Local strain evolution due to athermal γ→ε martensitic transformation in biomedical CoCrMo alloys

journal of the mechanical behavior of biomedical materials 32 (2014) 52 –61

Available online at www.sciencedirect.com

www.elsevier.com/locate/jmbbm

Research Paper

Local strain evolution due to athermal γ-ε martensitic transformation in biomedical Co–Cr–Mo alloys Kenta Yamanakaa,n, Manami Morib, Yuichiro Koizumia, Akihiko Chibaa a

Institute for Materials Research, Tohoku University, 2-1-1 Katahira, Aoba-ku, Sendai 980-8577, Japan Department of Materials and Environmental Engineering, Sendai National College of Technology, 48 Nodayama, Medeshima-Shiote, Natori 981-1239, Japan b

ar t ic l e in f o

abs tra ct

Article history:

Locally developed strains caused by athermal γ face-centered cubic (fcc)-ε hexagonal

Received 29 September 2013

close-packed (hcp) martensitic transformation were investigated for the γ matrix of Ni-free

Received in revised form

Co–29Cr–6Mo (wt%) alloys prepared with or without added nitrogen. Electron-backscatter-

4 December 2013

diffraction-(EBSD)-based strain analysis revealed that in addition to ε-martensite interiors,

Accepted 16 December 2013

the N-free alloy that had a duplex microstructure consisting of the γ matrix and athermal

Available online 24 December 2013

ε-martensite plates showed larger magnitudes of both elastic and plastic strains in the

Keywords:

γ phase matrix than the N-doped counterpart that did not have a ε-martensite phase.

Biomedical Co–Cr–Mo alloys

Transmission electron microscopy (TEM) results indicated that the ε-martensite micro-

Martensitic transformation

plates were aggregates of thin ε-layers, which were formed by three different {111}γ〈112〉γ

Local strain distribution

Shockley partial dislocations in accordance with a previously proposed mechanism

Electron backscatter diffraction

(Putaux and Chevalier, 1996) that canceled the shear strains of the individual variants.

(EBSD)

The plastic strains are believed to have originated from the martensitic transformation

Transmission electron microscopy

itself, and the activity of dislocations is believed to be the origin of the transformation. We

(TEM)

have revealed that the elastic strains in the γ matrix originate from interactions among the ε-martensite phase, extended dislocations, and/or thin ε-layers. The dislocations highly dissociated into stacking faults, making stress relaxation at intersections difficult and further introducing local strain evolution. & 2013 Elsevier Ltd. All rights reserved.

1.

Introduction

Co–Cr–Mo alloys are commonly used for orthopedic implants such as artificial hip and knee joints because the alloys exhibit excellent resistance to corrosion and wear (Niinomi, 2002; Buford and Goswami, 2004; Chiba et al., 2007). The metal-on-metal bearings produced with Co–Cr–Mo alloys n

Corresponding author. Tel.: þ81 22 215 2118; fax: þ81 22 215 2116. E-mail address: [email protected] (K. Yamanaka).

1751-6161/$ - see front matter & 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jmbbm.2013.12.019

enable the deployment of the large-diameter femoral head component of artificial joints, which can significantly improve upon the limited range of motion of metal-onceramic or metal-on-polymer hip devices (Kluess et al., 2007). Further improvement in alloy performance is essential in response to a strong demand for highly durable biomedical metallic materials in order to develop long-lifetime implant

journal of the mechanical behavior of biomedical materials 32 (2014) 52 –61

devices. Conventional Co–Cr–Mo alloys that are in compliance with the ASTM F75 standard are allowed to contain up to 0.35 wt% carbon to precipitate carbides (e.g., M23C6) as a strengthening phase (Rajan, 1982; Caudillo et al., 2002; Lee et al., 2006; Mineta et al., 2010). However, recent research has revealed that the hard carbide phase is detrimental to resistance against wear and corrosion and to biocompatibility of the implants (Chiba et al., 2007; Battini et al., 2011; Liao et al., 2012). Thus, novel approaches to designing and optimizing microstructures in Co–Cr–Mo alloys have previously been proposed in order to improve the performance of the alloys (Matsumoto et al., 2010; Yamanaka et al., 2009, 2011,2012a, 2012b, 2013; Mori et al., 2010, 2012; Lee et al., 2008; Salinas-Rodriguez and Rodriguez-Galicia, 1996; Mani and Salinas-Rodriguez López, 2011). It is well known that the characteristics of the matrix phase, e.g., phase stability, dislocations, grain sizes, and annealing twins, significantly affect the mechanical and tribological properties of Co–Cr–Mo alloys. In general, biomedical-grade Co–Cr–Mo alloys are composed of a metastable face-centered cubic (fcc) γ matrix and a hexagonal close-packed (hcp) ε-martensite phase, which forms during quenching, plastic deformation, and/or isothermal heat treatment (Chiba et al., 2007; Yamanaka et al., 2011, 2012b; Mori et al., 2010, 2012; Mani and Salinas-Rodriguez López, 2011; Kurosu et al., 2010). The ε-martensite phase plays a crucial role in improving the wear properties of Co–Cr–Mo alloys (Chiba et al., 2007) while diminishing plastic deformability (Yamanaka et al., 2012b; Mori et al., 2010; Mani and SalinasRodriguez López, 2011). Thus, it is intrinsically important to understand the formation kinetics, fraction, spatial distributions, and resulting microstructural changes that occur during ε-martensite transformation. For example, nickel has long been known to stabilize the γ phase by suppressing the martensitic transformation of cobalt and cobalt-based alloys. Consequently, Ni-containing alloys that do not undergo the γ-ε martensitic transformation exhibit excellent elongationto-failure (Chiba et al., 1999). High concentrations of Ni (410%), which may at least cause allergies and at worst cause cancer in living organisms (Denkhaus and Salnikow, 2002), have conventionally been incorporated into biomedical Co–Cr-based alloys for applications that require high deformability (designated ASTM F90 and F562) (Nagai et al., 2012; Marrey et al., 2006). Nitrogen, on the other hand, is a nontoxic γ-phase stabilizer that is attractive both from practical and scientific perspectives (Yamanaka et al., 2012b, 2013; Mori et al., 2012). We have investigated the structural and mechanical properties of hot-forged Co–29Cr–6Mo (wt%) alloys produced with and without added N (Yamanaka et al., 2012b). Hot forging significantly improved the strength of the alloys; however, N-free alloys containing an athermal ε-martensite phase exhibited premature fracturing for grains ranging from the submicron level to a few hundred micrometers before the onset of macroscopic necking. In contrast, N-doped alloys that had been subjected to hot deformation exhibited a better combination of high strength (1400 MPa under 0.2% proof stress) and large elongation-to-failure (420%) than their conventionally prepared counterparts (Yamanaka et al., 2013). We also found that the N-free duplex alloys

