Sei sulla pagina 1di 5

Scripta Materialia 49 (2003) 10071011 www.actamat-journals.

com

The inuence of triaxial stress on the ideal tensile strength of iron


D.M. Clatterbuck *, D.C. Chrzan, J.W. Morris Jr.
Center for Advanced Materials, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA Department of Materials Science and Engineering, University of California, Berkeley, CA 94720, USA Received 30 June 2003; received in revised form 22 July 2003; accepted 1 August 2003

Abstract The eect of a superimposed hydrostatic tension or compression on the ideal tensile strength of Fe has been computed ab initio using the Projector Augmented Wave method. Interestingly, a superimposed hydrostatic tension increases the ideal tensile strength while hydrostatic compression decreases it. This result can be explained by a simple crystallographic model. 2003 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Keywords: Iron; Mechanical properties; Ideal strength; Ab initio electron theory

1. Introduction The mechanical strength of a material is determined by the propagation of defects such as dislocations or cracks. However, in the absence of defects the strength of a crystal is limited by the stress at which the lattice itself becomes unstable [1]. This stress, often referred to as the ideal strength, is one of the few mechanical properties of a material that can be calculated ab initio [2]. The ideal strength is of practical interest since it sets an upper bound on the strength the material can have. Moreover, it is directly relevant in experi* Corresponding author. Address: Center for Advanced Materials, Lawrence Berkeley National Laboratory, 1 Cyclotron Road Mailstop 66, Berkeley, CA 94720, USA. Tel.: +1510-486-6035; fax: +1-510-486-4995. E-mail address: clatterb@uclink4.berkeley.edu (D.M. Clatterbuck).

mental situations where there are few mobile defects. For example, the shear stresses achieved in the nanoindentation of defect-free lms have been shown to approach the ideal shear strength [3]. While most studies of the ideal strength have focused on simple loading geometries such as uniaxial tension and pure shear [2], the stresses that govern many of the most interesting mechanical phenomena are multiaxial. The Hertzian stress beneath a nanoindenter is a case in point. A second example, of equal or greater importance, is the triaxial stress eld at the tip of a sharp crack or aw. The crack tip stress eld in mode I loading (tension perpendicular to the plane of the crack) is the sum of a uniaxial tension and a hydrostatic tension from the geometric constraint imposed by the bulk of the specimen. The hydrostatic tension can be large, as much as 45 times the uniaxial component [4,5]. The hydrostatic tension plays a very important role in the fracture of ductile

1359-6462/$ - see front matter 2003 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/S1359-6462(03)00490-1

1008

D.M. Clatterbuck et al. / Scripta Materialia 49 (2003) 10071011

materials since it has the consequence that the total tensile stress, which drives crack propagation to fracture, can be very high while the shear stress, which drives the plastic deformation that relieves the crack tip stress, remains relatively small. The boost provided by the hydrostatic tension plays a major role in the ductilebrittle transition behavior in bcc metals. Typical bcc metals fracture in a brittle mode when they are tested at low temperature after being cracked or notched, but may show considerable ductility at the same temperature when tested as wires or smooth-surfaced tensile specimens. If the tensile stress ahead of a crack tip reaches the ideal tensile strength, the material will necessarily fail and the crack will propagate. However, even in this case it does not automatically follow that the material will fracture in a brittle mode. Fcc metals such as Al and Cu fail in shear when stressed to the ideal strength in uniaxial tension [6]. Typical bcc metals such as Fe [7], W [8] and Mo [9] fail in tension (and, presumably, cleave) when stressed to the ideal strength in the weak direction (h1 0 0i in this case), but bcc Nb [9] fails in shear. It follows that to understand how a material might behave if it were loaded to its ideal strength in a crack tip stress eld we must address two issues: how the ideal tensile strength is changed by hydrostatic tension, and whether the failure mode is tension or shear. In the following we address these issues for the particular case of bcc Fe.

