Sie sind auf Seite 1von 16

Available online at www.sciencedirect.

com

Acta Materialia 59 (2011) 34313446 www.elsevier.com/locate/actamat

Deformation behaviour of commercially pure titanium at extreme strain rates


N.P. Gurao a, Rajeev Kapoor b, Satyam Suwas a,
a

Department of Materials Engineering, Indian Institute of Science, Bangalore 560 012, India b Materials Group, Bhabha Atomic Research Centre, Mumbai 400 085, India

Received 21 June 2010; received in revised form 30 December 2010; accepted 14 February 2011

Abstract The evolution of microstructure and texture during room temperature compression of commercially pure Ti with four dierent initial _ = 3 104 s1 all the dierent initial orientations were studied under quasi-static and dynamic loading conditions. At a low strain rate e textures yielded the same end texture, despite dierent microstructural evolution in terms of twin boundaries. High strain rate deforma_ = 1.5 103 s1 was characterized by extensive twinning and evolution of a texture that was similar to that at low strain rate with tion at e minor dierences. However, there was a signicant dierence in the strength of the texture for dierent orientations that was absent for low strain rate deformed samples at high strain rate. A viscoplastic self-consistent model with a secant approach was used to corroborate the experimental results by simulation. 2011 Published by Elsevier Ltd. on behalf of Acta Materialia Inc.
Keywords: Titanium; Texture; Twinning; Electron backscatter diraction

1. Introduction Hexagonal materials form a class of low symmetry materials that are extensively used in mechanical and structural applications. The deformation response of hexagonal close-packed (hcp) metals and alloys is generally guided by their axial (c/a) ratio [1]. The deformation behaviour of hcp materials is, however, very complex, due to activation of dierent type of slip as well as twinning systems under different conditions. The onset of twinning during the initial stages of deformation leads to a unique hardening response of these materials. Crystallographic texture also plays an important role in deciding the mechanical behaviour of hcp metals and alloys. Amongst the hcp metals, the majority of investigations have been directed at titanium [211] and zirconium [1219], due to their technological importance. In particular, titanium and its alloys nd extensive use in aerospace and biomedical applications due to their
Corresponding author. Tel.: +91 80 22933245; fax: +91 80 23600472.

E-mail address: satyamsuwas@materials.iisc.ernet.in (S. Suwas).

excellent mechanical and physical properties, like high specic strength, good ductility and excellent corrosion resistance [2]. Hence, it has been the subject of detailed investigation from the fundamental as well as engineering points of view. Various characterization techniques, such as electron backscatter diraction (EBSD) [20] and X-ray [21] and neutron diraction [22], have been employed to fully understand the evolution of microstructure and texture during dierent deformation processes. Attempts have been made to simulate the experimental stressstrain curves and texture evolution using various crystal plasticity models ranging from the simple Taylor model to the advanced crystal plasticity nite element (CPFEM) model. The mechanical response of titanium has been studied extensively for dierent deformation modes [37]. A good number of studies have focussed on the strain hardening characteristics of titanium during deformation [810]. Modelling texture evolution in hcp metals has been an intriguing problem due to a propensity for twinning during and co-workers [12,13] employed the deformation. Tome

1359-6454/$36.00 2011 Published by Elsevier Ltd. on behalf of Acta Materialia Inc. doi:10.1016/j.actamat.2011.02.018

3432

N.P. Gurao et al. / Acta Materialia 59 (2011) 34313446

volume fraction transfer (VFT) scheme of Van Houtte [23] and proposed the predominant twin reorientation (PTR) scheme to model the deformation behaviour of zirconium. Kalidindi and co-workers [10,11], on the other hand, employed a Taylor-type rate sensitive crystal plasticity model to simulate the deformation behaviour of Ti. The eect of initial orientation has been investigated extensively, with most studies being carried out on two dierent orientations, namely the through thickness orientation with strong basal texture and the in-plane orientation with prismatic texture [11,19]. It has been observed that the two extremely dierent initial textures lead to dierent mechanical responses. The basal orientation shows a higher yield strength but a lower strain hardening rate, while the prismatic orientation shows lower yield strength and a higher strain hardening rate. Crystal plasticity models have been able to predict dierent mechanical responses with respect to the initial orientation. The mechanical response of dierent initial orientations can be indirectly inuenced by the value of stacking fault energy (SFE), which is dierent for dierent planes in hcp metals like titanium (300 mJ m2 in the basal plane and 150 mJ m2 in the prismatic plane) [24]. The deformation of titanium by virtue of its less than ideal c/a ratio (1.588) is contributed by prismatic slip. However, since the basal and the prismatic slip systems provide only four independent slip systems, plastic deformation in Ti has to be accommodated by hc + ai slip or twinning. The major twinning systems that can operate in compression of a single crystal Ti are {1 0  1 2}h1 0  1 1i, {1 1  2 2}h1 1  2 3i and {1 1  2 1}h 1 1 2 6i. The choice of the twinning system depends on the initial orientation [5]. Twinning, being a polar transformation, is sensitive not only to the magnitude of shear but also to the direction of shear [25]. Depending on the loading direction, dierent twin systems become active and alter the deformation response, as well as the texture evolution, in Ti. A detailed study of the strain hardening behaviour of polycrystalline titanium under compression, plane strain compression and torsion has been carried out by Salem et al. [810]. The authors [8] categorized three distinct stages in the stressstrain curve, labelled A, B and C. The rst regime, stage A, was characterized by a decreasing strain hardening rate, similar to the dynamic recovery regime observed in high stacking fault energy (SFE) metals. This was followed by stage B, a region with increasing strain hardening rate in both plane strain and uniaxial compression. Stage C exhibited a decreasing strain hardening rate. During stage A, deformation was accommodated only by slip. The activation of cross slip between the basal and prism planes with a common slip direction leads to easy dynamic recovery and, hence, a lower strain hardening rate. The pyramidal hc + ai slip system, being considerably harder than the prism or basal slip systems, could be activated to make up the ve independent slip systems. In stage B, twinning led to an increase in strain hardening rate as twin activity increased with increasing strain. As the twin systems oper-

