Elsevier

Materials Science and Engineering: A

Volume 596, 24 February 2014, Pages 236-243
Materials Science and Engineering: A

Mechanisms of anisotropy of mechanical properties of α-titanium in tension conditions

https://doi.org/10.1016/j.msea.2013.12.061Get rights and content

Abstract

The plasticity of hexagonal materials is strongly anisotropic and involves different microscopic mechanisms such as mechanical twinning and dislocation glide. Twins are often considered to be responsible for various peculiar features of plastic flow, such as compression–tension anisotropy, a high work hardening rate, or a particular three-stage shape of deformation curves observed not only for single crystals but also for polycrystals. However, the role of twins remains a matter of debate. Moreover, most of the experimental results have been obtained in compression conditions and it is not clear whether the same features appear in other testing conditions. In the present work, tensile tests were performed on commercially pure α-Ti samples cut along the rolling and the transverse direction, and yielded several unexpected results. In particular, the work hardening rate was found to be lower in the latter case, although the EBSD measurements revealed a larger twin volume fraction in such samples. Besides, the two kinds of specimens showed an opposite sign of the strain-rate effect on the tendency to the three-stage work hardening. The possible role of twinning, dislocation glide, and aging of dislocations on the anisotropy of mechanical behavior of titanium is discussed. A mechanism based on the consideration of the dislocation glide anisotropy is suggested.

Introduction

Titanium alloys are widely used in the aeronautic and biomedical industries due to their high elastic limit-to-density ratio, high resistance to corrosion and good biocompatibility. Consequently, this material is the object of intense research since the past decades. Many studies are devoted to pure or commercially pure (CP) titanium, since its deformation mechanisms are still not well understood. At ambient temperature, titanium has a hexagonal close-packed (hcp) structure with a less-than ideal c/a ratio of 1.587. In general, hcp metals are characterized by high anisotropy of mechanical properties with regard to crystallographic directions. Contrarily to materials with more isotropic cubic structures, which often deform uniquely by dislocation glide, they do not possess enough easy slip systems to achieve arbitrary shape change. As a result, their plastic deformation is more complicated because of the interaction between dislocation glide and twinning [1]. In addition to this complexity, the role of twinning strongly varies between different hcp materials, in relation to the c/a ratio. For example, Mg (c/a=1.624) is known to show much more intense twinning than Ti (see, e.g., [2]). However, hcp materials may manifest many similar mechanical properties, e.g., an asymmetry of the yield stress with regard to the stress sign and material texture, a high work hardening rate, and, depending on the loading direction in relation to the texture, often an unusual concave shape of compression curves (see, e.g., [3]). These peculiarities are often considered from the viewpoint of twinning. However, even the experimental observations on one specific material remain a matter of controversy. As far as the totality of various data is concerned, it gives rise to diverse interpretations.

One point of controversy concerns the observation of the first occurrence of twins and, more specifically, the role of twinning in the material yielding. In particular, the phenomenon of tensile–compression asymmetry is usually referred to as a lower yield stress in compression compared with tension, as often observed for Mg alloys [4]. This effect was attributed to twinning providing an additional deformation mode in compression. At the same time, experiments on pure Ti showed either approximately equal values of the yield stress in tension and compression or a lower value in tension [5]. This difference between behaviors of two hcp materials may be explained by the fact that in Ref. [5] twinning was only observed after some plastic strain, whereas twins were detected immediately after yielding in Ref. [4]. On the other hand, such an explanation cannot account for another kind of asymmetry observed in tension of Zr (c/a=1.593) specimens: namely, the specimens deformed along the rolling direction displayed a lower yield stress than those deformed parallel to the transverse direction, although the onset of plastic flow in “transverse” specimens was found to be related to the start of deformation twinning [6].

Many authors discuss that twin boundaries acting as barriers to dislocations glide, likewise grain boundaries, can contribute to an increase in the flow stress. This suggestion is, however, not as obvious as it would seem. Indeed, the measurement of the reloaded yield stresses in pre-strained Ti samples showed that the yield stress in samples with a higher twin density was not significantly higher than that for samples with fewer twins [7]. Moreover, it was below the flow stress level manifested by the latter samples deformed to the corresponding strain. This observation requires consideration of the other aspects of the effect of twinning, e.g., facilitation of dislocation glide because of the reorientation of the lattice inside the twins.

Even more debate has been devoted to strain hardening behavior of hcp materials. An often reported feature found for various materials, in particular in compression conditions, consists in a three-stage (or “concave”) character of deformation curves [3], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17]. More precisely, the initial stage A characterized by a decreasing strain hardening rate Θ=dσ/dεp is followed by an increase in Θ (stage B) and, finally, a new decrease (stage C). In some cases the increase in Θ is not observed but the stage B is still detected as a slowing-down of the decrease with regard to the stage A [11], [13], thus testifying to a general nature of three-stage behavior. Moreover, a similar effect is also reported for fcc metals with low stacking fault energy, which also manifests twinning [18], [19], [20]. However, different authors disagree about the mechanisms responsible for these variations in the strain hardening rate. Concerning pure or CP-Ti, several distinct points of view have been discussed in the literature [8], [9], [10], [11], [12], [13]. Taking into account the strong texture, Kailas et al. suggested an analogy with the well-known hardening stages of single crystals, thus ascribing the three stages observed in CP-Ti, respectively, to easy glide, multiple dislocations slip, and dynamic recovery [8]. Nemat-Nasser et al. suggested that the stage B in CP-Ti could be ascribed to dynamic strain aging of dislocations, caused by their interaction with solute atoms [9]. Nevertheless, this explanation does not apply to ambient temperature because the dynamic strain aging is only essential in Ti in the temperature range of 500–850 K, as shown in Refs. [21], [22], [23] and discussed by Salem et al. [10], [11], [12] who performed an elaborated investigation of plasticity of Ti in relation to microstructure changes. These authors observed that stage B usually coincides with the onset of twinning in Ti. They ascribed the stage A to the usual dynamic-recovery regime observed in metals with a high stacking fault energy, the stage B to the effect of twins on the resistance to dislocation glide, and the stage C to saturation of the twin volume fraction.