53

had a microstructure that was more distorted than the N-containing alloy with a single γ matrix produced under identical conditions (Yamanaka et al., 2012b). The amount of strain accumulated in the γ phase directly affects the mechanical and tribological properties of alloys. Therefore, it is important to elucidate the mechanism of athermal martensitic transformation and its effect on strain evolution in the γ matrix in order to optimize the types and amounts of various microstructures in alloys. Electron backscatter diffraction (EBSD) is a powerful tool that is used to investigate the local strain distributions in crystalline materials. Wilkinson et al. (2006a, 2006b) have recently developed an EBSD-based method of accurately quantifying the magnitude of elastic strain and the degree of lattice rotation for various alloys. They applied this method to structural metallic materials such as Ni-base superalloys (Karamched and Wilkinson, 2011). Miyamoto et al. (2009) also used this method to accommodate transformation strain in the austenite matrix for various ferrous alloys containing α0 body-centered cubic (bcc) martensite phases. Although there have been some reports to date on the martensitic phase transformation (Mori et al., 2010; Yamanaka et al., 2013; Song et al., 2006; Huang and López, 1999; Koizumi et al., 2013), they mainly focused on the formation kinetics and its effects on macroscopic mechanical properties of alloys. However, there have been a few studies in which local strains that develop during the martensitic transformation have been evaluated. In this study, the elastic and plastic strains introduced into the γ matrix of Co–Cr–Mo alloys through the athermal γ-ε martensitic transformation were characterized using EBSD analysis. Complementary transmission electron microscopy (TEM) was performed to investigate the dislocation structures and how the ε-martensite phase evolves.

2.

Experimental

The Co–29Cr–6Mo (wt%1) ternary alloy, which accords with the ASTM F75 standard, was prepared using high-frequency induction melting in an argon atmosphere. For comparison, the Co–29Cr–6Mo–0.17N alloy was also prepared using Cr2N powder as a nitrogen source under the same conditions. The chemical compositions of the alloys are listed in Table 1. Hereafter, these alloys are denoted as N-free and N-doped alloys, respectively. The cast ingots (Φ15 mm) were subjected to heat treatment at 1473 K for 1.8 ks in order to homogenize the alloys throughout the ingots and were then hot-rolled to Φ9.6 mm (equivalent strainE0.89). The specimens were subsequently heated at 1473 K for 0.6 ks to remove the strains introduced by hot-rolling and were then quenched in water. The size of the γ grains in each alloy was approximately 100 μm. EBSD was measured using a field-emission scanning electron microscope (FE-SEM, FEI XL30S-FEG) operated at 15 kV. The EBSD data were accumulated and analyzed on a TSL (TexSEM Laboratories, Inc.) orientation image microscope (OIM) system. Lattice constants were measured 1

In this paper, all alloy compositions are expressed in wt%.

54

journal of the mechanical behavior of biomedical materials 32 (2014) 52 –61

Table 1 – Chemical compositions (wt%) of Co–Cr–Mo alloys used in present study. Alloy

Co

Cr

Mo

N

Ni

Mn

Si

C

N-free N-doped

Bal. Bal.

28.7 27.8

6.3 6.0

0.001 0.17

o0.01 o0.01

o0.01 o0.01

o0.1 o0.1

0.0016 0.0011

using X-ray diffraction (XRD, PANalytical X0 Pert MPD). The specimens used for the EBSD and XRD measurements were mechanochemically polished with a colloidal silica slurry and subsequently electropolished in a solution of sulfuric acid and methanol (1:9) to remove the surface-worked layer produced by grinding. The development of local strain in the N-free alloy was evaluated using CrossCourt 3 software (BLG Productions). The method relies on the fact that elastic strains and lattice rotations cause small shifts in the zone axis in the EBSD-Kikuchi patterns (Wilkinson et al., 2006a, 2006b). Thus, the small shift in the pattern was measured by comparing a pattern produced from a strained area with that produced at a reference point in a strain-free grain. The reference point for each grain was selected as the point in the grain that exhibited the lowest kernel average misorientation (KAM) in order to determine a suitable measurement point in a strain-free grain. The small shifts in the zone axis in the EBSD-Kikuchi patterns can be geometrically converted into displacement gradient tensors (aij), as follows: aij ¼