ative in hcp materials are non-coplanar with the slip systems, a considerable amount of hardening occurred during stage B. The decreasing strain hardening rate in stage C was explained due to increasing diculty in producing deformation twins with further straining. This theory proposed by the authors [8] was further supported by simulations obtained from a Taylor-type crystal plasticity model by Wu et al. [11]. However, most of these studies did not consider the eect of strain rate on the strain hardening response and texture evolution. Such studies are of paramount importance, as twinning and additional mechanisms such as shear banding are expected to play a very important role in the deformation behaviour of hexagonal materials at high strain rates. The present work is an attempt to complement the ndings of Salem et al. [8] for the quasi-static deformation of titanium and to explore the validity of these concepts in a dynamic strain rate regime. 2. Experimental Cylindrical samples of 6 mm diameter and 9 mm height were obtained from a rolled block of commercially pure (CP) Ti plate with a strong basal texture. Samples were machined in four dierent orientations (Fig. 1a), namely IIV, to ensure dierent starting textures. Samples IIII were machined along the RD, ND and TD directions of the rolled block while sample IV was machined at a 45 angle to the RDND plane. Each of these samples was subjected to compression testing at a strain rate of 3 104 s1 in a servo-hydraulic DARTEC mechanical testing set-up to a true strain of e = 0.36. The high strain rate tests at 1500 s1 were carried out using a split Hopkinson pressure bar (SHPB) with incident and transmission bars of 13 mm diameter and 1300 mm length. Cylindrical specimens of all four orientations with diameters and heights of 6 mm were used for the test. A MoS2 lubricant was used between the sample and the bar interfaces. For a general discussion see the relevant ASM Handbook [26]. The samples were then sectioned along the compression direction and subjected to metallographic preparation with electropolishing in a solution of 60 ml of perchloric acid in 600 ml of methanol and 400 ml of butycellosolve. The samples were etched with Krolls reagent for optical microscopy, while EBSD studies were carried out by eld emission gun scanning electron microscopy (FEG-SEM) (SIRION). A step size of 1 lm was used to capture large area scans for the starting and deformed samples. High resolution EBSD measurements were done using a step size of 50 nm for the samples deformed at high strain rate. Data acquisition and analysis were carried out using TSL software version 5.2. 3. Viscoplastic self-consistent simulations In the present investigation viscoplastic self-consistent simulation code VPSC-6 was used to simulate the experimental texture. The VPSC model is described in detail in

N.P. Gurao et al. / Acta Materialia 59 (2011) 34313446

3433

II
ND (z) TD (y) RD (x)

III

I IV

(a)
10 1 0 10 1 0

II

0002

21 10

0002

21 10

10 1 0

10 1 0

III

IV

0002

21 10

0002

21 10

(b)
Fig. 1. (a) Geometry of samples obtained from a Ti plate with initial strong basal texture and (b) corresponding compression direction (CD) inverse pole gures for samples with dierent orientations.

et al. [27] and the PTR scheme in Lebensohn and Tome [13]. The texture measured by EBSD was discretized Tome to obtain 2000 single orientations, that were used as input to the model. The critical resolved shear stress (CRSS) and hardening parameters for Voce type hardening were obtained by tting the stressstrain curve for sample II, for which minimum twin activity is observed. Wu et al. [11] obtained these values from shear tests on titanium, wherein twinning is minimal, from the hardening curves, as against stressstrain curves in the present case. In addition to extended Voce hardening (for statistical dislocations), hardening due to the presence of twin barriers is superimposed [13]. We have adopted the uncoupled approach as against the coupled grain approach used by Proust et al. [18]. In this approach the increase in twin activity and subsequent saturation with the amount of deformation in Zr was modelled by using negative hardening parameters for twinning. However, they followed this strategy to model up to a strain of 0.3. As a consequence, with the increase in amount of strain there was an increase in twinning activity. This is not the case for titanium, where it has been shown experimentally that twinning is replaced by pyramidal slip at higher strains. In the present work the minimum volume fraction of twins in a grain is assumed to be 0.1, while a saturation value of 0.5 has been used in the simula-

tions. The activation of any particular slip system as well as twin system is a strong function of initial orientation and, hence, dierent samples with dierent orientations show dissimilar stressstrain and strain hardening responses. In the present investigation, microstructural evidence of twin boundaries was used to ne tune the model. The fraction of the twin boundaries obtained from orientation imaging microscopy (OIM) by EBSD was used as an indication of the activity of that particular twin system. A higher fraction of a particular type of twin boundary indicates a higher activity of that particular twin system. 4. Results 4.1. Initial texture Inverse pole gures corresponding to the compression direction (CD) representing the initial texture of the starting samples as measured from large area EBSD scans are shown in Fig. 1b. The dierently oriented Ti samples extracted from the rolled plate ensured that the initial texture was signicantly dierent for each sample. Sample I, cut along RD, showed a majority of orientations along the h1 0  1 0i and h2  1 1 0i lines, with maximum intensity   at h3 1 2 0i, while sample II, cut along ND, had a strong near basal texture with a maxima for the h2  1 1 4i compo-

