Effect of martensite distribution on damage behaviour in DP600 dual phase steels

Materials Science and Engineering A 516 (2009) 7–16

Contents lists available at ScienceDirect

Materials Science and Engineering A journal homepage: www.elsevier.com/locate/msea

Effect of martensite distribution on damage behaviour in DP600 dual phase steels G. Avramovic-Cingara a,∗ , Y. Ososkov a , M.K. Jain b , D.S. Wilkinson a a b

Department of Materials Science and Engineering, McMaster University, 1280 Main Street West, Hamilton, Ont., Canada L8S 4L7 Department of Mechanical Engineering, McMaster University, Hamilton, Ont., Canada

a r t i c l e

i n f o

Article history: Received 6 October 2008 Received in revised form 9 February 2009 Accepted 23 March 2009 Keywords: Dual phase steel DP600 Damage Voids Martensite

a b s t r a c t The effect of martensite morphology and distribution in a ferrite matrix on the mechanical properties and the damage accumulation in uniaxial tension was investigated in two different automotive-grade dual phase DP600 steels. The two sheet steels had roughly 20% volume fraction of martensite but dissimilar chemical composition. A detailed analysis of microstructure and damage accumulation has been conducted as a function of strain. SEM analysis revealed that voids nucleation occurs by martensite cracking, separation of adjacent martensite regions, or by decohesion at the ferrite/martensite interface. Martensite morphology and distribution had a significant influence in the accumulation of damage. The steel with a more uniform distribution of martensite showed a slower rate of damage growth and a continuous void nucleation during the deformation process, which resulted in a higher void density before fracture. On the other hand, the steel with a centre-line of martensite through the sheet thickness exhibited accelerated void growth and catastrophic coalescence in the transverse orientation to the applied load. Crown Copyright © 2009 Published by Elsevier B.V. All rights reserved.

1. Introduction The need to reduce the fuel consumption and emissions, while maintaining vehicle safety, is the main driving force for the lower vehicle weight in the next new generation of automobile design. Application of high-strength dual phase (DP) steels combined with new production technologies is being considered as one of the most efficient ways to achieve the above goal. Dual phase steels exhibit low yield strength and a high workhardening rate, thus providing a high-strength steel of superior formability [1–3]. Despite the generic name “dual phase”, such steel may contain three or more phases. The matrix is typically soft ductile ferrite. The strengthening phase is martensite but may contain small amounts of bainite and retained austenite. It is now established that the martensite volume fraction is dominant in controlling tensile properties and increasing the amount of martensite decreases ductility. Studies have shown that morphology of martensite particles plays an important role in the strength and ductility of the dual phase steels [2–6]. However, most of the research work has been focused on comparison of martensite morphology produced by some variations of two basic heat treatments, the quenching (or step quenching) process from austenite region, or the intercritical annealing [3,5–12]. Martensite particles in the first treatment tend to have the same crystallographic orientation as the surrounding ferrite matrix [5,6]. As expected, there are remarkable differences

∗ Corresponding author. Tel.: +1 905 525 9140x27844; fax: +1 905 528 9295. E-mail address: [email protected] (G. Avramovic-Cingara).

in the resulting morphologies and most of the studies have compared the fine (uniform dispersion) and coarse (banded, or fibrous) martensite morphologies [3,6,8,10]. For a constant volume fraction of martensite, a microstructure of finely dispersed martensite has a better combination of strength and ductility. According to Kim and Thomas [3], cleavage fracture occurs at ferrite in a coarse martensite structure, whereas voids initiate at ferrite–martensite interfaces in a fine martensite structure. It has been demonstrated that optimum properties of DP steels are obtained at approximately 20% volume fraction of martensite [1,4]. In order to predict deformation behaviour of DP steels, one needs to have an understanding of constituent phases and also the partitioning of stress and strain between the two phases during deformation [13–16]. Different damage mechanisms of particular dual phase steels are related also to their chemical compositions, heat treatment history, and differences in their final microstructure [3–24]. Stevenson [12] reported that cracks initiate first in martensite under low strain and then propagate into ferrite. The fracture mechanism in a fine and coarse martensite morphology dual phase steels having 0.09% carbon and 17% martensite was studied by He et al. [11]. It was reported that in the coarse structures the initial void formation occurs due to cracking of the martensite at very low strain levels. At a higher strain this is followed by the formation of a second type of void by interfacial decohesion at ferrite/martensite interfaces. However, in the structures with fine martensite morphology, the majority of voids are formed by decohesion at the ferrite/martensite interface. This has been attributed to the widely different stress–strain characteristics of the two phases resulting in significant strain incompatibility [14,21]. The martensite under-

0921-5093/$ – see front matter. Crown Copyright © 2009 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2009.03.055

