real-time in situ method to characterize temperature increase occurring during plastic deformation, strain localization and/or phase transformations [40].

Saeed-Akbari et al. [41] investigated the deformation behavior of high manganese TWIP steels showing a strong Portevin Le Chatelier (PLC) effect during deformation at varying strain rates using infrared thermography. It was shown that the increase in stacking fault energy due to the adiabatic heating caused by the formation of the PLC band is dependent on the chemical composition. Bodelot et al. [42] performed so-called fully-coupled full-field measurements by a combination of thermography and digital image correlation on an austenitic stainless steel under tensile loading. Due to the high resolution obtained, a correlation between dissipated energy in form of increased temperature and local plastic deformation on the grain scale was achieved. Chen et al. [43] also combined DIC and IR thermography during hole expansion test of an austenitic TWIP and an interstitial free (IF) steel.

High-alloy CrMnNi cast TRIP/TWIP steels. Miniature flat tensile specimens were manufactured out of the solution annealed cast plates. The gauge length was 10 mm with a rectangular cross section of 2 × 4 mm². The specimens were carefully grinded and polished to obtain a deformation-free surface. Subsequently, the specimens were etched with a V2A agent (200 ml distilled water, 200 ml 32 pct hydrochloric acid, 20 ml 65 pct azotic acid, 0.6 ml Vogel’s special reagent) for 1 min at 60 °C providing an excellent surface contrast pattern required for DIC calculations. Tensile tests were performed both at room temperature (RT) and elevated temperatures (373 K, 473 K) using a miniature push-pull loading stage (Kammrath & Weiss, Dortmund, Germany) placed in the chamber of a high-resolution scanning electron microscope. The higher temperatures were realized by a heating plate mounted on the back side of the specimens. Temperature was controlled by a thermocouple and kept constant over the entire deformation test using PID control. The quasi in situ tensile deformation was performed step wise and after each deformation step of ∆l = 50 µm micrographs of predefined areas of interest (AOIs) were captured. The deformation continued until a global strain of 15% was reached. After tensile deformation, specimens were vibration polished using colloidal silica for 24 h (or multiples of) to remove etching pits on the surface. Subsequently, AOIs were measured again using electron backscatter diffraction. Recorded high-resolution SEM micrographs were evaluated using software package ARAMIS [82] to calculate strain distribution. For more experimental details see [31].

Steel with highest austenite stability (9 wt% nickel). Figure 15.2 shows the deformation of the X5CrMnNi16-6-9 steel at room temperature where a pronounced twin formation is expected according to the higher SFE (compare Table 15.1). Figure 15.2a shows the strain distribution according to the von Mises equivalent strain (ε_vM) at the beginning of deformation (4% global strain). The formation of deformation bands related to the activation of two different slip systems oriented parallel to \( \left( {111} \right) \) and \( \left( {11\bar{1}} \right) \) plane, respectively, was observed. The strain is well localized and homogeneously distributed within these bands (about ε_vM = 25%). With ongoing deformation, both the number of deformation bands as well as their thickness increase, in particular for the system belonging to \( \left( {11\bar{1}} \right) \) plane, which corresponds to the secondary slip system (µ = 0.36). Moreover, the strain localization within these bands increases continuously (see Fig. 15.2b). Figure 15.2c shows the calculated magnitude of shear for the secondary slip system in the direction of the Burgers vector of the Shockley partial dislocations with the highest Schmid factor (\( \vec{b} = \frac{a}{6}\left[ {1\bar{2}\bar{1}} \right] \), µ = 0.34). It turned out that not all deformation bands belonging to secondary slip system exhibit similar magnitude of shear. Only few of these bands show a relatively high value of about 0.75. These bands with the highest magnitude of shear were identified by EBSD measurements as regions with a twin orientation to the matrix (see Fig. 15.2e, f) indicated by red lines according to the misorientation relationship of \( 60^\circ \langle{111}\rangle\) for Σ3 (twin) boundaries. Figure 15.2d shows the deformation bands forming on the primary and secondary slip system in backscattered electron contrast. Slip bands or even twins were visible also for the primary slip system, but are quite narrow compared to the large and wide twins belonging to the secondary system. Furthermore, the pronounced distortion of the area of interest due to the favored deformation on the secondary system becomes apparent.

