- Email: [email protected]
The influence of Pd on tension–tension fatigue behavior of Zr-based bulk-metallic glasses
International Journal of Fatigue 32 (2010) 599–604
Contents lists available at ScienceDirect
International Journal of Fatigue journal homepage: www.elsevier.com/locate/ijfatigue
The influence of Pd on tension–tension fatigue behavior of Zr-based bulk-metallic glasses G.Y. Wang a,*, P.K. Liaw a, Y. Yokoyama b, M. Freels a, A. Inoue b a b
Department of Materials Science and Engineering, The University of Tennessee, Knoxville, TN 37996, USA Advanced Research Center of Metallic Glasses, Institute for Materials Research, Tohoku University, Sendai 980-8577, Japan
a r t i c l e
i n f o
Article history: Received 1 December 2008 Received in revised form 14 April 2009 Accepted 15 April 2009 Available online 22 April 2009 Keywords: Bulk-metallic glass Amorphous High cycle fatigue S–N curves Fatigue limit
a b s t r a c t Zr-based bulk-metallic glasses (BMGs) are being studied widely because Zr-based BMGs exhibit good glass-forming abilities and excellent properties, such as material strengths. In the current paper, the fatigue behaviors of Zr50Cu40xAl10Pdx [x: 0–7 atomic percent (at.%)] BMGs were investigated. The uniaxial tension–tension fatigue experiments were performed on the button-head rod fatigue specimens. The test environment was air at room temperature. The fatigue limit of Zr50Cu37Al10Pd3 was found to be the highest with a value of 945 MPa among the BMGs studied. A mechanistic understanding of the fatigue behavior of these Zr-based BMGs is suggested. The effect of the Pd content on the fatigue behavior was analyzed. A possible relationship between the fatigue limit (or the fatigue ratio) and the volume change, which probably corresponds to excessive free volume, was developed. Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction As excellent candidates for structural materials, bulk-metallic glasses (BMGs) are being studied extensively [1.2]. Because the fatigue behavior is a very important characteristic for the application of structural materials, the fatigue investigation of BMGs attracts attention [3–20]. However, the understanding of the fatigue behavior of BMGs is still limited. Gilbert et al. [3,4] were the first to report the fatigue results of BMGs. They performed four-point-bend fatigue experiments on Vitreloy 1 [Zr41.2Ti13.8Cu12.5Ni10Be22.5 (in atomic percent, at.%)] beam specimens and claimed that the fatigue limits, based on the stress range, were approximately 8% of the ultimate tensile strength. Menzel and Dauskardt also obtained similar fatigue results [5]. These values are very low compared to conventional crystalline alloys, such as high-strength steels, copper, and aluminum alloys, whose fatigue limits are typically 30– 40% of the ultimate tensile strength. On the other hand, tension–tension fatigue results of taper notched LM001 BMG samples that have the same composition as Vitreloy 1 revealed that the fatigue limits, based on the stress range, were 31% of the ultimate tensile strengths [6]. In addition, Peter et al. [7,8] and Wang et al. [9–13] performed tension–tension fatigue tests on notched Zr-based BMG samples, and Yokoyama et al. [14] did rotating-beam fatigue experiments on Zr-based BMGs. They found fatigue limits as high as 40–50% of the ultimate * Corresponding author. Tel.: +1 865 974 0245; fax: +1 865 974 4115. E-mail address: [email protected] (G.Y. Wang). 0142-1123/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijfatigue.2009.04.009
tensile strength. Recently, Nakai et al. conducted fatigue tests on smooth Zr-based BMGs specimens under fully reversed cyclic loading and reported that the fatigue limit, based on the stress amplitude, is 26% of the ultimate tensile strength [15]. According to these fatigue studies of Zr-based BMGs, it was found that Zr-based BMGs with various compositions exhibited very different fatigue lives. What caused such a large difference among these fatigue results of BMGs? Many factors could be involved, such as the material quality, mean stress, specimen geometry, chemical environment, temperature, cyclic frequency, residual stress, and surface condition. In fact, any processing that changes the static mechanical properties or microstructure will probably also affect the fatigue behavior of materials. However, some of factors mentioned above must play an important role in affecting the fatigue behavior of BMGs. The formation of shear bands is still unclear during the cyclic deformation of BMGs. How the fatigue crack initiates and propagates in metallic glasses needs to be solved. Thus, it is essential to perform the fundamental research on the fatigue behavior of BMGs. Zr50Cu40Al10 shows good mechanical behavior, and the fatigue behavior can be improved. In order to strengthen the fatigue behavior of Zr50Cu40Al10 BMGs, in general, some small additive elements like Ni and Pd were added into this alloy system [14]. In the current paper, uniaxial tension–tension fatigue experiments on the Zr50Cu40xAl10Pdx [x: 0–7 atomic percent (at.%)] were performed. The applied stress versus fatigue cycle (S–N) curves of these BMGs is presented. The fracture surfaces were observed and analyzed using scanning-electron microscopy (SEM). In addition, the factors
600
G.Y. Wang et al. / International Journal of Fatigue 32 (2010) 599–604
Fig. 1. Specimen geometry for tension–tension fatigue experiments (unit: mm).