3434

N.P. Gurao et al. / Acta Materialia 59 (2011) 34313446

nent. Sample III, cut along TD, was characterized by a spread of orientations along the h1 0  1 0i h 2  1 1 0i line. The texture was relatively weaker and showed higher intensities for the h6  3 3 2i and h1 0  1 0i orientations. For sample IV, cut at 45 to the NDRD plane, most of the crystallites were oriented towards the h2  1 1 0ih1 0  1 0i region of the inverse pole gure with a maximum intensity at h1 0  1 0i . 4.2. Mechanical behaviour The plastic region of the true stressstrains curve obtained from the compression tests at low strain rate, shown in Fig. 2a, indicates that the dierently oriented samples have distinctly dierent responses. The yield strength of sample II was higher than that of the other three samples and also showed a low strain hardening rate. In the case of sample I the true stress increased steadily with strain up to e = 0.15 and then increased steeply to reach a maximum value amongst all the studied samples at e = 0.36. The stress in the case of sample II increased gently to reach a value slightly less than that for sample I. Samples III and IV showed almost the same true stress at e = 0.36, which was lower than that for sample II. The stressstrain curve of sample IV coincided with the stressstrain curve of sample I up to e = 0.15 and then deviated to coincide with the sample III curve, giving the same stress at e = 0.36. The samples tested at high strain rate showed an increase in stress level at a given strain com-

r0 =d e normalized strain hardening rate d rG versus normal  rr0 ized stress for the low and high strain rate G

pared with the low strain rate deformed samples. The increase in yield strength was lower for samples I and II, while a substantial increase was observed for samples III and IV. However, the nature of the stressstrain curve remained similar for sample II at low and high strain rates, showing minimum strain hardening compared with the other three samples. In order to study the dependence of orientation on the strain hardening behaviour of titanium the strain hardening curves were calculated numerically. The strain hardening rate and stress were normalized with respect to the shear modulus of titanium (44   GPa) and the

deformed samples are plotted in Fig. 2b. As expected, the hardening curves showed the three regimes of strain hardening observed by Salem et al. [8]. Samples I, III and IV deformed at low strain rate showed the three stages of strain hardening quite distinctly, while sample II showed dominant stages A and C. The onset of stages B and C was dierent for these samples. Amongst these samples sample III showed the onset of stage B at higher values of strain, indicating late onset of twinning for this texture. Sample II with a strong basal texture was markedly distinct, with stage A dominant to large strains, indicating an absence of twinning. Also, the slope of the hardening curve showed the minimum strain hardening rate for sample II. Another important observation is that the hard-

1000

1000

LSR
800
800

HSR
Stress (MPa)
600 400 200 0

Stress (MPa)

600 400 200 0

I II III IV
0 0.1 0.2 0.3 0.4

I II III IV
0 0.1 0.2 0.3 0.4

Strain

(a)
0.12

Strain

Normalized Strain Hardening Rate

Normalized Strain Hardening Rate

0.06

LS R
0.05 0.04 0.03 0.02 0.01 0 0 0.005 0.01 0.015

I II III IV

HSR
0.1 0.08 0.06 0.04 0.02 0 0 0.005 0.01 0.015 0.02

I II III IV

0.02

0.025

Normalized Stress

(b)

Normalized Stress

Fig. 2. (a) True stressstrain curves and (b) hardening curves for samples IIV tested at low and high strain rates, respectively.

N.P. Gurao et al. / Acta Materialia 59 (2011) 34313446

3435

ening curve for sample II was shifted to the right, as it had got a higher yield point but a lower strain hardening rate due to the absence of twinning. The hardening curves for the samples tested at the high strain rate were similar to those at the low strain rate, however, there were some important distinctions. A higher hardening rate was observed at the high strain rate, in addition to a higher stress level. However, the dierence between various samples disappeared and the hardening curves of samples I, III and IV almost coincided. The strain hardening curves for all four orientations were similar, unlike at the low strain rate, where stage B is absent in sample II. 4.3. Evolution of texture Despite a marked dierence in the initial texture of the four samples, the overall texture evolution in all the cases was near basal at low strain rate (Fig. 3a). In addition, there was a clustering of orientations near h2  1 1 4i for all four samples. Some crystallites were oriented near the h2  1 1 0i corner of the inverse pole gure. A spread of ori-

entations from h2  1 1 4i to h1 0  1 4i was observed in samples I, II and IV, while sample III shows a spread of orientations along the h0 0 0 1ih2  1 1 0i line. The reorientation of crystallites towards the basal orientation was dominant in samples I and IV, while there was little change in texture of sample II, which had a strong basal texture. Although the texture in sample II was dierent from the remaining three samples, it is believed that with increasing strain it will also form the near basal h2  1 1 4i orientation, which is a stable end orientation. At a high strain rate, there was a slight weakening of texture and an additional, however weak, h1 0 1 0i component appeared that was absent from the samples deformed at a low strain rate (Fig. 3b). All the samples showed a texture characterized by the presence of orientations near the h2  1 1 0i and   h2 1 1 4i components. There was a spread of orientations from these components towards the line joining the lines h2  1 1 0ih1 0  1 0i and h0 0 0 1ih1 0  1 0i. It can be observed that the high strain rate deformed samples manifest a higher strength of the h2  1 1 0i component compared   with the h2 1 1 4i component, which is dominant at the low

10 1 0

10 1 0

II

0002

21 10

0002

21 10

21 14
10 1 0 10 1 0

IV

III

0002

21 10 10 1 0

0002

21 10 10 1 0

(a)
I II

0002

21 10 10 1 0

0002

21 10 10 1 0

III

IV

0002

21 10

0002

21 10

(b)
Fig. 3. Compression direction (CD) inverse pole gure showing nal texture after deformation of samples IIV at (a) low strain rate (LSR) and (b) high strain rate (HSR).