8

G. Avramovic-Cingara et al. / Materials Science and Engineering A 516 (2009) 7–16

Table 1 The chemical compositions of steels in wt.%. Steel

C

Mn

Mo

Si

Cr

Ni

Ti

A B

0.070 0.106

1.840 1.530

0.150 0.220

0.090 0.201

0.030 0.190

0.010 0.030

0.010 0.018

goes significantly less deformation than the ferrite and the voids form at the ferrite/martensite interface. On the other hand, significant plastic deformation of martensite occurs when its strength is reduced either by carbon content or by tempering. Szewczyk and Gurland [25] reported that extensive plastic deformation of martensite occurs mainly in the neck region of the tensile specimen. Localized deformation of martensite particles as a special distinctive void nucleation mechanism was studied by Erdogan [18] and Steinbrunner et al. [19]. Steinbrunner et al. [19] observed that the nucleation of voids due to localized deformation within the martensite or by martensite particle separation may be more complex than previously thought. A mechanism for separation of deformed martensite particles was proposed. In the present work, two different versions of commercially produced DP600 sheet steels are studied. Both steels have banded martensite morphology with different distribution of bands. The steels are produced by intercritical annealing, have similar mechanical properties, and slightly different chemical compositions. The main objective of this work is to compare deformation and damage behaviour of these two dual phase steel sheets during uniaxial tensile straining. In addition, an effort has been made to clarify the effect of banded martensite morphology on the damage nucleation and development, as well as fracture behaviour in the two steels, designated here as “DP600-A” and “DP600-B”. The damage accumulation was investigated by quantitative metallographic analysis of deformed uniaxial tensile specimens. Further, the formation of damage was evaluated using post-test scanning electron microscopy (SEM), as well as by in situ SEM. 2. Experimental procedure The two commercial high-strength dual phase DP600 steels, DP600-A and DP600-B, were studied in the as-received condition. The galvannealed steels were received in the form of 1.8 mm thick sheets. The galvannealing procedure has been widely used in the steel industry to promote inter-diffusion of zinc and iron, leading to an alloyed coating of better quality [26]. The chemical compositions of the two steels are shown in Table 1. The DP600-B steel contained higher levels of C, Mo, Si, and Cr compared to the DP600-A.

Tensile testing was performed at a crosshead speed of 1 mm/min on a servo-hydraulic MTS frame with 100 kN load-cell capacity. The tensile specimens were machined according to ASTM E8 standard with a gauge length of 25.4 mm in such a way that the applied tensile loading axis corresponded to either the rolling direction (RD) or transverse direction (TD) of the sheet [27]. As-received steel samples were etched in various etching solutions in order to clearly delineate the information about the dual phase steel constituents using light microscopy. Method 1 represents of a two-stage etching procedure consisting of 4% picral, followed by 10% aqueous sodium metabisulfite (SMB) solution [28,29]. It should be mentioned here that picral attacks interfaces between ferrite and carbide, therefore, carbides were better revealed by picral [29]. The distinct colour contrast between the martensite, bainite, and the ferrite matrix enabled measurement of the volume fraction of martensite through image analysis. Various samples were analyzed in all three-dimensional planes. Throughout the method 2 the samples were also separately etched with 2% nital to reveal ferrite grains and grain boundaries, as well as the martensite. The examination of steel microstructure and void spatial distribution was performed using a Zeiss Axioplan 2 light optical microscope. The area fractions of microstructural constituents were determined from optical micrographs using Northern Eclipse image analysis software [30]. The mean ferrite grain size was determined by the linear intercept method. SEM examination was conducted using a SEM Philips 515 and a JSM-7000F FESEM. X-ray diffraction analysis was applied to measure the retained austenite according to the ASTM E-975 standard. Metallographic analysis of damage accumulated along the gauge length after uniaxial tensile testing was carried out on failed samples, on cross-sections along the tensile axis. Fractured specimens were sectioned through-thickness along the mid-width in longitudinal direction. Both sides of failed tensile samples were characterized. To preserve any damage during specimen preparation, wire electrical discharge machining (WEDM) was used. The amount of damage was measured on polished samples, as a function of thickness reduction at different distances from the fracture surface and analyzed with Northern Eclipse image analysis software [30]. Consecutive fields along the same thickness strain were measured optically in order to determine the average through-thickness void data. The accuracy of small voids characterization by optical metallography was verified for DP600-B at higher magnification using a JSM-7000F FESEM. The SEM analysis of void nucleation mechanisms was carried out on the same samples etched later in 2.5% nital. For DP600-A, SEM in situ analysis was conducted by a screw-driven mini-tensile SEM stage (Ernest F. Fullam Inc., Lantham, NY) installed inside an environmental SEM (Electroscan

Fig. 1. Engineering stress–strain curves for (a) DP600-A and (b) DP600-B steels.

G. Avramovic-Cingara et al. / Materials Science and Engineering A 516 (2009) 7–16 Table 2 Uniaxial tensile test data (RD). Steel

UTSav (MPa)

 y (MPa)

εf (%)

n

K

DP600-A DP600-B

600 650

360 360

24 25

0.19 0.21

1020 1180

2020) which enabled observations of the sample surface during the deformation. The details of the SEM in situ analysis are reported in Ref. [31]. The Von Misses equivalent strain (εeq ) was calculated from current thickness (t) values and the initial thickness (t0 ) of the sheet using εeq = −2 ln(t/t0 ). 3. Results

9

morphological features of martensite phase are clearly revealed at higher magnification (nital etching). The average microstructural parameters for the two steels are summarized in Table 3. As noted, the average ferrite grain size of approximately 8 ␮m was obtained for both steels. The martensite volume fraction in the steel A determined by method 1 was 17.2 ± 3%, while 23.0 ± 3% was revealed by nital etching (method 2). In the steel B the volume fraction of martensite and bainite was determined by the two step etching method (method 1). The light microscopy conducted on a few samples at various planes of the steel B revealed 16.9 ± 5% of martensite and 6.8 ± 2% of bainite. Relatively large deviation in results is related to the fact that image analyses were conducted at higher magnifications including various areas of the samples. X-ray diffraction analysis of DP600-B exhibited about 1% of retained austenite and 7% bainite, while no retained austenite and bainite were obtained in DP600-A steel.