Figure 15.3 shows the deformation of X5CrMnNi16-6-9 steel at 473 K. The calculated von Mises equivalent strain at a global strain of 2% is shown in Fig. 15.3a. Here, the band-like distribution of the strain is recognizable as well. The deformation bands form on the primary slip system parallel to \( \left( {111} \right) \) and show von Mises equivalent strains of about ε_vM = 10%. With increasing global strain, the deformation band structure is still visible, but the areas between the bands also show increased strain values, as can be seen in Fig. 15.3b. When considering the AOI using backscattered electrons (Fig. 15.3c), the formation of band-like structures can be seen as well. Twin formation was not observed using EBSD. In general, the strain is more homogeneously distributed over the AOI. This is due to the increased slip of regular dislocations due to the increased stacking fault energy compared to room temperature. This also becomes visible when considering the magnitude of shear, which is shown in Fig. 15.3d. In contrast to deformation at room temperature, where the magnitude of shear was highest in bands related to twins, a more or less homogeneous magnitude of shear between 0 and 0.2 is observed at 473 K.

Steel with medium austenite stability (6 wt% nickel). Figure 15.4 shows the evolution of strain during tensile deformation of X5CrMnNi16-6-6 cast steel at room temperature. Figure 15.4a shows localized strain within deformation bands at the beginning of the tensile deformation at 3.6% global strain. Noteworthy is the high lateral resolution of the SEM-DIC. The width of the deformation bands which are recognizable by digital image correlation is less than 1 µm.

The strain is very homogeneously distributed within the deformations bands, except the area marked with a white arrow which shows strain localization up to ε_vM = 40%. With increase in macroscopic strain, the number of deformation bands increases, the deformation bands grow in their width and areas of local strain concentrations within these bands increase as well (see Fig. 15.4b). Figure 15.4d, e show the AOI in terms of backscattered electron (BSE) contrast and EBSD phase map, respectively. Both figures reveal the formation of α′-martensite inside deformation bands. In Fig. 15.4e, grey areas correspond to austenite, yellow to ε-martensite and blue to α′-martensite. In comparison with Fig. 15.4b it can be seen that the formation of the deformation bands correlates with the formation of ε-martensite and the strain-increased regions within the bands correlate with the formation of α′-martensite. Figure 15.4f shows that only a certain number of different orientations of α′-martensite grains have been formed within the deformation bands, which is an indication for a variant selection. A comparison with Fig. 15.4c shows that the magnitude of shear calculated in the \( \left[ {0\bar{1}1} \right] \) direction correlates with the orientation of the α′-martensite grains. Only one martensite variant (green color) exhibits the highest magnitude of shear of about 0.7, while the other variants (orange and violet color) show significantly smaller values of shear [31]. In comparison, regions corresponding to ε-martensite show nearly homogeneously distributed shear of about 0.3, which is due to the fact that only one single ε-martensite variant has been formed [31].

Figure 15.5 shows the evolution of strain during tensile deformation of X5CrMnNi16-6-6 cast steel during tensile deformation at 373 K. Due to the increased temperature and, consequently, the increased SFE as well as reduced driving force of ΔG (Gibb’s free energy), no formation of α′-martensite is observed. In Fig. 15.5a at 3.7% global strain, the localization of strain in bands was detected. The size of the bands which can be detected by DIC is below 1 µm and the strain distribution within these bands is quite homogeneous (Fig. 15.5b). With increase in macroscopic strain, the number of deformation bands further increases.

The von Mises strain localization within some bands is about 40% at ε = 15%, but it is in other bands significantly lower. The comparison with results from EBSD measurements (Fig. 15.5c) reveals that regions with highest strain localizations correspond to areas where twins have been formed. The magnitude of shear (Fig. 15.5d) was calculated on \( \left( {111} \right) \) plane in \( \left[ {\bar{2}11} \right] \) direction. The regions with highest shear correspond well with the regions of twins. However, the magnitude of shear is about 0.25 and is, therefore, significantly lower than that Schumann predicted theoretically to be of about 0.7 for twinning [83]. One reason for lower magnitudes of shear could be that the thickness of twin bundles within the broader shear bands is below resolution of the DIC measurements [15]. On the other hand, it is more likely that twins formed in this steel variant are the results of movement and accumulation of partial dislocations. The theoretically calculated magnitude of shear for partial dislocations is of 0.3, which would be in good agreement with the experimentally determined values of about 0.25. Here obviously the areas with high von Mises strain correlate with areas with high shear, in comparison to the deformation at room temperature, where different α′-martensite variants led to a partially inhomogeneous shear distribution depending on which martensite orientation has formed. Such behavior was not observed at 373 K because twins with a uniform orientation have been formed during deformation. As a result, the twins also have a homogeneous shear distribution. Finally, the X5CrMnNi16-6-6 cast steel with medium austenite stability behaves at 373 K like the steel X5CrMnNi16-6-9 with higher austenite stability at room temperature.