3436

N.P. Gurao et al. / Acta Materialia 59 (2011) 34313446

strain rate. Another important observation was the spread of orientations towards h1 0  1 0i from h2  1 1 0i in the high strain rate deformed samples. The texture was stronger for samples I and II while a weaker texture evolves in samples III and IV. 4.4. Evolution of microstructure Optical microscopy as well as scanning electron microscopy with EBSD was used to study evolution of microstructure during deformation and its correlation with the macroscopic hardening behaviour. The inverse pole gure (IPF) maps for the four dierent samples deformed at a low strain rate are shown in Fig. 4. It is observed that samples I, III and IV show profuse twinning while very few twins are observed in sample II. Fig. 5a shows a repre-

sentative grain boundary map depicting the high angle boundaries and the twin boundaries for sample IV deformed at low strain rate. It can be seen that only one type of twin system is operative in a grain of particular orientation. Extension twinning (green boundary) is operative in grains having their orientation away from the basal orientation, while contraction twinning (black boundary) is dominant in grains with a near basal orientation. It should be mentioned here that twinning is not observed in all grains of the above mentioned orientations. Another important observation is the absence of secondary twinning in the deformed sample. In titanium, the extension twin boundary is characterized by a 85 misorientation about the h2  1 1 0i axis while the contraction twin boundary is characterized by a misorientation of 65 about the h1 0  1 0i axis. Fig. 5b shows the axisangle distribution

Fig. 4. Inverse pole gure maps for deformed samples (a) I, (b) II, (c) III and (d) IV. The horizontal direction corresponds to the compression axis.

N.P. Gurao et al. / Acta Materialia 59 (2011) 34313446

3437

Misorientation Angle
0.6 0.5

Number Fraction

0.4 0.3 0.2 0.1 0.0 10 20 30 40 50 60 70 80 90 100

Misorientation Angle [degrees]

(b)
Fig. 5. (a) Blank map showing twin boundaries in titanium sample IV deformed at low strain rate. Reorientation of the crystallite due to twinning is shown. Contraction twinning is active in grains near the basal orientation while extension twinning operates in grains away from the basal orientation. (b) Axisangle pair showing that low angle boundaries have all possible orientations (orange colour) while the contraction twin is 65 about h1 0  1 0i and extension twin is 85 about h2  1 1 0i for sample IV deformed at low strain rate. (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)

for sample IV deformed at low strain rate. In this case the deformed microstructure is characterized by a high fraction of low angle grain boundaries (orange colour) with their axes spread throughout the inverse pole gure. The misorientation of 65 shows a clustering of axes near the h1 0  1 0i position of the inverse pole gure while the misorientation of 85 shows axes clustered near h2  1 1 0i. This is a clear indication that these are twin boundaries and, hence, can be used as a quantitative measure of twinning activity. EBSD measurements were carried out on high strain rate deformed samples using two step sizes, namely 1 lm and 50 nm. The data obtained from the large step size mea-

surements were used to calculate texture, while data obtained from the small step size scans were used to generate microstructural features that reveal the micro-mechanisms operative during high strain rate deformation. Optical micrographs of the samples show extensive twinning and grain fragmentation in the high strain rate deformed samples compared with the low strain rate deformed samples (Fig. 6). This eect is particularly dominant for sample II with a near basal orientation. A representative IPF map of sample III shown in Fig. 7 clearly indicates copious extension twins formed in a grain at a high strain rate.

3438

N.P. Gurao et al. / Acta Materialia 59 (2011) 34313446

Fig. 6. Optical micrographs of the samples deformed at (a) low strain rate and (b) high strain rate. Samples deformed at high strain rate are characterized by conspicuous twinning and grain fragmentation.

The fraction of twin boundaries pertaining to the two major twinning modes, namely the {1 0  1 2}h 1 0 1 1i    extension twin and {1 1 2 2}h1 1 2 3i contraction twin, were determined for samples deformed at a low strain rate (Fig. 8a). The EBSD scans were used to estimate grain average misorientation (GAM) for all samples.

Table 1 shows the GAM values for initial as well as deformed samples. It can be observed that the GAM value increases with deformation for all four samples. For the low strain rate deformed samples the highest intragranular misorientation build-up in terms of GAM is for sample II. The increase in GAM is much higher

N.P. Gurao et al. / Acta Materialia 59 (2011) 34313446

3439

Misorientation (degree)

120 100 80 60 40 20 0 0 2 4 6 8 Distance (micron)

Point to Point Point to Origin

10

12

Fig. 7. Inverse pole gure map of sample III with a step size of 50 nm showing multiple twinning in a single grain.

0.35 Contraction Twinning

Twin Boundary Fraction

0.3 0.25 0.2 0.15 0.1 0.05 0

Extension Twinning

Table 1 Variation in grain average misorientation () with deformation at low and high strain rates for samples with dierent orientation. Condition Sample I Start Low strain rate High strain rate 0.94 1.53 2.44 II 0.95 1.69 2.36 III 0.68 1.46 2.58 IV 0.82 1.28 2.51

II Sample

III

IV

(a)
0.45 0.4 0.35 Contraction Twinning Extension Twinning

in high strain rate deformed samples can be attributed to grain fragmentation, which is operative under dynamic loading conditions. 4.5. Simulations The compression direction (CD) inverse pole gures for the samples deformed at low and high strain rates as obtained by VPSC simulations are shown in Fig. 9. The simulated textures show a reasonable match with the experimental textures. There is a better agreement between the experimental and simulated textures for samples deformed at the low strain rate compared with the texture of samples deformed at the high strain rate. While the agreement between the experimental and simulation results is good for samples I, III and IV, only the major texture component is reproduced by simulations for sample II. The twin activity obtained from the simulation results (Fig. 8b) is in accordance with the twin boundary fraction obtained from EBSD (Fig. 8a). For the samples deformed at low strain rate a reasonable match has been observed between the experimental and simulated inverse pole gures, however, some deviation could be noticed in the stressstrain response (Fig. 10).