2. Computational method As discussed in [10], we dene the ideal strength as the stress at which a homogeneously deformed, perfect crystal becomes elastically unstable with respect to internal displacements. This is a sucient, though not necessary condition for failure; in particular experimental situations instability can be induced at lower stresses by the behavior of the loading mechanism [11,12]. However, for a given material and loading geometry this denition provides a unique measure of the ideal strength that sets an upper bound on the actual value. The computational method used to determine the ideal

strength is similar to that used in previous work [7,8]. The lattice parameters of the crystal are incrementally strained in the direction of the applied stress. After each increment of strain the crystal structure is relaxed to achieve the stress state desired. For uniaxial tension the incremental strain is in the direction of tension and the crystal is relaxed until all perpendicular and shear stresses vanish. The triaxial stress states investigated here include a triaxial tension in which we have taken r11 2r22 2r33 , and a uniaxial tension plus triaxial compression in which r11 r22 r33 . Although the ratios of the uniaxial stress to the hydrostatic stress in the two cases are not equal in magnitude, these stress states were chosen for convenience in order to study the trends associated with changing the hydrostatic stress. The total energy and HellmannFeynman stresses were computed using density functional theory and the generalized gradient approximation. The projector augmented wave method (PAW) was chosen for these calculations due to its reasonable accuracy and eciency [13]. Previous studies have shown that this method gives ideal strength values for iron which are similar to those found by full potential calculations [7,14]. The PAW method implementation within the Vienna Ab initio Simulation Package (VASP) was used [1517]. We used a scalar relativistic treatment for the 3d and 4s valence states, while treating the lower energy levels with the frozen core approximation. The plane-wave expansion had an energy cut-o of 25 Ry. The modied tetrahedron method of Blchl was used for integration in the Brillouin o zone [18], with the number of k-points chosen to converge the energies within 0.1 mRy and the HellmannFeynman stresses to within 0.1 GPa. The accuracy of the stresses produced by the tetrahedron method was conrmed by comparisons with calculations done using the integration scheme of Methfessel and Paxton [19]. The exchange correlation energy was computed using a method based on the generalized gradient approximation of Perdew and co-workers [20,21]. There are slight variations from this representation as described in [17]. The HellmannFeynman stresses are calculated at each increment of the strain in order to relax the crystal. The relaxation

D.M. Clatterbuck et al. / Scripta Materialia 49 (2003) 10071011

1009

is continued until the stresses are within 0.1 GPa of the chosen values. As the ideal strength in this work is dened in terms of a quasi-static homogeneous deformation at 0 K, these calculations do not consider dynamic instabilities such as soft phonons which may lower the ideal strength.

3. Results The calculations done here assumed a dominant tension, r11 , in a h0 0 1i direction, which is known to be the weak direction for bcc metals [2]. Fig. 1 shows the stress as a function of engineering strain in the [0 0 1] direction for the three dierent stress states considered. We see that a superimposed hydrostatic tension raises the ideal tensile strength relative to the uniaxial case while a superimposed hydrostatic compression lowers it. In conjunction with this we note that the strain corresponding to the ideal strength is increased in the tensile case and decreased in the compressive case.

These results contradict the intuitive view (which we and others have held) that hydrostatic tension would decrease the tensile strength of a material. However, the reason for this result becomes clear if one examines the eect of these stresses from the perspective of their associated strains. The application of a [0 0 1] tensile strain to a bcc metal causes the symmetry to be broken and the crystal structure becomes body centered tetragonal (bct). Upon increasing the c=a ratio to p 2, the fcc structure is reached (Fig. 2). This bcc to fcc transformation is commonly referred to as the Bain Transformation. Cubic symmetry requires that the [0 0 1] tensile stress goes to zero at both the bcc and fcc structures (the ferromagnetic order in the case of iron breaks this symmetry, but as we have previously shown, these structures are stress free in ferromagnetic iron [7]). As such, these structures are symmetry-induced extrema (or saddle points) on the multidimensional energystrain surface. We can model the stressstrain curve between the two energy extrema by a sinusoid as suggested by Orowon [22] r r0 sinep=emax ; 1

20.0 17.5

Compression Uniaxial Tension

Stress (GPa)

15.0 12.5 10.0 7.5 5.0 2.5 0.0 0.00 0.05 0.10 0.15 0.20 0.25 0.30

where r0 is the ideal strength, e is the engineering strain, and emax is the engineering strain of the second stress-free structure. Requiring that the Hookes law be obeyed at small strain, one nds that r0 Eh0 0 1i emax =p; 2

where Eh0 0 1i 1=s11 is the Youngs modulus in the h0 0 1i direction.