3.1. Tensile data 3.3. Damage accumulation Typical engineering uniaxial tensile stress–strain curves of DP600 steels A and B, along rolling and transverse directions, are given in Fig. 1(a) and (b). It is seen that there is no significant difference in the flow curves in RD, and TD directions for the two steels. Both steels are characterized by very uniform plastic flow until necking. Table 2 summarizes the tensile test data for steels A and B in terms of ultimate tensile strength (UTS), yield strength ( y ), strain at fracture (εf ), strain hardening exponent (n), and strength coefficient (K); the last two parameters were obtained by fitting the data to Hollomon equation  = Kεn . The material property data for DP600-A was reported earlier in Ref. [32] and is reproduced in Table 2. 3.2. Steel microstructures Fig. 2(a) shows typical optical microstructure of DP600-A in the centre part of the sheet through-thickness cross-section comprising of a ferrite matrix and martensite second phase. Fig. 2(b) shows SEM image of the microstructure from a similar area. It is seen that the spatial distribution of martensite is not uniform: the microstructure exhibits martensite banding located in the mid-thickness of the sheet and parallel to the rolling plane. Optical microstructure of the as-received DP600-B steel is shown in Fig. 3(a) to illustrate the distribution of the martensite bands throughout the sheet thickness. Fig. 3(c) obtained in normal plane (NP) clearly reveals martensite in brown colour, embedded in bainite in black. In the DP600-B steel, martensite appears linked in bands parallel to the rolling direction with a relatively uniform distribution through the sheet thickness, or as dispersed particles, with sizes in the range of 2 ␮m or more. Fig. 3(b) shows an SEM image of the microstructure of the same steel in the ND-RD plane, where

Optical micrographs of polished longitudinal cross-sections taken through the necks of deformed DP600-A and B samples, are shown in Fig. 4(a) and (b). These micrographs clearly reveal differences in voids distribution in the two materials. It is to be noted that in both cases, the horizontal and vertical axes corresponded to the rolling and thickness direction respectively. As shown in Fig. 4(a), the biggest voids are located at the centre of the sheet in DP600-A and are associated with the martensite centre-line. Other strings of much smaller voids are observed parallel to the centre-line. Fig. 4(b) illustrates a typical polished longitudinal cross-section of a fractured tensile sample of DP600-B steel. The strings of voids are distributed more uniformly through the thickness and the spacing between bands corresponds generally with the martensite distribution as shown earlier in Fig. 3(a). The void density increases toward the fracture surface. The void area fraction as a function of equivalent strain is shown in Fig. 5 for both steels. The scatter along the vertical axes for DP600B is the result of measurements carried out in different (top, centre, and bottom) regions of the sample at the same strain value. The data at the zero strain represents the initial void fraction measured in various locations in the as-received DP600-B. Two observations can be made: (i) a void area fraction increases more slowly with strain for DP600-B compared to DP600-A, and (ii) the extent of void damage at fracture is greater in DP600-B. 3.4. Void nucleation and growth SEM microstructure analysis of samples revealed three distinctive mechanisms of voids nucleation in DP600-A steel: cracking of the martensite, decohesion at ferrite–martensite interface, and sep-

Fig. 2. DP600-A steel microstructure in ND-RD plane: (a) optical micrograph showing martensite centre-line (dark regions), etched with picral followed by SMB. (b) SEM image showing martensite (light regions), etched with nital.

10

G. Avramovic-Cingara et al. / Materials Science and Engineering A 516 (2009) 7–16

Fig. 3. (a and b) DP600-B steel microstructure: (a) Optical micrograph showing martensite distribution in bands, ND-RD plane; (b) SEM image showing martensite, bainite and ferrite, ND-RD plane, etched with 2% nital. (c) Optical microstructure in normal plane (i.e. RD-TD plane) showing brown martensite and black bainite, etched with picral followed by SMB. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.) Table 3 Microstructural parameters. Steel

A B

Ferrite grain size (␮m)

7.8 8

Martensite area fraction (%)

Bainite area fraction (%)

Method 1

Method 2

Method 1

17.2 ± 3 16.9 ± 5

23.0 ± 3

– 6.8 ± 2

aration of adjacent martensite regions. Typical examples of these mechanisms are illustrated in Fig. 6(a) and (b) where Fig. 6(a) reveals voids at the martensite centre-line band (approximate location of this area is marked “A” in Fig. 4(a)). Fig. 6(b), taken away of

Retained austenite (%)

– 1

the centre-line (area marked “B” in Fig. 4(a)), reveals that majority of voids formed at ferrite–martensite interfaces were located between closely spaced martensite particles. These voids nucleated in the areas away from the centre-line were significantly smaller in

Fig. 4. Voids revealed by light microscopy in a polished through-thickness longitudinal cross-sections of (a) DP600-A and (b) DP600-B steels. A string of large voids is noticeable at the centre-line in DP600-A.