Twin Activity

0.3 0.25 0.2 0.15 0.1 0.05 0

II Sample

III

IV

(b)
Fig. 8. (a) Fraction of dierent twin boundaries obtained from an EBSD scan and (b) twin activity estimated from simulation results for dierent samples deformed at low strain rate.

at high strain rate than at low strain rate and the highest GAM value is obtained for sample IV, while sample II shows the lowest value. The higher GAM values obtained

3440

N.P. Gurao et al. / Acta Materialia 59 (2011) 34313446


10 1 0
10 1 0

II

0002

21 10

0002

21 10

10 1 0

10 1 0

III

IV

0002

21 10

0002

21 10

(a)
10 1 0 10 1 0

II

0002

21 10

0002

21 10

10 1 0

10 1 0

III

IV

0002

21 10

0002

21 10

(b)
Fig. 9. Simulated compression direction (CD) inverse pole gures for samples IIV deformed at (a) low strain rate and (b) high strain rate.

An increase in the propensity to twin was observed at the high strain rate and this has been modelled by increasing the CRSS for the slip system, keeping that of the twin systems constant. No tting was employed to match the stressstrain curves of the samples deformed at high strain rate, as it increases the number of variables in the model. Instead, it was assumed that the CRSS of all three slip systems increased by the same percentage (25%) while that of the two twin systems remained unchanged. This strategy reduced the number of variables in the model

but was still able to capture the experimentally observed increase in twin activity at the high strain rate. It should be mentioned here that the chosen CRSS values for the high strain rate yielded higher stress values for sample II and lower ones for the other samples. Although the simulated texture for sample II does not exactly match the experimental results, the parameters were not subjected to any tting. Owing to the physical processes occurring in the material at high strain rate that are not captured in the VPSC simulation, for example grain frag-

N.P. Gurao et al. / Acta Materialia 59 (2011) 34313446


900 800 700 600 500 400 300 200 100 0 0 900 True Stress (MPa) 800 700 600 500 400 300 200 100 0.05 0.1 0.15 0.2 True Strain 0.25 0.3 0.35 0 0 0.05 0.1 0.15 0.2 True Strain 0.25 0.3 0.35 Expt. Simu.

3441

True Stress (MPa)

Expt. Simu.

II

900 800 700 600 500 400 300 200 100 0 0

900 True Stress (MPa) 800 700 600 500 400 300 200 100 0 0.05 0.1 0.15 0.2 True Strain 0.25 0.3 0.35 Expt. Simu.

True Stress (MPa)

Expt. Simu.

II (a)

IV
0 0.05 0.1 0.15 0.2 True Strain 0.25 0.3 0.35

1200 1000 True Stress (MPa) 800 600 400 Expt. 200 0 0 0.05 0.1 0.15 0.2 True Strain 0.25 0.3 0.35 Simu.

1200

1000 True Stress (MPa) 800 600 400 200 0 0

II

Expt. Simu. 0.05 0.1 0.15 0.2 True Strain 0.25 0.3 0.35

1200 1000 True Stress (MPa) 800 600 400 Expt. 200 0 0 0.05 0.1 0.15 0.2 True Strain 0.25 0.3 0.35 Simu.

1200

III
1000 True Stress (MPa) 800 600 400

IV

Expt. 200 0 Simu.

(b)

0.05

0.1

0.15 0.2 True Strain

0.25

0.3

0.35

Fig. 10. Simulated and experimental stressstrain curves for the dierent samples deformed at (a) low strain rate and (b) high strain rate.

mentation, such a deviation is not surprising. This hampers the exact reproduction of texture by simulations for the high strain rate deformed samples. A typical instance is that the simulations were unable to reproduce the h1 0  1 0i component. The various parameters used for the simulation (Voce hardening law) are listed in Table 2.

5. Discussion Despite four dierent initial orientations, the deformation textures after compression at low and high strain rates were similar in terms of clustering of poles near the basal orientation. The only exception was the occurrence of an

3442

N.P. Gurao et al. / Acta Materialia 59 (2011) 34313446

Table 2 Voce hardening parameters for dierent slip and twin systems in titanium used in the present investigation. Slip/twin mode Basal Prismatic Pyramidal Extension twinning Contraction twinning s0 200/250 80/100 750/937.5 200 380 s1 5 20 5 220 120 H0 2 1200 12 1500 1000 h1 1 650 0.4 450 450

nent twinning. In addition, the propensity for cross slip is expected to reduce at high strain rates [28]. Therefore, in addition to twinning, restricted cross slip also contributes to hardening at high strain rates. 5.2. Texture evolution The evolution of a texture comprising components close to the basal orientation plus a clustering of orientations near the h2  1 1 0i component is a common observation for all four initial orientations, at the low as well as high strain rate. High strain rate deformation led to very weak texture in samples III and IV, but strong texture in samples I and II. The high strain rate deformed samples exhibited higher intragranular misorientations compared with the low strain rate deformed samples, as displayed by the trends in GAM. Canova et al. [29] predicted a weakening of texture during uniaxial tension or compression at high strain rates. This was attributed to the increased number of active slip systems that reduce plastic spin, thereby retarding texture evolution. The relation between the lattice spin (X), rigid body or material spin (b) and plastic spin (x) is given by [30]: Xbx Lij Lji 2 ;f s: f nf X ns X ms ij mji _ s; f xij c 2 f 1 s 1

hs1-s2-s3 = 1; hs-ttw = 1; hs-ctw = 1; httw-ttw httw-ctw = 1; httw-ctw hctw-ttw = 40.