Austenitic-martensitic-carbidic steel. Miniature flat tensile specimens were manufactured out of the Q&P treated steel X16CrNiMnSiN with same dimensions as described for the CrMnNi TRIP/TWIP steels. The surfaces of the specimens were grinded and vibration polished. Afterwards, AOIs were defined containing interdendritic austenitic grains, which were pre-characterized by EBSD. Subsequently the miniature specimens were etched for 2 min at 40 °C in V2A etching agent to achieve a structured surface. The quasi in situ deformation in the SEM was performed according to similar procedure as described above for CrMnNi TRIP/TWIP steels. After each deformation step several micrographs at two different magnifications were captured until uniform elongation of the specimen (ε = 19%). The sequence of SEM images was stabilized using the VirtualDub software package [84] and local strain fields were evaluated using the VEDDAC software [85]. After tensile deformation, the gauge length of specimens were cut-off and vibration polished for four hours. Subsequently, EBSD measurements of the AOIs were carried out to evaluate the development of the microstructure after tensile deformation. In addition, XRD measurements were performed on both undeformed and deformed specimens. The phase analysis was carried out by a Rietveld–like refinement of the whole diffraction pattern using the MAUD software package [86]. According to the microstructure, face-centered cubic and body-centered cubic iron were considered. The results are summarized in Table 15.3.

The XRD measurements on the initial state revealed 69 vol% of tempered martensite and 25 vol% of interdendritic austenite. In addition, 6 vol% of interlath austenite were identified based on a slightly increased lattice parameter of a = 0.3609 nm compared to the interdendritic austenite with a = 0.3597 nm. This difference of lattice constants for both austenitic phases is caused by different content of nitrogen and carbon due to various diffusion behavior. Diffusion occurs during partitioning from supersaturated, quenched martensite into austenite. The interdendritic austenite is characterized by grain sizes of >5 µm, whereas the grain size of interlath austenite is significantly smaller (<1 µm). Diffusion length of carbon was estimated by Ågren [87] to be less than 1 µm for partitioning at 723 K for 5 min. Due to the small size of the interlath austenite, it will have a higher carbon and nitrogen concentration than the interdendritic austenite. Thus, for interdendritic austenite the estimated diffusion length is too small to result in a significant increase in content of interstitial elements. The different lattice parameters are, therefore, explained by content of interstitial elements.

In the deformed state, no interdendritic austenite was identified by XRD and only 3 vol% of interlath austenite were remaining after tensile deformation up to ε = 19% with a lattice parameter of a = 0.3605 nm which is comparable to the undeformed state. This means, interlath austenite is much more stable during deformation. This is caused by the higher chemical stabilization due to the higher carbon and nitrogen content, but also due to the surrounding martensite which puts the austenite into a hydrostatic compression state, the so-called shielding effect [88]. In contrast, nearly all interdendritic austenite transformed during deformation. The amount of tempered martensite remains with 68 vol% nearly constant. After deformation, 29 vol% of fresh martensite were formed mostly from interdendritic austenite. The fresh martensite shows a small tetragonal distortion of about 0.6% in comparison to tempered martensite which is caused by higher carbon and nitrogen content, although interdendritic austenite does not gain significantly carbon and nitrogen during partitioning, as already mentioned above. However, the martensite existing before deformation has been tempered during partitioning, reducing the tetragonal distortion.