which influence the fatigue limit were discussed. Thus, the effect of the Pd content on the fatigue behavior of BMGs was clarified. 2. Experimental procedures Zr50Cu40xAl10Pdx (x: 0–7 at.%) BMGs were the ternary and quaternary alloys used in the present experiments. The ladle-hearth type arc-melt tilt-casting technique was employed to manufacture these BMGs. The master-alloy ingots were prepared by arc-melting mixtures of pure Zr, Cu, Al, and Pd metals in an argon atmosphere. A special Zr-crystal rod [<0.05 at.% oxygen] was employed in order to maintain the low-oxygen concentration of the alloys [21,22]. The tilt-casting technique has an advantage to control the molten alloy flow to restrict the formation of cold shuts, which probably induces early fatigue–crack initiation and propagation behavior [21,22]. Rod BMG ingots were produced with a length of 60 mm and a diameter of 8 mm. One taper notched specimen (Fig. 1) was machined from each ingot. The fatigue samples were, then, polished to minimize surface effects. The thermal properties of these BMG samples were measured in a Perkin–Elmer Diamond differential scanning calorimeter (DSC) at a heating rate of 20 K/min. The weight of samples used for DSC was in the range of 30–50 mg. The previous research result revealed a decrease in the free volume by annealing a BMG [23]. The free-volume changes could result in the variation of the BMG volume. Therefore, the volume change due to the structural relaxation at the glass-transition temperature (Tg) could be a convenient method to evaluate the freevolume difference in the glass structure. The volume change is defined as the volume change ratio of BMGs from the as-cast state to the annealed state at Tg. The following equation is used to determine the value of the volume change [24]:
DV ¼
1 q1 0 qTg 1 q0
where DV is the volume change, q0 is the density of the as-cast alloy, and qTg is the density of the alloy after annealing at Tg for 90 min. A computer-controlled Material Test System (MTS) servohydraulic testing machine was employed for fatigue studies. The machine was aligned prior to use, and as required. Samples were tested at various stress ranges with a R ratio (R = rmin/rmax, where rmin and rmax are the applied minimum and maximum stresses, respectively) of 0.1 under a load-control mode, using a sinusoidal waveform at a frequency of 10 Hz. Upon failures or 107 cycles, samples were removed and stored for later examinations by SEM. The fracture surfaces of selected specimens were examined, using a Leo 1526 SEM machine with the energy-dispersive spectroscopy (EDS) to provide fatigue and fracture mechanisms. 3. Results and discussion Fig. 2 shows the DSC thermograms obtained from the as-cast fatigue samples during continuous heating at a heating rate of 20 K/ min. An endothermic reaction, corresponding to the transition from a glassy state to a supercooled liquid state, and the following exothermic reaction, corresponding to crystallization are clearly observed. These parameters are labeled as the glass-transition temperature, Tg, and crystallization temperature, Tx, respectively. The characteristic temperatures, as well as heats of crystallization (DHx), are given in Table 1. In addition, the supercooled liquid region DTx = Tx Tg) was also calculated, as shown in Table 1. DTx may indicate that the larger the accessible supercooled liquid region is, the greater the glass-forming ability is [25]. The value of Tg increases with increasing the Pd content in these BMGs. However, the value of Tx and DHx exhibit slight decreases when the Pd content increases from 0% to 7% in these glassy alloys. The increase of Pd resulted in a slight decrease of the supercooled liquid region DTx), which may indicate that Pd has the negative effect on
Table 1 Glass-transition temperature (Tg), crystallization temperature (Tx), and heat of crystallization (DHx), as well as a supercooled liquid region (DTx = Tx Tg) for the Zr50Cu40xAl10Pdx (x: 0–7 at.%) alloys, as obtained from DSC at a heating rate of 20 K/ min.
Fig. 2. DSC thermograms of the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy alloys. The heating rate was 20 K/min. Locations of the characteristic temperatures (Tg and Tx) are indicated by arrows.