additional component h1 0  1 0i at the high strain rate. The factors governing the evolution of texture and microstructure at the two extreme strain rates are discussed in the following sections. 5.1. Strain hardening response The stressstrain curves for dierently oriented samples clearly indicated anisotropy in mechanical properties. Notable among these properties are the yield strength and strain hardening exponent. The responses of the dierently oriented titanium samples with regard to these properties were dierent for the two strain rate regimes. 5.1.1. Deformation at lower strain rates The stressstrain curves for the low strain rate deformation can be categorized into two types: (i) one with a higher yield strength but lower hardening rate (sample II) and (ii) the others with a lower yield strength but higher strain hardening rate (samples I, III and IV). The strain hardening curves (Fig. 2b) for all samples showed three distinct stages of hardening, namely stages A, B and C. Stage B, which is attributed to twinning, was minimal in sample II. The absence of twinning led to a lower strain hardening rate for sample II. In addition, owing to the higher SFE for the basal plane compared with the prismatic planes and the same Burgers vector of dislocations, cross slip from basal to prismatic planes is easier. This led to recovery and, therefore, a lower strain hardening rate for sample II. On the other hand, there is restricted cross slip from the low SFE prismatic to basal plane. Therefore, the strain hardening for samples I, III and IV, with non-basal orientations, is contributed by limited cross slip in addition to twinning. 5.1.2. Deformation at high strain rates At the high strain rate the hardening curves of samples I, III and IV were very close to each other while that of sample II was shifted to the right at higher stresses. The extent of shift in the hardening curve at high strain rate for sample II from the other samples was not, however, as much as in the case of low strain rate deformation of this sample. In sample II stage B is observed in the strain hardening curve for high strain rate deformation, which is an indication of increased twin activity. The optical micrographs of all the high strain rate deformed samples (Fig. 6) showed promi-

bij

Here L is the velocity gradient, m is the Schmid tensor, nf is the slip system family and ns is the individual slip system belonging to one family. For tension and compression b = 0, therefore, X decreases with a decrease in x. For torsion b 0, hence, X increases with a decrease in x. In the present investigation, texture weakened only for the samples III and IV, while a strong texture component h2  1 1 0i evolved in samples I and II at the high strain rate. Thus, initial texture plays an important role in determining the texture at high strain rates. The theoretical analysis provided by Canova et al. [29] is very simplied and does not necessarily hold good for low symmetry hexagonal materials like titanium that deform by twinning. 5.3. Twin activity Twinning plays an important role in determining texture in all the samples. Twinning activity has been conventionally estimated by measuring the fraction of twins in the microstructure by EBSD [8] or by neutron diraction [15]. However, with the progress of deformation the material inside the twinned region is further deformed by slip and/or twinning. The slip activity within a twinned region can alter the character of the twin boundary and might eventually lead to an under-estimation of the twin fraction in the sample. Under these circumstances it is quite dicult

N.P. Gurao et al. / Acta Materialia 59 (2011) 34313446

3443

to estimate twin activity using EBSD. In the present investigation we propose a unique way of accessing the twin activity. It is well known that a twin is characterized by a parent and a daughter grain separated by a twin boundary. Owing to the fact that dierent types of twins have dierent type of twin boundaries, one way of quantifying twinning activity is by considering the fraction of these boundaries that indicate the activity of a particular twinning system. The advantage of using such a scheme is that any change in character of the twin boundary can also be monitored. In fact, deviation from the ideal value of misorientation corresponding to a twin boundary can provide an estimate of slip activity [31]. It is well known that low energy coherent twin boundaries act as barriers to slip [32] and accumulation of dislocations at such boundaries leads to deviation from the ideal misorientation value of the twin boundary. The currently available OIM software cannot provide the deviation from ideal orientation for the co-incident site lattice (CSL) boundaries of hcp materials. Therefore, in the present investigation the misorientation distribution for a particular coherent twin boundary is plotted for a range
0.12

of misorientations. This would provide an estimate of deviation from the ideal misorientation. For instance, the misorientation distribution for the contraction twin boundary at 64.27 is plotted between 60 and 70, while for the extension twin boundary at 84.7 the same has been plotted in the range 8090 (Fig. 11). Although this approach does not consider a particular axis, which is mandatory to dene a boundary in terms of the axisangle pair, it gives a rst order estimate of deviation from the ideal misorientation. In all cases deviation for the extension twin boundary is higher (broader peak) than that for the contraction twin boundary. This is a clear indication of higher slip activity near extension twins. The deviation from ideal misorientation for the individual twin boundaries is higher in case of sample II, indicating higher slip activity in this sample. This analysis, however, could not be extended to the samples deformed at the higher strain rate, as the quality of EBSD data obtained was not good enough to draw similar conclusions. 5.4. Deformation mode activity The four dierent initial textures led to a similar texture after low strain rate deformation. The VPSC model was used to reproduce the experimental textures. The relative contribution of each slip/twin system can be estimated from the activity plots obtained from the VPSC simulations. The respective CRSS values and the hardening parameters for various slip and twin systems were chosen in such a way as to match the texture as well as the stressstrain curves. It should be recalled (Section 5.3) that the same parameters are used to model the texture and stressstrain responses of the four dierently oriented samples. The twin activities obtained from the simulations are in reasonable agreement with the fraction of twin boundaries experimentally measured from EBSD (see Fig. 8a and b). At low strain rates the simulation results clearly indicate that prismatic slip was dominant for all samples at the onset of plastic deformation. Its contribution gradually decreased with an increase in the contribution from pyramidal and basal slip (Fig. 12). In the case of sample II the contribution from prismatic slip and contraction twinning decreased instantly and was compensated for by the pyramidal slip system. Sample II showed very little extension twinning while signicant extension twinning was observed in samples I, III and IV. These samples showed minimal contraction twinning. The high strain rate deformed samples also indicate the occurrence of prismatic slip at the beginning of plastic deformation. However, these samples showed higher twin activity than that for the low strain rate deformed samples. This ensured that the pyramidal slip system was not as active as in the case of low strain rate deformation. The pyramidal slip system started to dominate at a higher strain level in samples I, III and IV. The overall contribution of the pyramidal slip system was reduced at larger strains in

I II
0.08

Fraction

III IV

0.04

0 80 82 84 86 88 90

Misorientation

(a)
0.12

I II
0.08

III IV

Fraction
0.04 0 60

62

64

66

68

70

Misorientation

(b)
Fig. 11. Misorientation development for the samples deformed at low strain rate near (a) an extension twin boundary and (b) a contraction twin boundary.