∂ui ∂xj

ði; j ¼ 1; 2; 3Þ

ð1Þ

where ui ¼(u1, u2, u3) is the displacement vector at position xi ¼(x1, x2, x3). According to the theory of elasticity (Timoshenko and Goodier, 1970), the following equations are used to determine the strain tensor, eij, and the lattice rotation tensor, wij, from the symmetric and asymmetric parts of aij, respectively: ! 1 ∂ui ∂uj þ eij ¼ ð2Þ 2 ∂xj ∂xi and wij ¼

1 ∂ui ∂uj  2 ∂xj ∂xi

! ð3Þ

The EBSD measurements for the local-strain analysis were performed using a 40-nm step. The EBSD-Kikuchi patterns were recorded at full resolution (1 k  1 k  12 bits deep). The athermal ε-martensite and dislocation substructures were investigated using TEM. The TEM observations were performed on a JEOL JEM-2000EX operated at 200 kV. The specimen used for TEM observations was produced by cutting a 3-mm-diameter disk from each specimen and grinding it with a dimple grinder to form a thin film. Ion-beam milling (Gatan 691 PIPS) was then used to prepare the thin foils.

3.

Results

3.1.

Conventional EBSD analysis

KAM maps of the (a–d) N-free and (e–h) N-doped alloys. Both alloys have equiaxed γ grains containing intragranular boundaries, i.e., athermal ε-martensite plates and annealing-twin boundaries. Note that the N-free alloy exhibits a multivariant ε-martensite structure. KAM was used to measure the magnitude of the residual plastic strain in the grains of the alloys (Fig. 1d and h). KAM represents an average misorientation angle between all adjacent measurement points within a grain and is correlated with the densities of geometrically necessary dislocation (Calcagnotto et al., 2010). In this study, KAM was calculated for up to five neighbor points with a maximum misorientation angle of 21 in accordance with the procedure given by Herrera et al. (2011). The results clearly indicate that the KAM of the N-free alloy is higher than that of the N-doped counterpart, as reported by Yamanaka et al. (2012b), meaning that the γ matrix is plastically deformed when athermal martensitic transformation occurs. The plastic strains must have been introduced in the N-free alloy during quenching, because the N-doped alloy subjected to the same heat treatment shows a much lower KAM. Strain evolution occurs not only inside the ε-martensite phase but also inside the γ phase. The magnified, high-resolution EBSD maps of the N-free alloys are represented in Fig. 2. Fig. 2a, b, and c show IQ, phase, and IPF maps of the as-quenched N-free Co–29Cr–6Mo alloy, respectively. The two γ orientations are represented in Fig. 2. The plate-like products existing inside the γ grains are the athermal ε-martensite phase. The relation between the crystallographic orientations of the γ matrix and the ε-martensite phase was determined from TEM analysis to be the ShojiNishiyama relation: {111}γ//(0001)ε, 〈110〉γ//〈1120〉ε. Fig. 2d shows a KAM map of the γ phase. It should be noted that although some areas have a high KAM, they could not be detected in Fig. 1 because of the limitation of the step size used for the EBSD measurements. The arrows in the IQ and KAM maps indicate contrasting areas that directly pass through the γ matrix and impinge on the ε-martensite plates in the same location. Although we cannot recognize from Fig. 2 whether these contrasting areas originate from the ε-martensite phase, they do form parallel to the traces of the {111}γ planes, as indicated by the red lines in Fig. 2d. Such planar evolution of KAM has not been found in previous studies on steels (Miyamoto et al., 2009; Sato and Zaefferer, 2009) and likely originates from highly planar dislocation structures in the γ phase, as discussed in Section 3.3.

3.2.

Fig. 1(a, e) show image quality (IQ) maps, (b, f) show inverse pole figure (IPF) maps, (c, g) show phase maps, and (d, h) show

EBSD-based local strain analysis

Fig. 3a and b represents the IPF and phase maps of the analyzed area, which include the γ/ε interfaces nearly perpendicular and parallel to the close-packed planes (i.e., the

55

journal of the mechanical behavior of biomedical materials 32 (2014) 52 –61

γ (fcc)

001

111

γ (fcc) ε (hcp)

ε (hcp) 1010

101 0001

0°

2°

2110

50 μm

50 μm

50 μm

50 μm

50 μm

50 μm

50 μm

50 μm

Fig. 1 – EBSD maps for (a–d) N-free and (e–h) N-doped Co–29Cr–6 Mo alloys annealed at 1473 K for 0.6 ks then quenched in water: (a, e) image quality (IQ) maps, (b, f) inverse pole figure (IPF) maps, (c, g) phase maps, and (d, h) kernel average misorientation (KAM) maps.

{111}γ traces

10 μm

10 μm

γ (fcc) ε (hcp)

10 μm γ (fcc)

001

10 μm 111

0°

2°

ε (hcp) 1010

101 0001

2110

Fig. 2 – EBSD maps for N-free Co–29Cr–6Mo alloy annealed at 1473 K for 0.6 ks then quenched in water: (a) image quality (IQ) map where light values indicate high IQ; i.e., low lattice distortions, (b) Phase map, (c) inverse pole figure (IPF) map, and (d) kernel average misorientation (KAM) of γ matrix.