The influence of the grain orientation on the local deformation behavior of the interdendritic austenite was investigated by DIC in combination with EBSD measurements. Here, two different grain orientations were selected: (i) \( \langle{101}\rangle \) and (ii) \( \langle{001}\rangle \) lattice directions, respectively, parallel to the loading axis. The first one was oriented for slip of partial dislocations and the latter one was oriented for preferential slip of regular dislocations. The microstructural evolution during tensile deformation of \( \left\langle {101} \right\rangle \) oriented interdendritic austenite grain is shown in Fig. 15.6a, b. At the beginning of deformation, a two-phase microstructure consisting of austenite and tempered martensite is present. With beginning of deformation, bands are evolving parallel to the primary slip system. Close to uniform elongation (ε = 19%) a massive martensitic phase transformation occurs starting from the interface of interdendritic austenite/tempered martensite. Figure 15.6b shows that both ε-martensite (hcp) and fresh α′-martensite (bcc) can be found within deformation bands. In contrast, Fig. 15.6c, d show the results for the deformation of the \( \langle{001}\rangle \) oriented interdendritic austenitic grain. Here, no formation of deformation bands was observed. The EBSD measurements reveal only fresh-formed α′-martensite and no ε-martensite was detected. This is what was expected for these two different grain orientations and can be understood by the different behavior of dislocations. Whereas dislocations in grain \( \langle{101}\rangle \) dissociate into Shockley partials leading to stacking faults, grain \( \langle{001}\rangle \) is oriented for preferred movement of regular dislocations. Figure 15.6d shows the \( \langle{001}\rangle \) oriented grain after 19% global strain. Within the austenite pronounced formation of fresh α′-martensite has occurred via the γ → α′ transformation. Thus, both grain orientations exhibit as expected different behavior in terms of the formation of fresh martensite during tensile deformation at RT.

Figure 15.7 shows the marked area of Fig. 15.6a (black rectangle) in more detail. Figure 15.7a shows the interdendritic austenitic grain with tempered martensite before deformation in secondary electron contrast after etching. Etching was optimized for a good contrast of the austenite. Therefore, both tempered martensite and δ-ferrite were slightly over etched (marked by bold and dashed arrows in Fig. 15.7a, respectively). Figure 15.7b shows the same AOI at ε = 19%. The formation of deformation bands within the interdendritic austenite is well observed. Furthermore, additional contrast changes within the deformation bands are recognizable. Figure 15.7c, d show the calculated von Mises equivalent strain at 11% and 19% of global strain, respectively. At ε = 11%, the local accumulation of strain within deformation bands is clearly visible. Furthermore, the strain is homogeneously distributed within these bands. During tensile deformation, ε-martensite is formed which is associated with a homogenous shear operation leading to homogenous von Mises equivalent strain values of about ε_vM = 0.3. With further tensile deformation, these bands grow in width and the activation of a secondary slip system is observed (see white lines in Fig. 15.7d). Furthermore, within the primary deformation bands, areas with enhanced strain localization are formed with values of ε_vM up to 0.7. Looking to the microstructure of the AOI at ε = 19% in Fig. 15.7b it can be seen that the areas of increased strain correlate well with the formation fresh α′-martensite.

Figure 15.6b showed also that α′-martensite formed during tensile deformation, in particular, at the previous interface of tempered martensite and interdendritic austenite. The formation of fresh martensite in this region could not be resolved by DIC. The very pronounced formation of α′-martensite resulted in a high local strain from one deformation step to the other resulting in a loss of the correlation between individual subsets in these areas. The preferred formation of fresh martensite at the austenite/tempered martensite interface can be assigned to the constraining effect [89]. Due to the high defect density at the interfaces caused by the formation of tempered martensite, the formation of fresh martensite is preferred.

Figure 15.8 shows the area of high strain values marked in Fig. 15.7d in more detail. It is visible that the areas of high strain values (see Fig. 15.8a) correlate well with the fresh-formed α′-martensite grains shown in Fig. 15.8b. Thus, two different α′-martensite variants have been formed exhibiting a twin orientation relationship (Σ3 boundaries marked by green line).