Materials
Tg (°C)
Tx (°C)
DHx (J/g)
DTx (°C)
Zr50Cu40Al10 Zr50Cu39Al10Pd1 Zr50Cu38Al10Pd2 Zr50Cu37Al10Pd3 Zr50Cu35Al10Pd5 Zr50Cu33Al10Pd7
417 428 431 431 434 435
507 507 507 507 503 504
43 43 39 40 34 33
90 79 76 76 69 69
G.Y. Wang et al. / International Journal of Fatigue 32 (2010) 599–604
Fig. 3. Stress-range/fatigue-life curves of the Zr50Cu40xAl10Pdx (x: 0–7 at.%) specimens tested with R = 0.1 at a frequency of 10 Hz in air. Note that the fatigue data of Zr50Cu40Al10 and Zr50Cu37Al10Pd3 alloys was obtained from the sharpnotched samples [9,13].
the glass-forming ability among Zr50Cu40xAl10Pdx (x: 0–7 at.%) BMGs. The S–N curves of the tension–tension fatigue results for the Zr50Cu40xAl10Pdx (x: 0–7 at.%) samples tested in air with R = 0.1 and a frequency of 10 Hz were presented in Fig. 3. Note that the fatigue data of Zr50Cu40Al10 and Zr50Cu37Al10Pd3 alloys was obtained from the sharp-notched samples [9,13] and the stress-concentration factor (Kt) is considered to be 1.49 at the notched section for these sharp-notched specimens based on the finite-element analysis [26]. New taper notched samples (Kt 1) were employed to avoid the stress-concentration-factor effect. The value of the stress range for these BMGs in the fatigue-life range of 103–105 cycles decreased significantly with increasing the cycles to failure. For these BMGs, the lifetime above the fatigue limit demonstrated no large differences. Nevertheless, the fatigue limits (rL), based on the applied stress range, for the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy alloys subjected to tension–tension loading exhibited clear differences, increasing from 537 to 945 MPa with the variation of the Pd content. The fatigue ratios (rL divided by the tensile strength) for these BMGs also improved obviously from 0.28 to 0.50 by changing the Pd content. The tensile strength, fatigue limit, and fatigue ratio, as well as the hardness, for the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy alloys were given in Table 2. It was obviously found that the tensile strength and hardness increased slightly with increasing the Pd content among the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy-alloy system [14]. However, the fatigue limits of the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy alloys showed a distinct trend, as seen in Fig. 4. The fatigue limits of the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy alloys decrease slightly when Pd content increases from 1% to 2% and, then, increase with increasing the Pd Table 2 Hardness, tensile strength, fatigue limit, and fatigue ratio for the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy alloys. Materials
Hardness (HV) [14]
Tensile strength (MPa) [14]
Fatigue limit (MPa)
Fatigue ratioa
Zr50Cu40Al10 Zr50Cu39Al10Pd1 Zr50Cu38Al10Pd2 Zr50Cu37Al10Pd3 Zr50Cu35Al10Pd5 Zr50Cu33Al10Pd7
506 505 504 503 512 528
1851 1909 1911 1899 1929 1952
723 597 537 945 608 776
0.39 0.31 0.28 0.50 0.32 0.40
a
Fatigue ratio = Fatigue limit/tensile strength.
601
Fig. 4. The relationship between the fatigue limit (fatigue ratio) and the Pd content among the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy alloys. Hollow points are from sharp-notched samples and solid points are from taper notched samples.
content. The fatigue ratios of these BMGs have the same trend as the fatigue limits. Why does the addition of Pd have slight effects on the thermal properties and static mechanical properties of Zr50Cu40xAl10Pdx (x: 0–7 at.%) in Tables 1 and 2, while it has obvious effects on the fatigue behavior of Zr50Cu40xAl10Pdx (x: 0–7 at.%) in Table 2? There are some important factors that influence the fatigue behavior. The deformation characteristic of metallic glasses can generally be classified as either inhomogeneous or homogeneous. The homogeneous deformation often occurs at high temperatures (e.g., T > 0.70 Tg), and the metallic glasses can demonstrate significant plasticity, especially in the supercooled liquid region [27]. The inhomogeneous deformation usually occurs when a metallic glass is deformed at low temperatures (e.g., room temperature) and is characterized by the formation of localized shear bands, followed by the rapid propagation of these bands, and sudden fracture. Typically, these shear bands are 10–20 nm in width [28]. A large amount of plastic strain occurs at these localized shear regions. However, the entire plastic deformation of the specimen is generally very low (2–3%) [29]. Two primary hypotheses are presented in the literature to explain the plastic flow and localized shear bands in metallic glasses [30]. The first hypothesis suggests that the viscosity in the shear bands decreases due to the formation of free volumes [31,32]. The second hypothesis suggests that the viscosity in the shear bands decreases due to the generation of the local adiabatic heating [33]. The generally accepted view on the inhomogeneous deformation in metallic glasses is the free-volume theory. Argon thought that the shear transformation at high stresses in metallic glasses narrowed down to a region between two short rows of 4–6 atoms around a free-volume site and resulted in the formation of strong shear localization, giving rise to the shear bands [30]. During cyclic experiments, it was also found that cracks initiated from shear bands [5,9,10,15]. This fact implied that the shear transformation confined into small regions near free-volume sites could produce the shear localization that lead to the formation of the shear bands when the metallic glasses were subjected to a repeated stress. It is suggested that the fatigue behavior could be related to the free volume in metallic glasses. Note that fatigue cracks grow unimpeded once they initiated due to the absence of microstructural barriers in metallic glasses [5]. Since the fatigue limit means that the fatigue sample does not fracture when the applied stress is below this value, it could also suggest that the fatigue sample does not initiate a crack in the case of metallic glasses. Therefore, it is possible that there is a relationship between the fatigue limit and the free volume in metallic glasses.