G. Avramovic-Cingara et al. / Materials Science and Engineering A 516 (2009) 7–16

Fig. 5. Comparison of damage accumulation curves in the DP600-A and the DP600-B steels.

comparison with voids formed at the martensite band. Both sets of voids, those originally nucleated at ferrite/martensite interface by decohesion and those formed by particles separation, elongate in the direction of applied load with increasing strain. This is accompanied by significant amount of plastic deformation in the adjacent ferrite grains. Observation of non-metallic inclusions on both as-polished cross-sections prior to tensile testing as well as on the fracture surfaces revealed roughly their homogeneous distribution with a small variation in size and shape. The results indicate that the area fraction is rather low, having an average value of 0.037%. EDX examination showed that the round inclusions were a mixture of manganese-sulphide and aluminum-oxide, whereas the elongated ones showed manganese-sulphide. Both types of inclusions contain a considerable amount of calcium. Most of non-metallic inclusions contained voids before tensile testing as a result of prior thermomechanical history. However, the inclusions do not appear to play a significant role in the overall damage development. This has been supported by SEM observations which showed that only a relatively small proportion of inclusions seem to be associated with the nucleation of micro-voids at the fracture surface. Fig. 7(a)–(d) shows the in situ damage process in DP600-A. Martensite band, shown in undeformed state in Fig. 7(a), starts to crack in Fig. 7(b), an earlier stages of deformation. Internal

11

martensite cracking preceded the ferrite–martensite decohesion in this material. Damage of martensite band started during uniform deformation, as evidenced in Fig. 7(b), started to grow in localized necking region, as revealed in Fig. 7(c) and (d). Extensive void growth at the martensite centre-line promotes the localization of strain through the sheet thickness. With increased loading, the deformation propagates into the neighboring ferrite grains and advances towards the surface. Two intersecting shear bands or a collection of intense slip bands formed in the ferrite are marked by arrows in Fig. 7(c) and (d). A critical local strain value of 0.029 for martensite cracking was obtained. Both events, martensite cracking and ferrite–martensite decohesion, occurred at tensile deformations significantly lower than the onset of macroscopic tensile instability. As reported earlier [31], significant strain partitioning between soft and hard areas was measured experimentally in the in situ experiments. Critical local true strain for void nucleation by ferrite–martensite decohesion of 0.09 was determined. The SEM observations of DP600-B, as shown in Fig. 8(a)–(f), clearly reveal the sites of void nucleation. Void nucleation in the deformed DP600-B samples occurred by four different processes. A small percentage of voids was nucleated by martensite particle cracking as shown by arrow in Fig. 8(b) and (c), corresponding to equivalent strains of 0.5 and 1.02 respectively. This mechanism occurs mostly on coarse martensite particles, or particles interconnected through band-like distribution of martensite phases. It is important to note that voids nucleated on fractured martensite do not seem to grow significantly Fig. 8(c) and (d). The main population of voids formed at higher strain was nucleated at the ferrite/martensite interface (or triple junction) by decohesion. Voids nucleation occurs generally on the interface perpendicular to the tensile axis. With increasing strain, the voids grow longitudinally along the grain boundaries, i.e. parallel to the direction of the applied tensile load, as illustrated in Fig. 8(d) and (e). In DP600-B a small number of voids are also formed on the titanium-nitride and the calcium-aluminum-silicate inclusions at low strains, as shown in Fig. 8(a). These voids tend to be large and thus contribute to the void area fraction as presented in Ref. [29]. The propagation of ductile crack occurs by the joining up of voids. The resulting coalescence along shear bands in a region close to fracture is illustrated in Fig. 8(f). 3.5. Fracture morphology Both DP600-A and DP600-B steels, displayed a “cup and cone” type of fracture. The morphology of the fracture surface of DP600-

Fig. 6. The SEM micrographs of void formation in DP600-A: (a) martensite cracking at the centre-line and separation of adjacent martensite areas; (b) ferrite/martensite decohesion in the area away from the sheet centre, RD-ND plane.

12

G. Avramovic-Cingara et al. / Materials Science and Engineering A 516 (2009) 7–16

Fig. 7. SEM in situ tensile tests showing sequences of damage nucleation and evolution at martensite centre-line in DP600-A steel: (a) undeformed state, (b) uniform elongation, and (c and d) void growth events at localized necking. The arrows indicate shear bands.

A is illustrated in Fig. 9 indicating a dimpled rupture as the prime failure mode. At low magnification (Fig. 9(a)), large elongated voids (some of them were up to 200 ␮m in size) are seen in the middle part of the fracture surface corresponding to the sheet centre-line. A number of small voids in the areas away from the centre-line are also seen. At higher magnification (Fig. 9(b)) elongated dimples of 3–5 ␮m average length are evident in the central part of

the fracture. A few ductile dimples associated with non-metallic inclusions were also noted. The fracture surface adjacent to the centre-line displays a smoother appearance at low magnification than the central part of the fracture. The elongated shear-type dimples were observed along the sides of the specimen suggesting that a substantial amount of plastic deformation occurred before complete separation of the fracture surface by localized shear. As seen

Fig. 8. Void nucleation in the DP600-B steel at various equivalent strains: (a) on inclusions; (b and c) martensite cracking; (d and e) decohesion at ferrite/martensite interface and propagation along the ferrite grain boundaries; (f) coalescence of voids close to fracture surface.