{111}γ and (0001)ε planes) of both the γ and ε phases. Hereafter, these are referred to as “interface 1” and “interface 2,” respectively (see Fig. 3a). Note that the growth direction of the ε-martensite 〈112〉γ phase is not parallel to the x1 axis at interface 1. Fig. 3c–k show the results of the EBSD-based local strain analysis of the γ matrix. Strains within71% are mapped in each figure. The elastic-strain distributions are shown in Fig. 3c–h. Among the normal elastic-strains (Fig. 3c–e),

components e22 and e33 appear to be high at interface 1. These strain components have opposite signs because of Poisson0 s effect. In-plane shear strains e23 and e31 also developed at interface 1 to produce strain distributions along the same direction as the normal strains, e33 and e22, respectively (Fig. 3g and h). Strain e12 increased along another γ/ε interface (i.e., interface 2). The lattice distortion components are represented in Fig. 3i–k. It should be noted that the magnitude of the w31 strain about the x2 axis (Fig. 3k) is apparently

56

journal of the mechanical behavior of biomedical materials 32 (2014) 52 –61

x2

Interface 2 {111}γ traces

x3 γ (fcc)

interface 1

500 nm

500 nm

001

0.01

x1 111

0

ε (hcp) 1010

101 0001

−0.01

2110

e11

e22

e33

e12

e23

e31

w12

w23

w31

Fig. 3 – Elastic strains and lattice rotations measured near athermal ε-martensite plates in N-free Co–29Cr–6Mo alloy annealed at 1473 K for 0.6 ks then quenched in water: (a) inverse pole figure (IPF) map, (b) phase map, (c) e11, (d) e22, (e) e33, (f) e12, (g) e23, (h) e31, (i) w12, (j) w23, and (k) w31.

larger at interface 2, which is parallel to the {111}γ/(0001)ε plane, than it is the grain interior far from the ε plates, as predicted from the KAM analysis (Fig. 2). The development of both the eij and wij strain components is also identified on the {111}γ plane trace far from the γ/ε interface (indicated by arrows in each figure).

3.3.

TEM observations

A representative bright-field TEM image of the γ/ε interfaces is shown in Fig. 4a. The incident beam of the transmission electron microscope was adjusted so that it was aligned with the [110]γ direction. Note that the dislocation structures and ε-martensite phases shown herein were formed during quenching, not by loading. The dislocations in the γ phase are entirely dissociated into Shockley partial dislocations (“partials”) bounding wide stacking faults (SFs), which can be observed as fringe contrasts in TEM, in accordance with

the following equation: a a a ½101γ - ½211γ þ SF þ ½1 12γ 2 6 6

ð4Þ

where a is the lattice constant of the γ phase. The wide SFs mean that single leading Shockley partial dislocations readily pass through the present alloys during cooling, whereas conventional fcc alloys usually exhibit paired Shockley partials bounding narrower SFs after the alloys have undergone solution heat treatment or even slight plastic deformation (Ojima et al., 2009). This is because the stacking fault energy (SFE) of Co–29Cr– 6Mo alloys is very low (Yamanaka et al., 2009, 2012a). Therefore, the present alloys favor the wide separation of SFs to reduce the amount of free energy in the system. The selected-area diffraction (SAD) patterns taken from the regions enclosed by white circles are superimposed in Fig. 4a (the electron diffraction patterns assigned to the ε and γ phases are indicated by the red and blue lines, respectively). The streaks parallel to the

journal of the mechanical behavior of biomedical materials 32 (2014) 52 –61

57

BD//[110]γ //[2110]ε edge-on {111}γ (plane trace) thin ε-layers

500 nm

200 nm

Fig. 4 – (a) TEM bright-field image showing interfaces between γ matrix and ε martensite in N-free Co–29Cr–6Mo alloys annealed at 1473 K for 0.6 ks then quenched in water. Corresponding selected-area diffraction (SAD) patterns taken from regions enclosed by white circles are superimposed in Fig. 4a. (b) Magnified image of γ/ε interface growing through nucleation and densification of thin ε-layers.

[0001]ε (//〈111〉γ) direction are understood to originate from planar defects on the (0001)ε planes, which can be identified in the bright-field image, because the directions of the streaks in the SAD patterns change depending on the ε variant. The ε-martensite plates form on all four {111}γ planes that are possible shear planes, and the plates consequently intersect with each other. Fig. 4b represents the magnified image of Fig. 4a. We can find thin ε-martensite layers (hereafter referred to as “ε-layers”) in addition to the abovementioned faulted ε-martensite plates. It is noteworthy that the ε-layers can also inhibit the glide of Shockley partial dislocations and the growth of ε-layers at their intersections (see Fig. 4b).

4.