602
G.Y. Wang et al. / International Journal of Fatigue 32 (2010) 599–604
Fig. 5. The relationship between the fatigue limit (fatigue ratio) and volume change among the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy alloys. Hollow points are from sharp-notched samples and solid points are from taper notched samples.
Metallic Glasses are solidified at high cooling rates below the glass-transition temperature. Thus, the free volume in a glassy alloy may change with the cooling rate, and the quantity of free volumes may also be changed by the alloy composition [24]. However, it is difficult to determine the absolute quantity of the free volume in a metallic glass. Therefore, the volume change is employed to evaluate the free-volume variation in the metallic glass [24]. Fig. 5 exhibited the relationship of the fatigue limit and the fatigue ratio with the volume change in the Zr50Cu40xAl10Pdx (x: 0–7 at.%)
glassy alloys. It seems that there is a possible trend between the fatigue limit (or the fatigue ratio) and the volume change, in which the fatigue limit increases with increasing the volume change. The larger value of the volume change could suggest more free-volume sites in metallic glasses and the existence of more free-volume sites could allow atoms to move more easily in metallic glasses. Thus, more shear bands could form and propagate easily. As a result, these shear bands disperse the localized stress. This trend could reduce the probability of the stress concentration during cyclic loading. Thus, it is difficult to initiate a fatigue crack, and the metallic glass could show a higher fatigue limit with increasing the volume change (Fig. 5). Yokoyama et al. [14] thought that the volume change indicated thermally unstable volumes existed below Tg in the amorphous structure, and it was generally assumed that the amorphous structure was composed of clusters and free volumes. Thermally unstable free volumes were probably effective to relax the localized stress because the atoms surrounding the void moved more easily than others, and the stress relaxation was beneficial to fracture and fatigue behavior [14]. In addition, Suh and Dauskardt [34] found that the plane-strain fracture toughness of a Zr–Ti–Ni–Cu–Be bulk-metallic glass was significantly decreased after annealing at 300 °C. They thought the anneal-out of free volumes resulted in the retarded atomic arrangement process for viscous flows, which led to the loss of the stress relief by viscous flows at the crack tip. As a result, the annealing embrittlement was observed [34]. In fact, it could be possible that the decrease in free volumes made it difficult for shear bands to produce and propagate so that the significant stress concentration occurred at the crack tip, which could generally result in lower fatigue limits with decreasing volume change (Fig. 5).
Fig. 6. (a) Overall fatigue fractography of the Zr50Cu38Al10Pd2 specimen tested at rmax = 663 MPa; (b) crack-initiation site; (c) boundary between the crack-growth region and final fast fracture regions and (d) local melting phenomena.