3444
0.7 0.6 0.5

N.P. Gurao et al. / Acta Materialia 59 (2011) 34313446


0.7

I
Activity

0.6 0.5 0.4

II

Activity

0.4 Prism 0.3 0.2 0.1 0 0 0.1 0.2 Strain 0.7 0.6 0.5 0.3 0.4 0.5 Basal Pyram. TTW CTW

Prism 0.3 0.2 0.1 0 0 0.1 0.2 Strain 0.7 0.3 0.4 0.5 Basal Pyram. TTW CTW

III
Activity

0.6 0.5 0.4

IV

Activity

0.4 Prism 0.3 0.2 0.1 0 0 0.1 0.2 Strain 0.3 0.4 0.5 Basal Pyram. TTW CTW

Prism 0.3 0.2 0.1 0 0 0.1 0.2 Strain 0.3 0.4 0.5 Basal Pyram. TTW CTW

(a)
0.7

0.7 0.6 0.5


Activity Activity

0.6 0.5 0.4

II

0.4 Prism 0.3 0.2 0.1 0


0 0.1 0.2 Strain 0.3 0.4 0.5

Prism 0.3 0.2 0.1 0


0 0.1 0.2 Strain 0.3 0.4 0.5

Basal Pyram. TTW CTW

Basal Pyram. TTW CTW

0.7 0.6 0.5


Activity

0.7

III
Activity

0.6 0.5 0.4

IV

0.4 Prism 0.3 0.2 0.1 0


0 0.1 0.2 Strain 0.3 0.4 0.5

Prism 0.3 0.2 0.1 0


0 0.1 0.2 Strain 0.3 0.4 0.5

Basal Pyram. TTW CTW

Basal Pyram. TTW CTW

(b)

Fig. 12. Activity of various slip and twinning systems for the four samples deformed at (a) low strain rate and (b) high strain rate. Prism, prismatic slip system; Pyra., pyramidal slip system; TTW, {1 0 1 2} extension twinning; CTW, {1 1 2 2} contraction twinning.

the high strain rate deformed samples which is compensated partly by twinning and later by higher basal slip. Thus, there was a higher contribution from the basal slip system to deformation in all four samples at the higher strain rate. The contribution of basal slip to room temper-

ature deformation of low c/a metals like zirconium and titanium is considered minimal [2], with beryllium as an exception [33]. However, it has been reported that in spite of the fact that the CRSS for basal slip is higher than that for prismatic slip in titanium, basal slip can be activated

N.P. Gurao et al. / Acta Materialia 59 (2011) 34313446


5000 Initial Final S22 (MPa) 2500 0 -5000 -2500 -2500 -5000 S11 (MPa) 5000 Initial Final S22 (MPa) 0 2500 5000 S22 (MPa) 5000

3445

Initial Final 2500

II

-5000

-2500
-2500

2500

5000

-5000 S11 (MPa)

III
2500 0

5000 Initial Final S22 (MPa) 2500 0 -5000 -2500 -2500 -5000 S11 (MPa)
5000 Initial Final 2500 S22 (MPa)

IV

-5000

-2500

2500

5000

2500

5000

-2500 -5000 S11 (MPa) 5000 Initial Final 2500 S22 (MPa) 0 -5000 -2500 -2500 -5000 S11 (MPa) 5000 Initial Final S22 (MPa) 0 2500 5000

(a) I

II

0 -5000 -2500 -2500 -5000 S11 (MPa) 5000 Initial Final 0 2500 5000

III
2500 0 S22 (MPa)

IV
2500 0

-5000

-2500
-2500

2500

5000

-5000

-2500

2500

5000

-2500 -5000 S11 (MPa)

-5000 S11 (MPa)

(b)

Fig. 13. Polycrystal yield locus for the four dierent samples deformed at (a) low strain rate and (b) high strain rate with the stress along the compression direction coming out of the plane of the paper.

[10,3437]. Although the extent of slip activity cannot be estimated experimentally, a good match between the experimental and simulated texture validates the slip/twin activity estimated by simulation. 5.5. Evolution of yield surface The evolution of the polycrystalline yield surface with deformation is governed by hardening of the slip systems that leads to expansion of the yield locus, while the shape of the yield locus is governed by the texture evolution. The initial shape of the yield locus for each of the four sam-

ples in the stress space normal to the compression direction (S33) was dierent due to dierent initial textures (Fig. 13). The yield locus changed after deformation at the low as well as the high strain rate for each of the samples. This was due to the dierent values of CRSS used to model the deformation behaviour at the two strain rates. The yield loci for all four samples were not asymmetrical due to the dominance of slip at these strains and lesser contribution from twinning. For the low strain rate deformation the yield locus for sample II exhibited greater anisotropy, having a higher value for S11 than S22. On the other hand, the yield loci for samples I, III and IV were not as aniso-