Engineering Strain
Fig. 1. Stress as a function of strain for ferromagnetic Fe under three dierent multiaxial loads: (N) uniaxial tension plus biaxial compression; (j) uniaxial tension; and (r) uniaxial tension plus biaxial tension. In all cases the largest tensile stress is in the [0 0 1] direction. The large symbols with arrows indicate the location of tetragonal to orthorhombic instabilities. The loading conguration which includes biaxial compression was not tested for this type of instability.

p Fig. 2. Increasing the c=a ratio of a bct crystal from 1 to 2 corresponds to transforming a bcc crystal into an fcc crystal.

1010

D.M. Clatterbuck et al. / Scripta Materialia 49 (2003) 10071011

Now consider the eect of a constant tensile stress applied in the [1 0 0] and [0 1 0] directions. The addition of the orthogonal tensile stresses increases the crystal lattice parameters in the [1 0 0] and [0 1 0] directions. Because of this, the [0 0 1] axis must be stretched farther to achieve the fcc structure, increasing the strain, emax . Eq. (2) then implies that a superimposed hydrostatic tension should increase the ideal strength. In a similar way, the addition of compressive stresses orthogonal to the tensile direction will have the opposite eect and decrease the strain needed to reach the fcc saddle point structure. While the sinusoidal approximation is not exact, its qualitative implications agree with the ab initio calculations. In the present work, the multiaxial stresses were not a constant but rather were proportional to the applied stress. This has the eect that the fcc structure must be stress-free and thus must correspond to the equilibrium fcc structure found in uniaxial tension. However, the above analysis can still be used to understand the changes in strength: the stress state at the peak of the stressstrain curve under proportional multi-axial loading would also occur during loading with constant multi-axial stresses if one makes a suitable choice for the constant multi-axial stress. An additional eect not explicitly included in Eq. (2), is caused by the pressure dependence of the Youngs modulus. It would be possible for triaxial tensile stresses to lower the ideal tensile strength of a bcc metal if the additional stresses decreased the Youngs modulus more than they increased the strain needed to reach the fcc structure. From the initial slopes in Fig. 1, we see that triaxial tensile stresses increases the Youngs modulus in iron, which increases the ideal strength even more than if the Youngs modulus were independent of pressure. This increase in the Youngs modulus with pressure is consistent with the experimentally determined pressure dependence of the elastic constants [23,24]. The second issue that needs to be addressed is the fracture mode. It has been previously found that at large strains the body centered tetragonal structure becomes elastically unstable with respect to perturbation into a face centered orthorhombic structure [9,7]. If this transition occurs before the

ideal strength is reached, the ideal strength will be determined by the elastic instability associated with this orthorhombic structure. In contrast to the elastic instability along the Bain path, which is tensile in nature, the orthorhombic instability has a shear character and corresponds physically to h1 1 1if1 1 2g shear. Under uniaxial tension, Fe remains tetragonal to strains that are beyond its stability limit, and, hence, fails in tension (cleavage). The calculations done here (Fig. 1) show that a superimposed triaxial tension increases the strain at which the tetragonal to orthorhombic transition occurs; however, the increase is less than the increase in the strain associated with the peak in the stress strain curve (see Fig. 1). The eect is thus to cause the orthorhombic instability to occur closer to the maxima in the stressstrain curve. While cleavage is predicted, the margin favoring cleavage is less than in the uniaxial case. In this work, we have only considered the case of ferromagnetic iron, and, hence, have not studied the eect of hydrostatic pressure on the magnetic transformations that can occur along the Bain path. In the uniaxial case, the elastic instability occurs at smaller strains than these magnetic transitions and thus it dictates the ideal strength [7]. To a rst-order we expect that the addition of orthogonal tensile stresses would enhance the stability of the ordered magnetic phases over the non-spin polarized (paramagnetic) phases phase due to a narrowing of the bandwidth. Further insight can be obtained by examining the energy as a function of volume for the various magnetic phases at xed applied strains. As the volume decreases, the lowest energy magnetic phase changes from ferromagnetic to antiferromagnetic and nally to non-spin polarized. This trend is seen explicitly in [14] for Fe in the fcc crystal structure and was found to hold at lower applied strains as well. Thus, orthogonal tensile stresses would increase the stability of the ferromagnetic phase moving the magnetic transformation to larger strain than that found in the uniaxial case. It is not clear, however, if the additional stresses would change the strain associated with the magnetic phase transitions more or less than the strain associated with the elastic instability.