G.Y. Wang et al. / International Journal of Fatigue 32 (2010) 599–604
The volume change was assumed to be the degree of volume expansion from the fully relaxed state and could only indicate the relative quantity of free-volume sites. This situation could be one of the reasons why the linear relationship between the fatigue limit (the fatigue ratio) and the volume change is not so strong in metallic glasses (Fig. 5). Other factors, such as the surface condition, sample quality, and residual stress, could also affect the fatigue limit. Therefore, complex factors need to be considered during studying the fatigue behavior of BMGs. The fractured surface of a Zr50Cu38Al10Pd2 BMG sample with SEM observations after tension–tension fatigue experiments is shown in Fig. 6. It is found that the whole fatigue-fracture surface clearly includes the fatigue–crack initiation, crack propagation, final fast fracture, and apparent melting areas. The fracture surface is basically perpendicular to the loading direction. In general, the fatigue crack initiates from the outer surface of the BMG sample. The area near the crack-initiation site with shear slips is relatively rough, which could result from the shear transformation (Fig. 6a and b). Then, the fatigue crack propagates towards the inside of the specimen. The fatigue-fractured crack-propagation region shows a thumb-nail shape and exhibits a striation-type fracture, which could be due to a repeated resharpening and blunting process during cyclic loading (Fig. 6a and c). The final fast fracture region (Fig. 6a and c) was similar to the tensile facture surface, which forms because the crack grows very fast and the sample breaks immediately. There is a distinct boundary between the crack-propagation and fast fracture regions (Fig. 6c), which reveals that the fatigue and tensile fracture are probably controlled by different fracture mechanisms. The distinct melting mark, droplet, and vein pattern can be observed in the apparent melting region (Fig. 6d). There is no melting mark, droplet, or vein pattern observed in the crack-propagation region (Fig. 6c), which indicates that the melting phenomenon did not occur at the tip of the fatigue crack during the crack-propagation stage. It suggests that the released elastic energy due to crack propagation is too low to melt the metallic glass locally. However, in the melting region the vein-like structure and droplets appeared (Fig. 6d), which suggests that most of the elastic and plastic energies were released, and the temperature increased quickly at the moment of the fracture so that a part of the BMG sample was melted. The same fracture morphology was also found in other Zr-based BMG fatigue-fractured samples. Based on the above discussion, the fatigue and fracture process of BMGs during tension–tension cyclic loading could be summarized: First, shear bands initiate due to shear localization that resulted from the shear transformation confined into small regions near free-volume sites. Then, some shear bands grow into a crack under cyclic stress. This step forms the relative rough crack-initiation region. Secondly, the crack will propagate via a blunting and resharpening process under cyclic loading. This step forms the relative flat crack-propagation region with a striation-type structure. Lastly, the fatigue crack advances very quickly, and the sample fractures shortly. This process forms the fast fracture region with the typical overload feature and the melting region with the melting mark, droplet, and vein pattern.
There is a possible trend between the fatigue limit (or the fatigue ratio) and the volume change. The volume change could indicate the relative quantity of free-volume sites. In general, the fatigue limit (or the fatigue ratio) increased when the volume change became larger. The fatigue crack could initiate from shear bands, which resulted from the shear localization. The crack advanced because of the blunting and resharpening under cyclic loading, which formed the relative flat crack-propagation region with a striation-type structure. At last, the sample fractures quickly, which produces the fast fracture region with the typical overload feature and the melting region with the melting mark, droplet, and vein pattern because most of the elastic and plastic energies were released and the temperature increased quickly at the moment of the fracture. Acknowledgements We would like to acknowledge the financial support of the National Science Foundation: (1) the Division of the Design, Manufacture, and Industrial Innovation Program, under Grant No. DMI9724476, (2) the Combined Research-Curriculum Development (CRCD) Programs, under EEC-9527527 and EEC-0203415, (3) the Integrative Graduate Education and Research Training (IGERT) Program, under DGE-9987548, (4) the International Materials Institutes (IMI) Program, under DMR-0231320, and (5) the Major Research Instrumentation (MRI) Program, under DMR-0421219, to the University of Tennessee, Knoxville, with Dr. D. Durham, Ms. M. Poats, Dr. C.J. Van Hartesveldt, Dr. D. Dutta, Dr. W. Jennings, Dr. L. Goldberg, Dr. C. Huber, and Dr. C.R. Bouldin as Program Directors, respectively. We are also thankful to Mr. Douglas E. Fielden for helping us on machining samples. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20]
4. Conclusions According to the fatigue experiments and the above discussion, the following results could be summarized: In the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy-alloy systems, the additions of Pd have only slight effects on the thermal properties and static mechanical properties. However, it has obvious effects on the fatigue behavior.