Measurements of acoustic emission were performed in situ during tensile deformation of high-alloy X5CrMnNi16-6-x (x = 3, 6 or 9 wt%) cast steels in order to study the temporal evolution of ongoing deformation processes. Therefore, threshold-less, continuous AE signal acquisition method using a single AE sensor was applied. Different individual deformation mechanisms were identified as sources of acoustic emission and their evolution over the entire deformation process was evaluated. In addition, the influence of temperature on operating deformation mechanisms was evaluated. Details on experimental setup were published elsewhere [45, 90]. As demonstrated in [45], the detailed analysis of AE data obtained during tensile tests both in the time as well as frequency domain allowed for the separation of the following deformation mechanisms: (i) dislocation glide (movement of regular dislocations), (ii) formation of stacking faults (partial dislocation movement), (iii) twin nucleation, (iv) formation of α′-martensite, and (v) formation of Portevin Le Chatelier bands. Each of these mechanisms can be characterized by a specific set of AE parameters such as median frequency f_m and energy E of AE signals forming individual clusters of signals. The evolution of these individual clusters was followed over the entire deformation process. Thus, it was possible to evaluate which deformation mechanism becomes dominant at what stress or strain level during the tensile deformation. An example out of the variety of results is shown for the steel with the highest austenite stability (9 wt% nickel) in Fig. 15.9. For this steel it is known from microstructural investigations that mechanical twinning is the dominant deformation mechanism (Fig. 15.2), whereas at temperature T > 80 °C the movement of regular dislocations becomes dominant. Figure 15.9a, c show the results obtained at RT and Fig. 15.9b, d the results from tests at 373 K. The results are plotted either in terms of the cumulated AE energy E_Σ (Fig. 15.9a, b) or the cumulated number of AE cluster elements (Fig. 15.9c, d) always combined with the stress versus time response. These results illustrate well both the differences between the two AE sources (i) dislocation glide (low energy signals) and (ii) twin nucleation (high energy signals) as well as the influence of temperature on these mechanisms. Thus, both mechanisms are active at RT and 373 K. However, the mechanical twinning is dominant at RT whereas its activity is reduced at 373 K. In contrast, the dislocation glide becomes more important at 373 K compared to RT.

Furthermore, it was shown that the evolution of the identified clusters related to individual AE sources correlates well with other microstructural parameters or properties. Thus, it was shown in [45] that the evolution of the identified cluster for the α′-martensite formation correlates well with the evolution of the ferromagnetic phase fraction in case of steels X5CrMnNi16-6-6 and X5CrMnNi16-6-3 with medium and low austenite stability, respectively. Figure 15.10 summarizes the results from AE data analysis (magenta lines), the volume fraction of α′-martensite obtained from ferromagnetic phase fraction measurements using a feritescope (symbols) and the calculation of the volume fraction of α′-martensite according to the Olson-Cohen model [91] (dashed lines).

Figure 15.10 demonstrates an excellent qualitative agreement between experimentally determined volume fraction of strain-induced α′-martensite with the evolution of the cumulated AE energy of the cluster related to α′-martensite formation. Both the onset of martensite formation as well as the different kinetics of α′-martensite formation are very well described by the evolution of the cumulated AE energy of signals related to the cluster corresponding to α′-martensite formation. In addition, both experimental data—AE and feritscope measurements—fit well with the Olson-Cohen model description of the kinetics of α′-martensite formation.

The high-alloy X5CrMnNi17-7-4 cast steel exhibits an excellent strengthening behavior and high ductility yielding, as a consequence, a delay of the macroscopic strain localizations resulting in retarded necking and, finally, failure of the material at higher strain values. However, TRIP steel with the lowest nickel content (3 wt%) exhibits, in addition, serrated plastic flow known as the Portevin Le Chatelier (PLC) effect [93].