G. Avramovic-Cingara et al. / Materials Science and Engineering A 516 (2009) 7–16

13

Fig. 9. SEM fractography of the DP600-A steel: (a) overview of the fracture surface; (b) higher magnification fractograph of the rectangular region marked in (a) showing morphology of large dimples; (c) martensite and ferrite areas.

in Fig. 9(a) and (b), preferential void growth and coalescence occurs along the transverse plane normal to the applied load in DP600-A. SEM fractographs of a DP600-B tensile sample at high magnifications from different regions of the fracture surface, as per low magnification fracture surface in Fig. 10(a), are shown in Fig. 10(b)–(f). The pattern and size distribution of the dimples on fracture surface reflect the distribution of originating particles. A uniform background of smaller dimples, (Fig. 10(e) and (d)), is due to the interfacial nucleation sites, and the parallel striations reflect the void nucleation in fibrous martensite bands. Different dimple sizes in ridges and valleys are also observed. SEM observation reveals marked shear distortion along the sides of the specimen, Fig. 10(c), as indicative of the ductile fracture mechanism. The dimples in vicinity of the shear zone are elongated in the direction of shear. The elongated voids are shallow, as compared with to DP600-A. A few large voids were observed in the vicinity of inclusions (calcium-aluminum-silicate or titanium-nitride), as shown in Fig. 10(g). In summary, a careful examination of the whole DP600B steel fracture surface did not reveal very large dimples similar to those in DP600-A steel. 4. Discussion As shown in Table 2, there was no significant difference in the uniaxial tensile data between the DP600-A and the -B steels. However, the damage and fracture behaviour is significantly influenced by the original microstructure. In attempting to understand the results presented here, it is useful to consider the role of processing and composition. Both strip steels have been processed by hot rolling followed by cold-rolling and intercritical annealing. The exact processing conditions of these commercial materials are proprietary. The microstructure of these coated galvannealed commercial dual phase steels before the intercritical annealing consisted of cold-rolled, elongated ferrite grains and pearlite colonies. Much of the martensite is therefore inherited structure. Martensite banding morphology has been observed by others and is a characteristic of as-rolled dual phase steels with manganese segregation [33,34]. The DP600-A steel had banded microstructure with distinctive central band. Martensite distribution in the form of relatively uniformly distributed bands and small dispersed particles was revealed in the DP600-B steel, including 7% of bainite as well.

The difference in the martensite content in two steels is not significant and is in the range of standard deviation. Ferrite grains in both steels were fully recrystallized, as confirmed by the SEM/EBSD analysis. 4.1. Damage mechanism in the DP600-A steel The DP600-A steel exhibited four different distinctive damage mechanisms: cracking of martensite, decohesion at ferrite/martensite interface, separation of martensite particles, and void nucleation on inclusions. Non-metallic inclusions did not play a significant role in the overall damage development due to their low fraction. The extent of void nucleation was related to the martensite connectivity and its spatial distribution. The SEM in situ tensile tests results demonstrated that martensite cracking started initially at very low local strains (0.029). Immediately after initiation, these cracks propagate rapidly throughout the volume occupied by the cracked martensite regions giving rise to an initial increase in the void area fraction. As deformation proceeds, the voids initiated at martensite bands continue to grow by further separation of the cracked martensite areas rather than by void coalescence, resulting in the increase in the area fraction of the biggest void (Fig. 4(a)). At larger strains voids open up between adjacent martensite particles and at ferrite–martensite interface where the compatibility of plastic deformation can no longer be maintained. As per SEM in situ evidence, local true strain for void nucleation by ferrite/martensite decohesion of 0.09 was obtained. These voids propagate into the ferrite and elongate in the direction of straining, giving a pronounced increase in the void density and total void area fraction. The result is in agreement with the work of He et al. [11] who reported that this mechanism of void formation was particularly intense at high strains in the necked region, and the growth and coalescence of these voids was much slower. In the work of He et al., the microstructure observations show two different possibilities for the linkage of voids depending on the stress triaxiality. At low triaxiality, only a few voids exhibit the void-sheeting mechanism of coalescence, showing the linkage at 45◦ orientation to the tensile direction. This implies that, in the present case, the interaction between voids is rather weak and does not significantly affect the void growth in the direction of straining. In the localized neck area

14

G. Avramovic-Cingara et al. / Materials Science and Engineering A 516 (2009) 7–16

Fig. 10. SEM fractography of the DP600-B steel: SEM fractography of the DP600-B fracture surface: (a–f) ductile dimples; (e and d) different dimple size in ridges and valleys; (g) inclusions present within the dimples.