603
[21] [22] [23] [24] [25] [26]
Johnson WL. MRS Bull 1999;10:42. Inoue A. Acta Mater 2000;48:279. Gilbert CJ, Lippmann JM, Ritchie RO. Scripta Mater 1998;38:537. Gilbert CJ, Schroeder V, Ritchie RO. Metal Mater Trans A 1999;30A:1739. Menzel BC, Dauskardt RH. Acta Mater 2006;54:935. Wang GY, Liaw PK, Peker A, Freels M, Peter WH, Buchanan RA, et al. Intermetallics 2006;14:1091. Peter WH, Liaw PK, Buchanan RA, Liu CT, Brooks CR, Horton Jr JA, et al. Intermetallics 2002;10:1125. Peter WH, Buchanan RA, Liu CT, Liaw PK. J Non-Cryst Solids 2003;317:187. Wang GY, Liaw PK, Peter WH, Yang B, Yokoyama Y, Benson ML, et al. Intermetallics 2004;12:885. Wang GY, Liaw PK, Peter WH, Yang B, Freels M, Yokoyama Y, et al. Intermetallics 2004;12:1219. Wang GY, Liaw PK, Peker A, Yang B, Benson ML, Yuan W, et al. Intermetallics 2005;13:429. Wang GY, Liaw PK, Yokoyama Y, Peter WH, Yang B, Freels M, et al. J Alloy Compd 2007;434–435:68. Wang GY, Liaw PK, Yokoyama Y, Peker A, Peter WH, Yang B, et al. Intermetallics 2007;15:663. Yokoyama Y, Liaw PK, Nishijima M, Hiraga K, Buchanan RA, Inoue A. Mater Trans JIM 2006;47:1286. Nakai Y, Hosomi S. Mater Trans JIM 2007;48:1770. Flores KM, Johnson WL, Dauskardt RH. Scripta Mater 2003;49:1181. Hess PA, Menzel BC, Dauskardt RH. Scripta Mater 2006;54:355. Freels M, Liaw PK, Wang GY, Zhang QS, Hu ZQ. J Mater Res 2007;22:374. Wang GY, Liaw PK, Yokoyama Y, Freels M, Buchanan RA, Inoue A, et al. J Mater Res 2007;22:493. Qiao DC, Wang GY, Liaw PK, Ponnambalam V, Poon SJ, Shiflet GJ. J Mater Res 2007;22:544. Yokoyama Y, Inoue K, Fukaura K. Mater Trans JIM 2002;43:2316. Yokoyama Y, Akeno Y, Yamasaki T, Liaw PK, Buchanan RA, Inoue A. Mater Trans JIM 2005;46:2755. Flores KM, Suh D, Asoka-Kumar P, Sterne PA, Howell R, Dauskardt RH. J Mater Res 2002;17:1153. Yokoyama Y, Yamasaki T, Liaw PK, Buchanan RA, Inoue A. J Alloy Comd 2007;434–435:160. Inoue A. Mater Trans JIM 1995;36:866. Wang GY, Liaw PK, Smyth A, Denda M, Peker A, Freels M, Buchanan RA, Brooks CR. Rev Adv Mater Sci 2008;18:18.
604
G.Y. Wang et al. / International Journal of Fatigue 32 (2010) 599–604
[27] Nieh TG, Mukai T, Liu CT, Wadsworth J. Scripta Mater 1999;40:1021. [28] Pekarskaya E, Kim CP, Johnson WL. J Mater Res 2001;16:2513. [29] Liu CT, Heatherly L, Easton DS, Carmichael CA, Schneibel JH, Chen CH, et al. Metall Mater Trans A 1998;29A:1811. [30] Wright WJ, Saha R, Nix WD. Mater Trans JIM 2001;42:642.
[31] [32] [33] [34]
Spaepen F. Acta Mater 1977;25:407. Argon AS. Acta Mater 1979;27:47. Leamy HJ, Chen HS, Wang TT. Metall Trans 1972;3:699. Suh D, Dauskardt RH. J Non-Cryst Solids 2003;317:181.