The PLC effect was studied on the high-alloy CrMnNi steel X5CrMnNi17-7-4 with 0.05% C, 17% Cr, 7% Mn, 3.7% Ni in the as cast condition after solution heat treatment (0.5 h at 1323 K followed by N₂ gas quenching). Due to the low austenite stability (M_s = 333 K), the initial microstructure consists beside the metastable austenite of about 15 vol% of martensite formed during cooling and 10 vol% of δ-ferrite due to ferritic solidification. The tensile tests were performed at room temperature using an electro-mechanical testing system (Zwick, Germany) using flat tensile specimens having a rectangular cross section of 8 × 4 mm² and a gauge length of 35 mm at a total length of 205 mm. The tests were carried out under crosshead displacement control with different nominal initial strain rates (10⁻⁴ s⁻¹, 10⁻³ s⁻¹, 10⁻² s⁻¹ and 10⁻¹ s⁻¹, respectively). The mechanical tests were complemented by full-field measurements using in situ thermography measurements with the infrared thermography system VarioCamhr (InfraTec, Dresden, Germany) and digital image correlation using digital camera (Canon EOS 420D). One side of the gauge parts of the specimens was prepared for thermographic measurements by mechanical grinding and, subsequently, covered with a black thermo lacquer enabling a constant thermal emission of 0.96. The back side was prepared for digital image correlation using etching procedure as described above for SEM-DIC. Ferromagnetic measurements using feritescope were performed before and after the tensile tests in order to evaluate the volume fraction of strain-induced α′-martensite. Finally, the tensile tests were corroborated, in addition, by in situ acoustic emission measurements. For more experimental details see [93].

The mechanical tests revealed that for strain rates in the range of 10⁻⁴ ≤ \( \dot{\varepsilon } \) ≤ 10⁻² s⁻¹ pronounced inhomogeneous macroscopic flow was observed characterized by numerous serrations occurring in stress-strain curves. Moreover, a shift of the onset of serrated plastic flow within the stress-strain curve was recorded with an increase in strain rate. The tensile deformation was accompanied by both an increase in global temperature as well as volume fraction of strain-induced martensite. No serrated plastic flow was observed at the highest strain rate (\( \dot{\varepsilon } \) = 10⁻¹ s⁻¹), c.f. [93]. The serrations can be related, according to the common understanding in the literature [94], to the localization of plastic strain in macroscopic bands of several micrometer thickness propagating along the gauge length—known as PLC bands. The formation of these macroscopic bands of strain localizations is accompanied by temperature increase which can be recorded using infrared thermography [93]. Figure 15.11 shows exemplarily the results of in situ thermographic measurements during tensile test at room temperature and a nominal strain rate of 10⁻³ s⁻¹. Figure 15.11a shows the true stress versus true strain curve (black) in combination with temperature evolution profiles evaluated at three different points along the gauge length (green, grey, red). For better visibility, the temperature-time curves are shifted by ∆T = 5 K. A continuous temperature increase was recorded in the first part of the stress-strain curve (ε < 0.15) which can be correlated directly with both the onset of plastic deformation as well as the martensitic phase transformation. It is well known from previous investigations [95] that the martensitic phase transformation in this CrMnNi steel variant with 3 wt% Ni starts immediately after passing the yield point. The course of temperature-time curves at higher strain levels (ε > 0.15) reveals clearly an oscillating behavior related to the formation and movement of PLC bands along the gauge length. The PLC bands form on the left side of the gauge length (green curve) and move along the gauge length (grey and red curve). The temperature maxima for the three different points can be followed indicating that one band is followed by another one. In addition, Fig. 15.11b shows a sequence of thermograms taken at different subsequent strain levels (given as ε_t) illustrating the development and propagation of two individual bands along the gauge length. The macroscopic localization of strain inside these bands is related to an increase of temperature of few Kelvin. Moreover, the frequency of emerging PLC bands is quite high. Thus, the specimen gauge length cannot cool down completely before the next band appears resulting in a continuous increase of temperature over the whole test. The analysis of the recorded thermograms allowed for the evaluation of the number of PLC bands, the maximum temperature increase within individual bands as well as their velocity propagating along the gauge length in dependence on the nominal strain rate. It was revealed that with increasing strain rate the number of emerging bands is decreasing whereas the temperature itself is increasing [96].