Discussion

In the present study, the local strain accumulation in the γ matrix, which is caused by the athermal γ-ε martensitic transformation during quenching, was evaluated using EBSD. The detailed mechanism of local strain evolution is discussed here. It is well accepted that the ε-martensite phase consists of an aggregate of regularly overlapped SFs. The athermal γ-ε martensitic transformation, which results from a change in the stacking sequence of close-packed {111}γ planes from ABCAB… to ABABA…, develops by passing Shockley partials through Burgers vector a/6〈112〉γ on every second {111}γ plane. This means that the ε-martensite phase is produced by the motion of Shockley partials. In other words, the partials glide in advance of not only the formation of the strain-induced martensite phase but also athermal martensitic transformation. It should first be noted that the γ-ε phase transformation involves almost zero change in molar density. From the XRD measurements, the lattice parameters of the present N-free Co–29Cr–6Mo alloy were determined as aγ ¼ 0.3573 nm for the γ phase and aε ¼ 0.2534 nm and cε ¼ 0.4091 nm for the ε phase. The change in volume of the alloy and the magnitude of the strain perpendicular to {111}γ (i.e., along the 〈111〉γ directions)

for the γ-ε phase transformation were therefore calculated as  0.25% and  0.85% (contraction), respectively. Thus, the evolved strains in the γ phase cannot be explained by volumetric changes during martensitic transformation. However, a single-variant ε-martensite phase whose shear strain is 8  1/2 ( ¼ 0.354) is produced in the 〈112〉γ direction when a single pair of partial dislocations slips (Bhadeshia, 2006), and the shear strain remains during deformationinduced martensitic transformation (Bergeon et al., 1998). However, a specimen should undergo zero macroscopic deformation during athermal martensitic transformation. Putaux and Chevalier (1996) therefore developed a model of the nucleation and growth of the ε phase undergoing athermal martensitic transformation, as schematically shown in Fig. 5. In their model, a thin ε-layer containing a correlated periodic array of three types of Shockley partials with different shear directions 〈112〉γ forms in a (111)γ plane to locally cancel the shear strains of each other (Fig. 5a). The ε-martensite plates are believed to develop from the successive nucleation of the new ε-layers in close proximity to the preexisting ones (Fig. 5b). However, the thickening of the SFs that are already formed is unfavorable because ε-layers that are only a few atomic planes thick are advantageous to minimize the amount of elastic strain energy (Bergeon et al., 1998; Putaux and Chevalier, 1996). The γ phase consequently remains between the ε-layers and forms planar defects in the ε-martensite phase, as observed in various steels (Putaux and Chevalier, 1996). We can actually find isolated thin ε-layers and highly faulted ε-martensite phases that fit this hypothesis (Fig. 4), and we recognize that this model is also reasonable for Co–Cr–Mo alloys. Therefore, the magnitude of the shear strain, theoretically 8  1/2, due to the martensitic transformation would be reduced in accordance with this mechanism. This is why the elastic and plastic strains measured using EBSD have much smaller magnitudes than the theoretical strains (Fig. 3). All four {111}γ planes in the fcc structure are possible shear planes, and each of them has three 〈112〉γ directions. Activating

58

journal of the mechanical behavior of biomedical materials 32 (2014) 52 –61

[101] Shockley partial dislocation Stacking fault

1/6[211] 1/6[112] 1/6[121]

1/6[121]

1/6[211]

thin ε-layer

1/6[112] 1/6[121]

thin ε-layer

Grain boundaries Nucleation

Densification

Faulted martensite

Fig. 5 – (a) Slip system of Shockley partial dislocations forming zero-macroscopic-strain athermal ε martensite, and (b) schematic illustration showing development of athermal ε-martensite plate through successive nucleation of thin ε-layers.

the slip system of Shockley partials numerous times consequently leads to a multivariant ε-martensite structure, as observed in Figs. 1 and 4a, and the Shockley partials and the subsequently formed athermal ε-martensite plates ultimately intersect with each other. In this study, the elastic strains and lattice distortions were separately evaluated using the EBSD-based method. No dislocations are generated by the impingement of leading Shockley partial dislocations in the vicinities of the γ/ε interfaces (Fig. 4a). Thus, this interaction should mainly be understood to produce elastic strains (indeed, eij was detected in Fig. 3c–h). Note that elastic strains cannot be detected using conventional EBSD analysis. A similar phenomenon would occur when the dislocations and/or ε-layers encounter the thin ε-layers, because TEM observations clearly revealed that the ε-layers stopped the glide or growth of the partial dislocations and the ε-layers at points where they intersected. However, both elastic and plastic strain components develop on the {111}γ planes in some areas (indicated by the arrow in Fig. 3, for example). The ε-martensite phase in these areas could not be detected using EBSD. However, TEM observations (Fig. 4) showed numerous SFs/ε-layers as the precursors of the ε-martensite plates inside the γ matrix. Of course, the glide of Shockley partial dislocations causes local plastic strains (Fig. 3i–k). In addition, the subsequent self-formation of the ε-martensite phase would produce lattice curvature, since the ε-martensite phase is formed through the collective motions of Shockley partial dislocations. The wij sometimes originate from the elastic deformation that changes both components

(Miyamoto et al., 2009). However, we can identify a high KAM region in Fig. 2d and a slip trace in the IQ map (Fig. 2a). These two findings taken together clearly indicate that the lattice distortion was due to dislocation activity. We emphasize that the impingement between the ε-martensite plates plays a crucial role in elastic strain evolution in the γ matrix. The ε-martensite plates act as strong obstacles against the motion of perfect dislocation and Shockley partial dislocations (Sato et al., 1982). It is especially difficult for the slip transfer of gliding Shockley partial dislocations at γ/ε interfaces to operate as a stress concentration relaxation process in Co–Cr–Mo alloys. Matsumoto et al. (2009) have previously reported that the ε phase in a polycrystalline Co–27Cr–5Mo alloy deformed through basal 〈a〉 and prismatic 〈a〉 slips. Since the axial ratio (c/a) of the ε phase in the present alloy (¼ 1.614) is almost identical to that reported in their study (¼1.610; Matsumoto et al., 2010), neither nonbasal 〈cþa〉 slips nor deformation twining, which produce deformation along the c-axis in the hcp structure, would be activated. In addition, there is no evidence in support of reactions at intersections between the ε-martensite and ε-martensite/Shockley partial dislocations (e.g., nucleation of the γ phase or twinning in the ε phase (Zhang et al., 2011)). Thus, the preexisting ε-martensite phase can completely block the gliding of partial dislocations, {111}γ〈112〉γ, and the growth of the ε-martensite phase. The very low SFE of the present Co–CrMo alloys should also be emphasized. It becomes negative below  1123 K (Yamanaka et al., 2009, 2012a), which simply means that the ε phase is thermodynamically stabler than the γ phase (Yamanaka et al., 2012a). Therefore, the SFs