Contents lists available at ScienceDirect
International Journal of Fatigue journal homepage: www.elsevier.com/locate/ijfatigue
The influence of Pd on tension–tension fatigue behavior of Zr-based bulk-metallic glasses G.Y. Wang a,*, P.K. Liaw a, Y. Yokoyama b, M. Freels a, A. Inoue b a b
Department of Materials Science and Engineering, The University of Tennessee, Knoxville, TN 37996, USA Advanced Research Center of Metallic Glasses, Institute for Materials Research, Tohoku University, Sendai 980-8577, Japan
a r t i c l e
i n f o
Article history: Received 1 December 2008 Received in revised form 14 April 2009 Accepted 15 April 2009 Available online 22 April 2009 Keywords: Bulk-metallic glass Amorphous High cycle fatigue S–N curves Fatigue limit
a b s t r a c t Zr-based bulk-metallic glasses (BMGs) are being studied widely because Zr-based BMGs exhibit good glass-forming abilities and excellent properties, such as material strengths. In the current paper, the fatigue behaviors of Zr50Cu40xAl10Pdx [x: 0–7 atomic percent (at.%)] BMGs were investigated. The uniaxial tension–tension fatigue experiments were performed on the button-head rod fatigue specimens. The test environment was air at room temperature. The fatigue limit of Zr50Cu37Al10Pd3 was found to be the highest with a value of 945 MPa among the BMGs studied. A mechanistic understanding of the fatigue behavior of these Zr-based BMGs is suggested. The effect of the Pd content on the fatigue behavior was analyzed. A possible relationship between the fatigue limit (or the fatigue ratio) and the volume change, which probably corresponds to excessive free volume, was developed. Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction As excellent candidates for structural materials, bulk-metallic glasses (BMGs) are being studied extensively [1.2]. Because the fatigue behavior is a very important characteristic for the application of structural materials, the fatigue investigation of BMGs attracts attention [3–20]. However, the understanding of the fatigue behavior of BMGs is still limited. Gilbert et al. [3,4] were the first to report the fatigue results of BMGs. They performed four-point-bend fatigue experiments on Vitreloy 1 [Zr41.2Ti13.8Cu12.5Ni10Be22.5 (in atomic percent, at.%)] beam specimens and claimed that the fatigue limits, based on the stress range, were approximately 8% of the ultimate tensile strength. Menzel and Dauskardt also obtained similar fatigue results [5]. These values are very low compared to conventional crystalline alloys, such as high-strength steels, copper, and aluminum alloys, whose fatigue limits are typically 30– 40% of the ultimate tensile strength. On the other hand, tension–tension fatigue results of taper notched LM001 BMG samples that have the same composition as Vitreloy 1 revealed that the fatigue limits, based on the stress range, were 31% of the ultimate tensile strengths [6]. In addition, Peter et al. [7,8] and Wang et al. [9–13] performed tension–tension fatigue tests on notched Zr-based BMG samples, and Yokoyama et al. [14] did rotating-beam fatigue experiments on Zr-based BMGs. They found fatigue limits as high as 40–50% of the ultimate * Corresponding author. Tel.: +1 865 974 0245; fax: +1 865 974 4115. E-mail address: [email protected] (G.Y. Wang). 0142-1123/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijfatigue.2009.04.009
tensile strength. Recently, Nakai et al. conducted fatigue tests on smooth Zr-based BMGs specimens under fully reversed cyclic loading and reported that the fatigue limit, based on the stress amplitude, is 26% of the ultimate tensile strength [15]. According to these fatigue studies of Zr-based BMGs, it was found that Zr-based BMGs with various compositions exhibited very different fatigue lives. What caused such a large difference among these fatigue results of BMGs? Many factors could be involved, such as the material quality, mean stress, specimen geometry, chemical environment, temperature, cyclic frequency, residual stress, and surface condition. In fact, any processing that changes the static mechanical properties or microstructure will probably also affect the fatigue behavior of materials. However, some of factors mentioned above must play an important role in affecting the fatigue behavior of BMGs. The formation of shear bands is still unclear during the cyclic deformation of BMGs. How the fatigue crack initiates and propagates in metallic glasses needs to be solved. Thus, it is essential to perform the fundamental research on the fatigue behavior of BMGs. Zr50Cu40Al10 shows good mechanical behavior, and the fatigue behavior can be improved. In order to strengthen the fatigue behavior of Zr50Cu40Al10 BMGs, in general, some small additive elements like Ni and Pd were added into this alloy system [14]. In the current paper, uniaxial tension–tension fatigue experiments on the Zr50Cu40xAl10Pdx [x: 0–7 atomic percent (at.%)] were performed. The applied stress versus fatigue cycle (S–N) curves of these BMGs is presented. The fracture surfaces were observed and analyzed using scanning-electron microscopy (SEM). In addition, the factors
600
G.Y. Wang et al. / International Journal of Fatigue 32 (2010) 599–604
Fig. 1. Specimen geometry for tension–tension fatigue experiments (unit: mm).