Some additional mechanisms were proposed in references devoted to other hcp materials [14], [15], [16], [17]. For instance, Sarker and Chen found an increasing twin volume fraction during stage A in compression of a AM30 Mg alloy, followed by its decrease (detwinning) after the end of this stage [16]. They supposed that stage A is caused by intensive twinning, the following increase in the work hardening (stage B) is due to interaction of dislocations with the twins generated during stage A, and the detwinning eventually leads to a decrease in Θ (stage C). Wang et al. studied compression of an AZ31 Mg alloy and suggested that the main contribution to increasing Θ stems from texture strengthening as a result of twinning, which rotates grains into hard orientations [17].

It should be underlined that the majority of the above data on the strain hardening behavior of Ti were obtained under conditions of compression. Most works in tension conditions presented examples of usual parabolic-like tensile curves typical of polycrystals (e.g., [5], [24]), except for a very recent paper by Becker and Pantleon, who identified three different work-hardening stages during tensile deformation of CP-Ti [13]. Nevertheless, the observed effect was relatively weak, in the sense that Θ showed a slowed-down decrease during stage B in all their tests. Hence, it is still not totally clear whether the conclusion on the three-stage nature of plastic deformation may also be relevant in tension. The present paper aims at characterizing the mechanical anisotropy of commercially pure titanium and especially its strain hardening behavior in tensile conditions. As the propensity to twinning is highly dependent on the crystallographic texture of hexagonal materials, the tests are realized for two orientations of the tensile axis of specimens with regard to the rolling direction. The use of the tensile geometry also made it possible to combine the conventional tensile tests with high resolution local extensometry. This effort was provoked, in particular, by the conjecture of the role of aging of dislocations. Indeed, the aging is known to be responsible for a yield plateau associated with propagation of a Piobert–Lüders deformation band, and may thus explain low Θ behavior during stage A [25].

The paper is organized as follows. Section 2 presents experimental details. Section 3 first describes macroscopic mechanical behavior of two kinds of specimens, with an accent put on the three-stage character of deformation, as reflected in the tension curves and the evolution of the work hardening rate. The local extensometry results are then presented and the section is completed with the results of the microstructure analysis. The overall discussion and a hypothesis regarding the physical mechanism, which may explain the totality of experimental data, are presented in Section 4. Concluding remarks follow.

Section snippets

Experimental technique

The as-received commercially-pure titanium sheet with the composition given in Table 1 had an average grain size of around 9 µm. The samples with a gage section of 30×7×1.62 mm3 were cut parallel to either the rolling direction (hereinafter referred to as RD-samples) or the transverse direction (TD-samples). The specimens were deformed at room temperature at a constant crosshead velocity corresponding to initial applied strain rates ε̇a ranging from 5×10−5 s−1 to 8×10−3 s−1. Local strain-rate

Deformation curves

Two to four samples were tested for each strain rate and each orientation, and displayed very good reproducibility. Young's moduli were measured using the initial linear portions of the deformation curves obtained in the tests with the sensor arm extensometer covering the entire gage length of the specimens. The calculations yielded a value of 109.5±2.5 GPa for TD-samples and 103.7±2.8 GPa for RD-specimens, without noticeable dependence on the applied strain rate. These values agree with various

Discussion and conclusions

To summarize, the following main observations can be underlined:

  • Very few twins are found in RD-samples deformed in tension. More specifically, the observed twin volume fraction was always less than 0.5%. Twinning is much more significant for the transverse direction (Fig. 7). However, even in this case the quantitative estimates reveal the twin volume fraction of about several percents, which is an order of magnitude less than reported for compression data for slightly larger grain size in the

Acknowledgments

This work was supported by the French State through the program "Investment in the future" operated by the National Research Agency (ANR) and referenced by ANR-11-LABX-0008-01 (LabEx DAMAS), and through the ANR project PHIRCILE (ANR 2010 JCJC 0914 01).

References (37)

  • J.W. Christian et al.

    Prog. Mater. Sci.

    (1995)
  • A. Ghaderi et al.

    Acta Mater.

    (2011)
  • J. Jiang et al.

    Mater. Sci. Eng. A

    (2008)
  • M.E. Nixon et al.

    Int. J. Plast.

    (2010)
  • D.R. Chichili et al.

    Acta Mater.

    (1998)
  • S. Nemat-Nasser et al.

    Acta Mater.

    (1999)
  • A.A. Salem et al.

    Scr. Mater.

    (2002)
  • A.A. Salem et al.

    Acta Mater.

    (2003)
  • H. Becker et al.

    Comput. Mater. Sci.

    (2013)
  • F. Coghe et al.

    Mater. Sci. Eng. A

    (2012)
  • M.R. Barnett et al.

    Acta Mater.

    (2004)
  • D. Sarker et al.

    Scr. Mater.

    (2012)
  • B. Wang et al.

    Mater. Sci. Eng. A

    (2012)
  • S.R. Kalidindi

    Int. J. Plast.

    (1998)
  • T.A. Lebedkina et al.

    Mater. Sci. Eng. A

    (2009)
  • K. Renard et al.

    Mater. Sci. Eng. A

    (2012)
  • R.E. Reed-Hill et al.

    Scr. Metall. Mater.

    (1995)
  • J.S. Lecomte et al.

    Mater. Sci. Eng. A

    (1997)
  • Cited by (93)

    View all citing articles on Scopus
    View full text