(high stress triaxiality), void distribution suggests that void growth and coalescence is occurring preferably in the plane normal to the applied axial load (compare Figs. 4(a) and 11). Fig. 11, obtained in polished condition on the transverse cross-section in the area below the fracture surface, reveals the coalescence of cracks under the fracture area. No evidence of accelerated coalescence at high equivalent strain has been found on the longitudinal plane (Fig. 4(a)) in the areas adjacent to the fracture, in comparison to the transverse plane of the sample (Fig. 11), which illustrates the highly localized and catastrophic character of the final coalescence process. Finally, as the central cavity forms, the material loses its load carrying capacity, which is the last stage. The specimens after failure reveal a series of large dimples elongated along the transverse direction. For comparison, DP600-B reveals a more uniform size and shape of dimples through the sheet thickness (compare Figs. 9(a) and 10(a)). In addition, the deformation localization process may result in the eventual nucleation of “secondary” micro-

voids within the shear localization band joining the primary voids. As the central fracture advances toward the surface, eventually a condition is reached where the shear band extend to the surface. A catastrophic propagation of these bands will result in the formation of the shear portion of the fracture surface. The presence of ductile shear dimples in that area illustrates a considerable amount of accumulated damage in the void shear linkage process. There are only a few in situ SEM studies on dual phase steels [7,37]. Suh et al. examined in situ the dual phase steel with 0.13% carbon in order to compare fine and coarse structures obtained by different heat treatments [7]. They reported that, in the case of dual phase steels having coarse martensite, shear bands were formed inside ferrites only, while martensite hardly deformed until fracture. The deformation was predominantly localized in the ferrite matrix. It is clear from the results shown in Fig. 7 that, in DP600-A, martensite deforms first and deformation propagates into ferrite grains. It has to be emphasized here that the nucleation strains for martensite cracking (0.029) and decohesion (0.09) are local microscopic strains [31]. Plastic deformation of a single crystal due to dislocations is inherently heterogeneous at a microscopic level [40,41]. Each individual grain in a polycrystalline aggregate deforms differently from an imposed macroscopically deformation field. The origin of significant strain fluctuation in dual phase steels can be attributed to non-homogeneous deformation of ferrite grains and especially incompatibility of soft ferrite and hard martensite phases [31]. 4.2. Damage mechanism in the DP600-B steel

Fig. 11. Voids located on the transverse section under the fracture surface, polished condition. Tensile axis is normal to the picture, DP600-A.

Primary mechanism for void formation in DP600-B steel is decohesion of the ferrite/martensite interface. Majority of voids formed on the ferrite/martensite interface grew along the ferrite grain boundaries, parallel to the applied tensile load. The voids formed

G. Avramovic-Cingara et al. / Materials Science and Engineering A 516 (2009) 7–16

under uniaxial tensile stress conditions tend to elongate in the tensile direction without coalescing. Thus, the ferrite phase can continue to deform without fracture. Voids that form under triaxial stress conditions such as those found in the neck of a tensile test specimen tend to grow in the transverse direction and eventually coalesce resulting in the fracture of the steel (Fig. 8(f)). Voids nucleated on fractured martensite were formed on the interface perpendicular to the tensile axis, but they did not grow significantly. Szewczyk and Gurland [25] analyzed the ductile fracture mechanism in a dual phase steel with 0.12% C (16 wt.% fraction of martensite, ferrite grain size of 7.3 ␮m), which is similar to the DP600-B. They found that damage begins with void formation at martensite/ferrite interfaces, and most frequently at the poles of closely spaced martensite particles situated at ferrite grain boundaries. In this steel voids were classified into two groups, nucleated and grown voids, and behaviour of both groups was analyzed and reported elsewhere [29]. Fig. 12, from Ref. [29], shows the void density of two categories of void sizes found in the DP600-B, those having diameter greater than 1 ␮m and the others having a diameter less than or equal to 1 ␮m. When such a classification is applied, it is clear that the density of small as well as larger voids increases with strain. The presence of the smaller voids category at all strains means that the nucleation process is continuous during deformation. Since voids nucleated on inclusions do not follow the same mechanism, they are recorded separately from martensite particles. Fig. 13, from Ref. [29], represents variation of average void area fraction as a function of thickness strain. Two different void size groups are described by separate curves. For the lower size range curve, the intercept along the thickness strain axis occurs at a thickness strain value of 0.11 (or an equivalent strain of 0.22). It is to be noted that nucleation in DP600-B occurs at an average (macroscopic) equivalent strain of 0.22. Voids were observed in the DP600-B at equivalent strains lower than 0.22. However, it is difficult to separate initial void fraction in the as-received material and that obtained at small strains. The rate of void nucleation is expected to be a function of many parameters such as the volume fraction, the size and the distribution of martensite, as well as hydrostatic stress. The void area fraction of grown voids increases exponentially with strain.

Fig. 12. Average void density vs. thickness strain for two categories of void size [29].

15

Fig. 13. Void area fraction vs. thickness strain for two categories of void size [29].