journal of the mechanical behavior of biomedical materials 32 (2014) 52 –61

Shockley partial dislocation

Stacking fault

{111}γ Dislocation and/or thin ε-layers

impingement

Preexisting ε

thin ε-layers

Origin of strain evolution in γ matrix (a) interaction between ε martensite (b) dislocations that impinge on ε martensite (c) interaction among dislocations and thin ε-layers (d) glide of Shockley partial dislocations/ growth of ε martenstie

Fig. 6 – Schematic illustration showing interactions among preexisting ε plate, growing ε martensite, and gliding Shockley partial dislocations.

bounded by the Shockley partial dislocations introduced in the present alloys favor wide separation and form the ε-martensite phase, as shown in Fig. 4. This kind of dislocation structure strongly suppresses cross slip, and the extended planar dislocations impinged against grain boundaries and the γ/ε interfaces would introduce significant stress concentration. Namely, the low SFE must further enhance strain evolution. Fig. 6 shows the schematic illustration of microstructure evolution during athermal γ-ε martensitic transformation. The wide SFs should be formed during quenching because the thermodynamic stability of the ε phase becomes higher than that of the γ phase with decreasing temperature (i.e., during quenching). Thus, the only leading partials that impinge against the ε-martensite plate and/or the thin ε-layers are displayed in Fig. 6. Using multiple slipping of partial dislocations to ensure zero macroscopic deformation causes the interactions among a preexisting ε-martensite plate, a growing ε-martensite phase, and gliding Shockley partial dislocations. These interactions result in the development of elastic strain in the γ phase. The highly extended dislocations and resulting formation of the ε-martensite phase, which are enhanced by the very low SFE of the alloy, must increase the magnitude of the elastic strain in the vicinity where they impinge on each other when athermal martensitic transformation occurs. The results of the present study suggest that dislocations that act as precursors of the ε-martensite phase are the origin of the strain evolution discussed in this article. In addition, we have recently identified short-range ordering (SRO) between Cr and N atoms or Cr2N nanoprecipitates in the γ matrix of the N-containing alloys (Yamanaka et al., 2013). These nanostructural inhomogeneities were revealed to function as obstacles to the gliding of partial dislocations (and consequently, they significantly affect the kinetics of the

59

γ-ε martensitic transformation). Thus, adding nitrogen to the alloys affects the activity of dislocations and the γ-ε martensitic transformation and would reduce the magnitude of the plastic strain induced during quenching. Dislocation reactions, e.g., the Lomer-Cottrell dislocation reaction, may similarly contribute to the strain evolution. The results of the present study indicate that the activity of dislocations and the resulting interactions between lattice dislocations and the ε-martensite phase would result in a highly strained microstructure (Yamanaka et al., 2012b). The “self-hardening” and grain refinement effects of the γ/ε multiphase structures must increase the alloy strength (Yamanaka et al., 2014). This is also very important for improving the wear resistance of the alloys because the kinetics of strain-induced martensitic transformation, which facilitates the wear resistance of Co–Cr–Mo alloys (Chiba et al., 2007), depends on how dislocations behave in the γ matrix. In addition, a similar hardening phenomenon; that is, early nucleation of microcracks and fracture occurring during tensile loading, would occur during plastic deformation even where strain-induced ε martensite is involved, as observed in previous studies (Mori et al., 2010; Kurosu et al., 2011; Lee et al., 2011).

5.

Conclusions

In this study, EBSD and TEM were used to investigate the microstructural evolution caused by athermal γ-ε martensitic transformation in biomedical Co–29Cr–6Mo alloys. The N-free alloy that had a γ/ε duplex microstructure exhibited a higher residual plastic strain in the γ phase than that in the N-doped alloy that did not have an athermal εε-martensite phase. The shear strain due to the martensitic transformation was accommodated by operating a multivariant structure derived from a system of numerous {111}γ〈112〉γ Shockley partial dislocations. The slip activity itself, as a precursor of the martensitic transformation, produced plastic strain. In addition, the present study revealed the development of elastic strain at the γ matrix/ε-martensite interfaces and along the traces of the {111}γ planes. The highly planar dislocation structures consisting of wide SFs must further enhance strain evolution by suppressing relaxation.