3446

N.P. Gurao et al. / Acta Materialia 59 (2011) 34313446

tropic, in spite of a higher value for S11 than S22 (Fig. 13). The markedly dierent behaviour of sample II from the other samples is attributed to a higher contribution from pyramidal slip as against extension twinning in the other samples. The dierence between sample II and the other three samples is maintained for the high strain rate deformation condition. The higher stress levels at the high strain rate ensured that the polycrystalline yield locus expanded more in this case than for low strain rate deformation. 6. Summary and conclusions CP titanium with four dierent initial orientations was subjected to deformation at extreme strain rates, namely, 3 104 and 1.5 103 s1. From the examination of microstructure and texture of the deformed samples the following conclusions can be drawn. 1. At the low strain rate a similar texture with a strong h2  1 1 4i component (close to basal h0 0 0 1i) and a weak h2  1 1 0i component evolves, despite dierent initial textures. At the high strain rate the texture evolution is similar, with an additional h1 0  1 0i component for all samples. However, the texture of the high strain rate deformed samples is weaker than the low strain rate deformed samples. 2. The strain hardening response and twin activity are different for samples with dierent orientations at the low strain rate. The sample with basal orientations undergoes minimum twinning and, hence, exhibits least hardening, but a higher yield strength due to activation of the harder pyramidal slip system. At higher strain rates the strain hardening response is complicated due to enhanced twinning. The dierences in the strain hardening response of the dierently oriented samples is reduced at the high strain rate. 3. A viscoplastic self-consistent model was successful in predicting texture evolution, stressstrain response and twin activity for the titanium samples with dierent initial orientations. The agreement between the experimental and simulated textures is better at the low strain rate than the high strain rate. 4. The simulations indicate that the activation of basal slip is necessary to reproduce the experimental textures and the stressstrain response. The activation of basal slip with a higher CRSS value than prismatic slip can be attributed to higher stress levels achieved locally during the course of deformation. 5. The yield loci obtained from the simulations indicate that the contribution to the overall deformation is more from slip than twinning for all four samples at the low as well as the high strain rate. Acknowledgements To th for providThe authors acknowledge Prof. Laszlo ing the VPSC-6 code and also for elucidating various com-

plexities of crystal plasticity modelling during his stay at IISc, Bangalore. Thanks are due to Dr. Benoit Beausir for many discussions. The discussion with Prof. Surya Kalidindi has also been very useful. The microtexture studies were carried out using the SEM facility at the Institute of Nano-Science Initiative, Indian Institute of Science, Bangalore, for which the authors acknowledge a DSTFIST grant. The authors thank Mr. Sashidhara for help in carrying out the compression tests using DARTEC. References
[1] Gottstein G. Physical foundations of materials science. Berlin: Springer Verlag; 2004. p. 224. [2] Hanson BH. Mater Des 1986;7:301. [3] Conard H. Prog Mater Sci 1981;26:123. [4] Minonishi Y, Morozumi S, Yoshinaga H. Scripta Metall 1982;16:427. [5] Munroe N, Tan X, Gu H. Scripta Mater 1997;36:1383. [6] Chichili DR, Ramesh KT, Hemker KT. Acta Mater 1998;46:1025. [7] Nemat-Nasser S, Guo WG, Cheng JY. Acta Mater 1999;47:3705. [8] Salem AA, Kalidindi SR, Doherty RD. Scripta Mater 2002;46:419. [9] Salem AA, Kalidindi SR, Doherty RD. Acta Mater 2003;51:4225. [10] Salem AA, Kalidindi SR, Semiatin SL. Acta Mater 2005;53:3495. [11] Wu X, Kalidindi SR, Necker C, Salem AA. Acta Mater 2007;55: 423. CN, Lebensohn RA, Kocks UF. Acta Metall Mater [12] Tome 1991;39:2667. CN. Acta Metall Mater 1993;41:2611. [13] Lebensohn RA, Tome [14] Kaschner GC, Bingert JF, Liu C, Lovato ML, Maudlin PJ, Stout CN. Acta Mater 2001;49:3097. MG, Tome CN, Mauldin PJ, Lebensohn RA, Kaschner GC. Acta Mater [15] Tome 2001;49:3085. CN, Beyerlein IJ, Vogel SC, Brown DW, [16] Kaschner GC, Tome McCabe RJ. Acta Mater 2006;58:2887. CN, McCabe RJ, Misra A, Vogel SC, Brown [17] Kaschner GC, Tome DW. Mater Sci Eng 2006;463A:122. CN, Kaschner GC. Acta Mater 2007;55:2137. [18] Proust G, Tome CN. Int J Plasticity 2008;24:867. [19] Beyerlein IJ, Tome [20] Yang H, Yin S, Huang C, Zhang Z, Wu S, Li S, Liu Y. Adv Eng Mater 2008;10:955. [21] Cheneau-Spa th N, Fillit RY, Driver JH. J Appl Cryst 1994;27:980. [22] Rangaswamy P, Bourke MAM, Brown DW, Kaschner GC, Rogge CN. Metall Mater Trans 2002;33A:757. RB, Stout MG, Tome [23] Van Houtte P. Acta Metall 1978;26:591. [24] Patridge PG. Metall Rev 1967;12:169. [25] Christian JW, Mahajan S. Prog Mater Sci 1995;39:1. [26] ASM. High strain rate compression testing. Mechanical testing, vol. 8. West Conshohocken, PA: ASM International; 1985. p. 190. CN, Canova GR, Kocks UF. Acta Metall Mater 1984;32:1637. [27] Tome [28] Bhattacharyya A, Rittel D, Ravichandran G. Metall Mater Trans 2006;37A:1137. [29] Canova GR, Fressengeas C, Molinari A, Kocks UF. Acta Metall 1988;36:1961. [30] Beausir B, Toth LS, Neale KW. Acta Mater 2007;55:2695. [31] Bhowmik A, Biswas S, Suwas S, Ray RK, Bhattacharjee D. Metall Mater Trans 2009;40A:2729. [32] Reed-Hill RE. Physical metallurgy principles. Canada: D. Van Nostrand. [33] Brown DW, Abeln SP, Blumenthal WR, Bourke MAM, Mataya MC, CN. Metall Mater Trans 2005;36A:929. Tome [34] Anderson EA, Jillson DC, Dunbar SR. Trans AIME 1953;197:1191. [35] Oliver EC, Daymond MR, Quinta da Fonseca J, Withers PJ. J Neutron Res 2004;12:33. th LS, Fundenberger J-J, Gottstein G. Acta [36] Suwas S, Beausir B, To Mater 2011;59:1121. [37] Yapici GG, Karaman I. Mater Sci Eng 2009;503:78.

Das könnte Ihnen auch gefallen