Present results indicate different damage mechanisms in DP600-A and B steels, influenced in principal primarily by martensite morphology. The damage mechanisms of nucleation, growth and coalescence in DP600-A and B steels showed rather dissimilar characteristics leading to distinctive fracture morphologies of steels in question. Second important fact to consider is also the chemical composition of steels, especially the carbon content is in order. The DP600-A contained a lower level of carbon (0.070% vs. 0.106% for DP600-B), which might influence behaviour of martensite in this steel. Mazinani and Poole [15,16] have recently analyzed the deformation of martensite in a low-carbon dual phase steel with 0.06% C. They reported that martensite plasticity is favoured when its strength is reduced by lowering its carbon content, or when the martensite islands morphology is changed from equiaxed to banded. The effect of carbon in the two materials and in particular its effect on the strength and ductility of the respective martensite phase can be explained in the following way. A lower carbon content in DP600-A, and consequently a lower carbon content of its martensite phase, is expected to lead to lower strength and higher ductility of this phase. On the other hand, a vice versa situation exists for DP600-B. Therefore, it could be reasoned that the martensite phase in DP600-A is likely to deform more compatibly with the matrix and will likely result in failure after significant elongation of the phase. Also, the decohesion at the ferrite/martensite interface will be significantly reduced. For DP600-B, however, a strong martensite phase with reduced ductility will likely lead to interfacial decohesion and cracking. Evidence of this for both materials has been reported in this paper and is in conformity with the recent work of Mazinani and Poole [15]. Steinbrunner et al. [19] investigated a dual phase steel with 0.08% of C and observed the nucleation of voids at fractured martensite particles at thickness strains as low as 0.05 (equivalent strain 0.1). They proposed the localized deformation within the martensite as a new distinctive void nucleation mechanism. In the intercritically annealed dual phase steels with similar carbon contents as in this work, Baburamani et al. [35] reported voids nucleation at the ferrite/martensite interface at remarkably low plastic strains of 0.05–0.1. It has to be noticed that in general the martensite (austenite) volume fraction is increasing with an increase in the intercritical annealing temperature. Also, the martensite hardness as a function of carbon content might be higher

16

G. Avramovic-Cingara et al. / Materials Science and Engineering A 516 (2009) 7–16

with increasing of this temperature. So, there is a possible impact of C content within the martensite in the case of drastically different intercritical annealing temperatures, which is not the case here. Now, the question as to why steel with more carbon does not show a higher ultimate tensile strength, or in other words, why should the carbon content influence the martensite only and not the other material properties? Bortsov and Fonshteyn [36] found that redistribution of the strains between the ferrite and martensite results in nucleation of voids at the early stage of deformation. The lack of any influence of carbon content within the martensite on the ultimate tensile strength of the material is a consequence of redistribution of the strains between the phases. Third factor of consequence to be considered here is the location and orientation of martensite bands with respect to the applied stress. Goods and Brown [38] reported that the particles whose major axes lie closest to the tensile axis will fracture preferentially, according to so called fibre loading mechanism. Lindley et al. [39] found that there was an increasing incidence of carbide cracking with decreasing angle to the tensile axis. This supports our results where the martensite centre-line band in the DP600-A steel, located in the middle of the sheet thickness, and in the direction of tensile stress, had the most prominent internal failure. Finally, it appears that the higher carbon martensite of DP600-B steel is less plastic, has favourable morphology than the lower carbon martensite, and thus has a higher tendency for void nucleation at the interfaces. It is to be noted that the present study is based on uniaxial tension. Damage would be even more significant under biaxial tensile strain paths. 5. Conclusions Present study indicates that dual phase steel such as DP600 from two different sources exhibits significantly different damage mechanisms even when their uniaxial tensile material properties, in terms of strength and workhardening, are nearly identical. The results emphasize that composition, processing and microstructure can play a significant role in determining the formability of these materials. (1) DP600-A steel microstructure exhibits evident martensite banding at the centre-line of the sheet thickness. The nucleation of voids occurs by localized cracking of martensite, by decohesion at the ferrite–martensite interface, and by separation of adjacent martensite regions. As per the evidence of the SEM in situ tensile tests, martensite cracking starts initially at local strain of 0.029 and void nucleation by ferrite/martensite decohesion at the strain of 0.090. An accelerated void growth at martensite centre-line was revealed, as opposed to the areas located away from the sheet centre. The preferable void growth and coalescence along the transverse plane normal to the applied load was recorded in the localized neck region. Damage process resulted in fracture surface morphology exhibiting very large and deep dimples related to martensite. (2) The quantitative damage analysis of DP600-B steel suggests that the void nucleation occurs continuously during the entire deformation process with an almost constant rate and this rate reduces before fracture. The major void nucleation mechanism is decohesion at the ferrite/martensite interface, and those voids grow along ferrite grain boundaries. Fewer voids are nucleated at cracked martensite particles, and very small number of large voids is nucleated on inclusions. A nucleation strain of 0.15 has been estimated for this material. This steel exhibited typical ductile fracture characteristics.