Acknowledgments The authors would like to thank Shun Ito and Sho Suzuki for their assistance with TEM observations and EBSD analysis. This research was financially supported by Grant-in-Aid for JSPS Fellows, the Global COE Program “Materials Integration (International Center of Education and Research), Tohoku University” and by a grant from the Regional Innovation Cluster Program of the Ministry of Education, Culture, Sports, Science, and Technology (MEXT), Japan.

r e f e r e nc e s

Battini, E., Eriksson, T., Bostro¨m, M., Leygraf, C., Pan, J., 2011. Influence of metal carbides on dissolution behavior of

60

journal of the mechanical behavior of biomedical materials 32 (2014) 52 –61

biomedical CoCrMo alloy: SEM, TEM and AFM studies. Electrochim. Acta 56, 9413–9419. Bergeon, N., Guenin, G., Esnouf, C., 1998. Microstructural analysis of the stress-induced ε martensite in a Fe–Mn–Si–Cr–Ni shape memory alloy: part I – calculated description of the microstructure. Mater. Sci. Eng. A 242, 77–86. Bhadeshia, H.K.D.H., 2006. Worked Examples in the Geometry of Crystals, Second ed. The Institute of Metals, Brookfield. Buford, A., Goswami, T., 2004. Review of wear mechanisms in hip implants: paper I – general. Mater. Des. 25, 385–393. Calcagnotto, M., Ponge, D., Demir, E., Raabe, D., 2010. Orientation gradients and geometrically necessary dislocations in ultrafine grained dual-phase steels studied by 2D and 3D EBSD. Mater. Sci. Eng. A 527, 2738–2746. Caudillo, M., Herrera–Trejo, M., Castro, M.R., Ramı´rez, E., Gonza´lez, C.R., Jua´rez, J.I., 2002. On carbide dissolution in an as-cast ASTM F-75 alloy. J. Biomed. Mater. Res. 59, 378–385. Chiba, A., Li, X.G., Kim, M.S., 1999. High work-hardening rate and deformation twinning of Co–Ni-based superalloy at elevated temperatures. Philos. Mag. A 79, 1533–1554. Chiba, A., Kumagai, K., Nomura, N., Miyakawa, S., 2007. Pin-ondisk wear behavior in a like-on-like configuration in a biological environment of high carbon cast and low carbon forged Co–29Cr–6Mo alloys. Acta Mater. 55, 1309–1318. Denkhaus, E., Salnikow, K., 2002. Nickel essentiality, toxicity, and carcinogenicity. Crit. Rev. Oncol./Hematol. 42, 35–56. Herrera, C., Ponge, D., Raabe, D., 2011. Design of a novel Mn-based 1 GPa duplex stainless TRIP steel with 60% ductility by a reduction of austenite stability. Acta Mater. 59, 4653–4664. Huang, P., Lo´pez, H.F., 1999. Athermal ε-martensite in a Co–Cr–Mo alloy: grain size effects. Mater. Lett. 39, 249–253. Karamched, P.S., Wilkinson, A.J., 2011. High resolution electron back-scatter diffraction analysis of thermally and mechanically induced strains near carbide inclusions in a superalloy. Acta Mater. 59, 263–272. Kluess, D., Martin, H., Mittelmeier, W., Schmitz, K.P., Bader, R., 2007. Influence of femoral head size on impingement, dislocation and stress distribution in total hip replacement. Med. Eng. Phys. 29, 465–471. Koizumi, Y., Suzuki, S., Yamanaka, K., Lee, B.S., Sato, K., Li, Y.P., Kurosu, S., Matsumoto, H., Chiba, A., 2013. Strain-induced martensitic transformation near twin boundaries in biomedical Co–Cr–Mo alloy with negative stacking fault energy. Acta Mater. 61, 1648–1661. Kurosu, S., Matsumoto, H., Chiba, A., 2010. Isothermal phase transformation in biomedical Co–29Cr–6Mo alloy without addition of carbon or nitrogen. Metall. Mater. Trans. A 41A, 2613–2625. Kurosu, S., Matsumoto, H., Chiba, A., Landron, C., Fabregue, D., Maire, E., 2011. The damage process in a biomedical Co–29Cr– 6Mo–0.14N alloy analyzed by X-ray tomography and electron backscattered diffraction. Scr. Mater. 64, 367–370. Lee, B.S., Matsumoto, H., Chiba, A., 2011. Fractures in tensile deformation of biomedical Co–Cr–Mo–N alloys. Mater. Lett. 65, 843–846. Lee, S.H., Takahashi, E., Nomura, N., Chiba, A., 2006. Effect of carbon addition on microstructure and mechanical properties of a wrought Co–Cr–Mo implant alloy. Mater. Trans. 47, 287–290. Lee, S.H., Nomura, N., Chiba, A., 2008. Significant improvement in mechanical properties of biomedical Co–Cr–Mo alloys with a combination of N addition and Cr enrichment. Mater. Trans. 49, 260–264. Liao, Y., Pourzal, R., Stemmer, P., Wimmer, M.A., Jacobs, J.J., Fischer, A., Marks, L.D., 2012. New insights into hard phases of