Furthermore, the evolution of local strain fields related to PLC bands was evaluated by digital image correlation. For this purpose, optical micrographs were taken using a digital camera during the tensile tests at RT and different nominal strain rates. The strain fields were recorded on the opposite side of specimens used for IR-TG measurements. The initiation and propagation of bands along the gauge length is accompanied not only by an increase in temperature, as described already above, but also by an increase in strain. This is exemplarily shown in Fig. 15.12a–c. Figure 15.12a shows not the entire deformation process. Instead, only a short sequence between 0.22 and 0.32 of true strain is plotted. The thermograms and the calculated strain fields are related to numbers 1 to 6 indicated in the stress-strain curve showing the sequence of two consecutive PLC bands. In the present case, the PLC bands initiated always at the lower part of the gauge length and propagated in bottom-up direction on both sides of the specimen as seen from the thermograms and DIC images in Fig. 15.12b–c. The local strain fields are provided in terms of von Mises equivalent strain ε_vM. It turned out from the fully-coupled measurements that the PLC bands are characterized by both temperature and strain localization. Thus, the localized strain within an individual band is about 3%. In addition, the nominal strain rate has a significant influence on the formation of PLC bands. In general, it can be summarized that as lower the strain rate is as more frequently PLC bands occurs. In contrast, both the increase in temperature as well as the velocity of PLC bands are smaller than at higher strain rates. Moreover, different band propagation directions were observed. Thus, at lower strain rates, bands nucleate at different places and run either top-down or bottom-up. At higher strain rates, bands nucleate either at the upper or lower part of the gauge length and propagate then in one direction only [96].

In addition to microstructure evolution and kinetics of deformation processes, the local mechanical behavior of CrMnNi TRIP steel with medium austenite stability X5CrMnNi16-6-6 was investigated by nanoindentation. The focus was set on the local hardness of individual microstructural components (γ-austenite, ε-martensite and α′-martensite). To investigate the strain-hardening behavior, initial hardness of the austenite was determined on an undeformed reference specimen and compared to the local hardness of tensile pre-deformed specimen (ε = 15%). For this purpose, different grain orientations were chosen both with respect to the loading axis during tensile pre-deformation as well as with respect to the indentation axis. Details on the experimental setup can be found in [60].

In the initial state, the measured indentation hardness is between 4.1 and 4.9 GPa and only slightly dependent on grain orientation. In accordance with literature [97], grain orientations with indentation axis near \( \langle{111}\rangle \) show the highest indentation hardness. In some of the indentation experiments, pronounced pop-in events occurred during the indentation of undeformed austenite as shown exemplarily by a load-displacement curve in Fig. 15.13a. At about 550 µN a pronounced deviation from the Hertzian elastic contact solution (dashed line) is observed—the so-called pop-in event. The displacement of this single pop-in event is quite large of about 30 nm. In order to clarify whether the pop-in is the result of an avalanche-like movement of dislocations or of an indentation-induced martensitic phase transformation, a lamella was extracted across the indentation area using focused ion beam (FIB) technique. The lamella was investigated in TEM and the results are shown in Fig. 15.13b.

Figure 15.13b shows that directly below the indent an α′-martensite grain has been formed which was proofed by the diffraction pattern obtained from the Fast-Fourier Transformation of high-resolution TEM image (see insert). In the direct vicinity to the α′-martensite nucleus, stacking faults were found which are assumed to be caused by the indentation of the austenite. In order to examine whether α′-martensite formation can be regarded as the reason for the pop-in event, the theoretical displacement caused by α′-martensite transformation was calculated. The maximum shape strain of the martensitic transformation is assumed to be of 0.24 [98] for the case that the indentation axis is parallel to the direction of maximum shape change. The thickness of the α′-martensite nucleus was measured to be 50 nm which corresponds to a pop-in displacement of about 12 nm. Consequently, α′-martensite formation cannot be the only reason for the pop-in event which corresponds to a displacement of more than 30 nm. More likely, the combination of the formation of stacking faults and α′-martensite together caused this huge displacement. Ahn et al. [99] observed similar pop-in behavior in a high nitrogen steel and excluded also the α′-martensite formation as the only reason for the pop-in event.

In order to determine the indentation hardness of the microstructural constituents in the tensile pre-deformed state, indentation maps were placed within austenitic grains of different orientations covering all three microstructural constituents in a pre-deformed state of ε = 15%: (i) austenite, (ii) deformation bands with ε-martensite, and (iii) α′-martensite grains. The indentation was performed with a load of 1 mN and the indents were placed with a minimum distance of 1.5 µm. An example for a grain with \( \langle{125}\rangle \) as indentation axis is shown in Fig. 15.14 (see [60]).