(3) The initial microstructure had an effect on the damage accumulation rates of both DP600-A and B steels. There is an initial slow increase in total damage, followed by a rapid increase in damage accumulation rate immediately prior to fracture. Damage is more rapid in the DP600-A with overall damage evolution mainly due to the growth of voids previously nucleated at martensite centre-line and secondary nucleation of small voids in the areas away from the centre at higher plastic strains. The higher carbon martensite in DP600-B steel is less plastic than the lower carbon martensite of DP600-A steel, and thus has a higher tendency for void nucleation at the interfaces. Difference in martensite morphology and carbon content of these DP600A and DP600-B steels thus significantly influence variations in their damage and fracture behaviour. Acknowledgement The authors gratefully acknowledge the financial support from the Canadian Auto21 Network of Centers of Excellence. References [1] R.G. Davies, Metall. Trans. 9A (1978) 41. [2] G. Thomas, J.Y. Koo, in: R.A. Kot, J.W. Morris (Eds.), Structure and Properties of Dual-phase Steels, TMS-AIME, Warrendale, PA, 1979, p. 183. [3] N.J. Kim, G. Thomas, Metall. Trans. 12A (1981) 483. [4] A.R. Marder, Metall. Trans. 13A (1982) 85. [5] A.H. Nakagawa, G. Thomas, Metall. Trans. 16A (1985) 831–840. ˇ [6] L. Sidjanin, S. Miyasato, Mater. Sci. Technol. 5 (1989) 1200. [7] D. Suh, D. Kwon, S. Lee, N.J. Kim, Metall. Trans. 28A (1997) 504. [8] A. Bag, K.K. Ray, E.S. Dwarakadasa, Metall. Trans. 30A (1999) 1193. [9] S. Kim, S. Lee, Metall. Trans. 12A (2000) 483. [10] H.S. Lee, B. Hwang, S. Lee, C.G. Lee, S.J. Kim, Metall. Trans. 35A (2004) 2371. [11] X.J. He, N. Terao, A. Berghezan, Metal Sci. 18 (1984) 367. [12] R. Stevenson, in: A.T. Davenport (Ed.), Formable HSLA and Dual Phase Steels, TMS-AIME, Warrendale, PA, 1977, p. 99. [13] D.A. Korzekwa, R.D. Lawson, D.K. Matlock, G. Krauss, Scripta Metall. 14 (1980) 1023. [14] A.M. Sarosiek, M. Grujicic, W.S. Owen, Scripta Metall. 18 (1984) 353. [15] M. Mazinani, W.J. Poole, Metall. Trans. 38A (2007) 328. [16] M. Mazinani, W.J. Poole, Adv. Mater. Res. 15–17 (2007) 774. [17] M. Erdogan, S. Tekeli, Mater. Des. 23 (2002) 597. [18] M. Erdogan, J. Mater. Sci. 37 (2002) 3623. [19] D.L. Steinbrunner, D.K. Matlock, G. Krauss, Metall. Trans. 19A (1988) 579. [20] M. Sarwar, R. Priestner, J. Mater. Sci. 31 (1996) 2091. [21] S.K. Han, H. Margolin, Mater. Sci. Eng. A112 (1989) 133. [22] D. Boyd, Q. Poruks, I. Yakubsov, in: W.O. Soboyejo, J.J. Lewandowski, R.O. Ritchie (Eds.), The John Knott Symposium, TMS, Warrendale, PA, 2002, p. 187. [23] P. Poruks, I. Yakubtsov, J.D. Boyd, Scripta Mater. 54 (2006) 41. [24] G. Liu, J.D. Embury, F.E. Hassani, in: W.O. Soboyejo, J.J. Lewandowski, R.O. Ritchie (Eds.), The John Knott Symposium, TMS, Warrendale, PA, 2002, p. 137. [25] A.F. Szewczyk, J. Gurland, Metall. Trans. 3A (1982) 1821. [26] E. Almeida, M. Morcillo, Surf. Coat. Technol. 124 (2000) 180. [27] ASTM E8-04, Standard test methods of tension testing of metallic materials (2004). [28] A.K. De, J.G. Speer, D.K. Matlock, Adv. Mater. Process. (February) (2003) 27. [29] G. Avramovic-Cingara, Ch. A.R. Saleh, M.K. Jain, D.S. Wilkinson, Metallurgical and Materials Transactions A, (2009), manuscript under review. [30] Northern Eclipse v.6.0, Operating Manual, Empix Imaging Inc. (2001). [31] Y. Ososkov, D.S. Wilkinson, M.K. Jain, T. Simpson, Int. J. Mater. Res. (formerly Zeitschrift für Metallkunde) 98 (8) (2007) 664. [32] Y. Ososkov, K. Inal, M.K. Jain, D.S. Wilkinson, K.W. Neale, SAE Technical Paper Series (SP-1953) (2005) 1. [33] G.R. Speich, R.L. Miller, in: R.A. Kot, J.W. Morris (Eds.), Structure and Properties of Dual-phase Steels, TMS-AIME, New York, NY, 1979, p. 145. [34] D.Z. Yang, E.L. Brown, D.K. Matlock, G. Krauss, Metall. Trans. 16A (1985) 1385. [35] P.S. Baburamani, R.A. Jago, R.M. Hobbs, ICSMA 6 Int. Conf., vol. 1, Melbourne, Australia, 1982, p. 115. [36] A.N. Bortsov, N.M. Fonshteyn, Phys. Met. Metall. 61 (2) (1986) 74. [37] S. Lee, K.S. Sohn, I.M. Park, K. Cho, Met. Mater. 1 (1995) 37. [38] S.H. Goods, L.M. Brown, Acta Metall. 27 (1979) 1. [39] T.C. Lindley, T. Oates, C.E. Richards, Acta Metall. 8 (1970) 1127. [40] W. Tong, J. Mech. Phys. Solids 46 (10) (1998) 2087. [41] M. Jain, D.J. Loyd, S.R. Macewen, Int. J. Mech. Sci. 38 (2) (1996) 219.