D.M. Clatterbuck et al. / Scripta Materialia 49 (2003) 10071011

1011

4. Conclusions The eects of multiaxial loading on the ideal tensile strength of iron have been investigated using density functional calculations. The superposition of a hydrostatic tension increases the ideal strength because the perpendicular tensile stresses increase the strain needed to reach the stress-free saddle point structure along the loading path. The superposition of a hydrostatic compression has the opposite eect; the ideal tensile strength decreases. While the biaxial tensile stresses slightly modify the strain associated with the competing shear failure, the tensile failure found in the uniaxial case is preserved suggesting that iron loaded at low temperature should fail by cleavage in the crack tip stress eld, as it does.

References
[1] Kelly A, Macmillan NH. Strong solids. 3rd ed. Oxford: Clarendon Press; 1986. [2] Morris Jr JW, Krenn CR, Roundy D, Cohen ML. In: Turchi PE, Gonis A, editors. Phase transformations and evolution in materials. Warrendale, PA: The Minerals, Metals and Materials Society; 2000. p. 187207. [3] Krenn CR, Roundy D, Cohen Marvin L, Chrzan DC, Morris Jr JW. Phys Rev B 2002;65:13411.

[4] Morris Jr JW, Guo Z, Krenn CR, Kim YH. ISIJ Int 2001;41:599. [5] McMeeking RM, Parks DM. In: Landes JD, Begley JA, Clarke GA, editors. Elasticplastic fracture. ASTM STP, vol. 668. Philadelphia, PA: American Society for Testing and Materials; 1979. p. 17594. [6] Morris Jr JW, Krenn CR, Roundy D, Cohen Marvin L. Mater Sci Eng A 2001;309:121. [7] Clatterbuck DM, Chrzan DC, Morris Jr JW. Acta Mater 2003;51:2271. [8] Roundy D, Krenn CR, Cohen Marvin L, Morris Jr JW. Philos Mag A 2001;81:1725. [9] Luo W, Roundy D, Cohen Marvin L, Morris Jr JW. Phys Rev B 2002;66:94110. [10] Morris Jr JW, Krenn CR. Philos Mag A 2000;80:2827. [11] Hill R. Math Proc Camb Philos Soc 1975;77:225. [12] Hill R, Milstein F. Phys Rev B 1977;15:3087. [13] Bl chl PE. Phys Rev B 1994;50:17853. o [14] Clatterbuck DM, Chrzan DC, Morris Jr JW. Philos Mag Lett 2002;82:141. [15] Kresse G, Furthmller J. Phys Rev B 1996;54:11169. u [16] Kresse G, Hafner J. J Phys Condens Matter 1994;6:8245. [17] Kresse G, Joubert D. Phys Rev B 1999;59:1758. [18] Bl chl PE, Jepsen O, Andersen OK. Phys Rev B o 1994;49:16223. [19] Methfessel M, Paxton AT. Phys Rev B 1989;40:3616. [20] Perdew JP, Chevary JA, Vosko SH, Jackson KA, Pederson MR, Singh DJ, et al. Phys Rev B 1992;46:6671. [21] Perdew JP, Wang Y. Phys Rev B 1992;45:13244. [22] Orowon E. Rept Prog Phys 1949;12:323. [23] Rotter CA, Smith CS. J Phys Chem Solids 1966;27:267. [24] Guinan MW, Beshers DN. J Phys Chem Solids 1968;29:541.

Potrebbero piacerti anche