CoCrMo metal-on-metal hip replacements. J. Mech. Behav. Biomed. Mater. 12, 39–49. Mani, A., Salinas-Rodriguez Lo´pez, H.F., 2011. Deformation induced FCC to HCP transformation in a Co–27Cr–5Mo–0.05C alloy. Mater. Sci. Eng. A 528, 3037–3043. Marrey, R.V., Burgermeister, R., Grishaber, R.B., Ritchie, R.O., 2006. Fatigue and life prediction for cobalt–chromium stents: a fracture mechanics analysis. Biomaterials 27, 1988–2000. Matsumoto, H., Kurosu, S., Lee, B.S., Li, Y.P., Chiba, A., 2010. Deformation mode in biomedical Co–27% Cr–5% Mo alloy consisting of a single hexagonal close-packed structure. Scr. Mater. 63, 1092–1095. Mineta, S., Namba, S., Yoneda, T., Ueda, K., Narushima, T., 2010. Carbide formation and dissolution in biomedical Co–Cr–Mo alloys with different carbon contents during solution treatment. Metall. Mater. Trans. A 41A, 2129–2138. Miyamoto, G., Shibata, A., Maki, T., Furuhara, T., 2009. Precise measurement of strain accommodation in austenite matrix surrounding martensite in ferrous alloys by electron backscatter diffraction analysis. Acta Mater. 57, 1120–1131. Mori, M., Yamanaka, K., Matsumoto, H., Chiba, A., 2010. Evolution of cold-rolled microstructures of biomedical Co–Cr–Mo alloys with and without N doping. Mater. Sci. Eng. A 528, 614–621. Mori, M., Yamanaka, K., Sato, S., Wagatsuma, K., Chiba, A., 2012. Microstructures and mechanical properties of biomedical Co–Cr–Mo alloys processed by hot rolling. Metall. Mater. Trans. A 43A, 3108–3119. Nagai, A., Tsutsumi, Y., Suzuki, Y., Katayama, K., Hanawa, T., Yamashita, K., 2012. Characterization of air-formed surface oxide film on a Co–Ni–Cr–Mo alloy (MP35N) and its change in Hanks0 solution. Appl. Surf. Sci. 258, 5490–5498. Niinomi, M., 2002. Recent metallic materials for biomedical applications. Metall. Mater. Trans. A 33, 477–486. Ojima, M., Adachi, Y., Tomota, Y., Ikeda, K., Kamiyama, T., Katada, Y., 2009. Work hardening mechanism in high nitrogen austenitic steel studied by in situ neutron diffraction and in situ electron backscattering diffraction. Mater. Sci. Eng. A 527, 16–24. Putaux, J.L., Chevalier, J.P., 1996. HREM study of selfaccommodated thermal ε-martensite in an Fe–Mn–Si–Cr–Ni shape memory alloy. Acta Mater. 44, 1701–1716. Rajan, K., 1982. Phase transformation in a wrought Co–Cr–Mo alloy. Metall. Trans. A 13A, 1161–1166. Salinas-Rodriguez, A., Rodriguez-Galicia, J.L., 1996. Deformation behavior of low-carbon Co–Cr–Mo alloys for low-friction implant applications. J. Biomed. Mater. Res. 31, 409–419. Sato, A., Soma, K., Mori, T., 1982. Hardening due to pre-existing ε-martensite in an Fe–30Mn–1Si alloy single crystal. Acta Metall. 30, 1901–1907. Sato, H., Zaefferer, S., 2009. A study on the formation mechanisms of butterfly-type martensite in Fe–30% Ni alloy using EBSD-based orientation microscopy. Acta Mater. 57, 1931–1937. Song, C.B., Park, H.B., Seong, H.G., Lo´pez, H.F., 2006. Development of athermal ε-martensite in atomized Co–Cr–Mo–C implant alloy powders. Acta Biomater. 2, 685–691. Timoshenko, S.P., Goodier, J.N., 1970. Theory of Elasticity, third ed. McGraw-Hill, New York. Wilkinson, A.J., Meaden, G.B., Dingley, D.J., 2006a. High-resolution elastic strain measurement from electron backscatter diffraction patterns: new levels of sensitivity. Ultramicroscopy 106, 307–313. Wilkinson, A.J., Meaden, G.B., Dingley, D.J., 2006b. High resolution mapping of strains and rotations using electron backscatter diffraction. Mater. Sci. Technol. 22, 1271–1278. Yamanaka, K., Mori, M., Kurosu, S., Matsumoto, H., Chiba, A., 2009. Ultrafine grain refinement of biomedical Co–29Cr–6Mo

journal of the mechanical behavior of biomedical materials 32 (2014) 52 –61

alloy during conventional hot-compression deformation. Metall. Mater. Trans. A 40A, 1980–1994. Yamanaka, K., Mori, M., Chiba, A., 2011. Mechanical properties of as-forged Ni-free Co–29Cr–6Mo alloys with ultrafine-grained microstructure. Mater. Sci. Eng. A 528, 5961–5966. Yamanaka, K., Mori, M., Chiba, A., 2012a. Origin of significant grain refinement of biomedical Co–29Cr–6Mo alloy without severe plastic deformation. Metall. Mater. Trans. A 43A, 4875–4887. Yamanaka, K., Mori, M., Chiba, A., 2012b. Enhanced mechanical properties of as-forged Co–Cr–Mo–N alloys with ultrafinegrained structures. Metall. Mater. Trans. A 43A, 5243–5257.

61

Yamanaka, K., Mori, M., Chiba, A., 2013. Nanoarchitectured Co–Cr–Mo alloys: nitrogen-enhanced nanostructural evolution and its effect on phase stability. Acta Biomater. 9, 6259–6267. Yamanaka, K., Mori, M., Chiba, A., 2014. Effects of nitrogen addition on microstructure and mechanical behavior of biomedical Co–Cr–Mo alloys. J. Mech. Behav. Biomed. Mater. 29, 417–426. Zhang, X., Sawaguchi, T., Ogawa, K., Yin, F., Zhao, X., 2011. A structure created by intersecting ε martensite variant plates in a high-manganese steel. Philos. Mag. 91, 4410–4426.