Here, Fig. 15.14a, b show results of EBSD measurements for the area with indentation map and Fig. 15.14c of a single indent placed within the area of a deformation band yielding a pop-in event in the load-displacement curve. The area of FIB lamella for TEM investigations is indicated by white rectangle. Figure 15.14d–f show the corresponding t-SEM and TEM results. In the EBSD maps the austenitic phase is shown in grey, the ε-martensite bands in yellow and the α′-martensite in blue color. Details on all load-displacement curves and indentation hardness and modulus are summarized in [60]. In general, it can be summarized that the overall indentation hardness values of the tensile pre-deformed state are increased in average of about 19% compared to the non-deformed state as it was expected. In addition, an orientation dependency was found for indentation hardness values in the pre-deformed specimen. Thus, austenitic grains with \( \langle{101}\rangle \) or \( \langle{111}\rangle \) orientation of the loading axis during the tensile pre-deformation showed higher indentation hardness values compared to other grain orientations such as \( \langle{001}\rangle \). These grain orientations are favored for the formation of largely extended stacking faults resulting in a significant reduction of mean free path of dislocations yielding pronounced strain hardening. In contrast, grain orientations close to \( \langle{001}\rangle \) are favored for slip of regular dislocations resulting in a lower density of deformation bands and α′-martensite nuclei causing lower strain hardening effect. Such an example is shown in Fig. 15.14a for orientation \( \langle{125}\rangle \) of loading axis.

In addition, austenite orientations which showed increased slip of regular dislocations can experience clear differences in the strain hardening. Figure 15.14b shows a kernal average misorientation (KAM) map with marked indents within austenite. Red triangles correspond to indents within an austenite with high KAM and black triangles correspond to indents within austenite with low KAM. Thus, areas with increased KAM show an average indentation hardness increase of about 12% compared to indents within lower KAM. This is due to the higher defect density; the higher KAM is a result of a higher number of geometrically necessary dislocations and thus correlates with a higher dislocation density [100].

The indentation hardness of ε-martensite was always higher than the hardness of the deformed austenite and was between 5.7 and 8.5 GPa. A dependency of the hardness on the orientation of the ε-martensite was recognizable. (0001) perpendicular to the direction of indentation showed the highest indentation hardness. In addition, some ε-martensite orientations showed also a distinct pop-in behavior. Figure 15.14c shows a band contrast map with a deformation band consisting of ε-martensite and an α′-martensite nucleus which has formed by the pre-deformation. The green and red lines mark the Kurdjumov-Sachs orientation relationship [101] between α′/ε and α′/γ. The blue triangle marks the position of an indent with occurring pop-in event within ε-martensite. In order to investigate the reason for the pop-in event a FIB lamella was extracted (marked by white box). Figure 15.14d, e show the cross section of the extracted FIB lamella observed in SEM in transmission mode (t-SEM) and in TEM, respectively.

The lamella shows numerous bcc nuclei with a plate-like structure which are inclined at an angle of approximately 45°. By selected area diffraction (SAED) it was shown that all of that plate-like α′-martensite nuclei had the same orientation. Due to the similar orientation and morphology, it can be concluded that these α′-martensite nuclei are the result of the pre-deformation. Directly beneath the indent an additional bcc grain has formed. In contrast to all other martensite nuclei, this one has a different morphology and also a different orientation, which implies that this α′-martensite is the result of indentation. In addition, the 45° tilted α′-martensite grain shows a small misalignment which is marked by the white box in Fig. 15.14e and is shown in detail in Fig. 15.14f. This misalignment (marked by orange lines) is caused by the indentation-induced martensite. The expanding α′-martensite grain ran up to or even into the already existing deformation-induced α′ martensite and caused a small shearing effect. Furthermore, the indentation-induced martensite contains a grain boundary (marked by blue line). SAED observations and the fact that just one pop-in event was noticeable in the load displacement curve indicate that the formed indentation-induced α′-martensite grains are twins [60].

The hardness of the α′-martensite introduced by the pre-deformation was determined to 6.3 GPa up to 10.5 GPa. However, the hardness of α′-martensite depends on several aspects. On the one hand, hardness is determined by the orientation of the crystal. For example, α′-martensite grains with an orientation close to \( \langle{111}\rangle\) showed the highest indentation hardness. Furthermore, the grain size does affect the determined indentation hardness since additional interfacial effects between dislocations and grain boundaries have to be considered. In addition, also the defect density of the studied α′-martensite additionally influences the determined indentation hardness [60].