which influence the fatigue limit were discussed. Thus, the effect of the Pd content on the fatigue behavior of BMGs was clarified. 2. Experimental procedures Zr50Cu40xAl10Pdx (x: 0–7 at.%) BMGs were the ternary and quaternary alloys used in the present experiments. The ladle-hearth type arc-melt tilt-casting technique was employed to manufacture these BMGs. The master-alloy ingots were prepared by arc-melting mixtures of pure Zr, Cu, Al, and Pd metals in an argon atmosphere. A special Zr-crystal rod [<0.05 at.% oxygen] was employed in order to maintain the low-oxygen concentration of the alloys [21,22]. The tilt-casting technique has an advantage to control the molten alloy flow to restrict the formation of cold shuts, which probably induces early fatigue–crack initiation and propagation behavior [21,22]. Rod BMG ingots were produced with a length of 60 mm and a diameter of 8 mm. One taper notched specimen (Fig. 1) was machined from each ingot. The fatigue samples were, then, polished to minimize surface effects. The thermal properties of these BMG samples were measured in a Perkin–Elmer Diamond differential scanning calorimeter (DSC) at a heating rate of 20 K/min. The weight of samples used for DSC was in the range of 30–50 mg. The previous research result revealed a decrease in the free volume by annealing a BMG [23]. The free-volume changes could result in the variation of the BMG volume. Therefore, the volume change due to the structural relaxation at the glass-transition temperature (Tg) could be a convenient method to evaluate the freevolume difference in the glass structure. The volume change is defined as the volume change ratio of BMGs from the as-cast state to the annealed state at Tg. The following equation is used to determine the value of the volume change [24]:
DV ¼
1 q1 0 qTg 1 q0
where DV is the volume change, q0 is the density of the as-cast alloy, and qTg is the density of the alloy after annealing at Tg for 90 min. A computer-controlled Material Test System (MTS) servohydraulic testing machine was employed for fatigue studies. The machine was aligned prior to use, and as required. Samples were tested at various stress ranges with a R ratio (R = rmin/rmax, where rmin and rmax are the applied minimum and maximum stresses, respectively) of 0.1 under a load-control mode, using a sinusoidal waveform at a frequency of 10 Hz. Upon failures or 107 cycles, samples were removed and stored for later examinations by SEM. The fracture surfaces of selected specimens were examined, using a Leo 1526 SEM machine with the energy-dispersive spectroscopy (EDS) to provide fatigue and fracture mechanisms. 3. Results and discussion Fig. 2 shows the DSC thermograms obtained from the as-cast fatigue samples during continuous heating at a heating rate of 20 K/ min. An endothermic reaction, corresponding to the transition from a glassy state to a supercooled liquid state, and the following exothermic reaction, corresponding to crystallization are clearly observed. These parameters are labeled as the glass-transition temperature, Tg, and crystallization temperature, Tx, respectively. The characteristic temperatures, as well as heats of crystallization (DHx), are given in Table 1. In addition, the supercooled liquid region DTx = Tx Tg) was also calculated, as shown in Table 1. DTx may indicate that the larger the accessible supercooled liquid region is, the greater the glass-forming ability is [25]. The value of Tg increases with increasing the Pd content in these BMGs. However, the value of Tx and DHx exhibit slight decreases when the Pd content increases from 0% to 7% in these glassy alloys. The increase of Pd resulted in a slight decrease of the supercooled liquid region DTx), which may indicate that Pd has the negative effect on
Table 1 Glass-transition temperature (Tg), crystallization temperature (Tx), and heat of crystallization (DHx), as well as a supercooled liquid region (DTx = Tx Tg) for the Zr50Cu40xAl10Pdx (x: 0–7 at.%) alloys, as obtained from DSC at a heating rate of 20 K/ min.
Fig. 2. DSC thermograms of the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy alloys. The heating rate was 20 K/min. Locations of the characteristic temperatures (Tg and Tx) are indicated by arrows.
Materials
Tg (°C)
Tx (°C)
DHx (J/g)
DTx (°C)
Zr50Cu40Al10 Zr50Cu39Al10Pd1 Zr50Cu38Al10Pd2 Zr50Cu37Al10Pd3 Zr50Cu35Al10Pd5 Zr50Cu33Al10Pd7
417 428 431 431 434 435
507 507 507 507 503 504
43 43 39 40 34 33
90 79 76 76 69 69
G.Y. Wang et al. / International Journal of Fatigue 32 (2010) 599–604
Fig. 3. Stress-range/fatigue-life curves of the Zr50Cu40xAl10Pdx (x: 0–7 at.%) specimens tested with R = 0.1 at a frequency of 10 Hz in air. Note that the fatigue data of Zr50Cu40Al10 and Zr50Cu37Al10Pd3 alloys was obtained from the sharpnotched samples [9,13].
the glass-forming ability among Zr50Cu40xAl10Pdx (x: 0–7 at.%) BMGs. The S–N curves of the tension–tension fatigue results for the Zr50Cu40xAl10Pdx (x: 0–7 at.%) samples tested in air with R = 0.1 and a frequency of 10 Hz were presented in Fig. 3. Note that the fatigue data of Zr50Cu40Al10 and Zr50Cu37Al10Pd3 alloys was obtained from the sharp-notched samples [9,13] and the stress-concentration factor (Kt) is considered to be 1.49 at the notched section for these sharp-notched specimens based on the finite-element analysis [26]. New taper notched samples (Kt 1) were employed to avoid the stress-concentration-factor effect. The value of the stress range for these BMGs in the fatigue-life range of 103–105 cycles decreased significantly with increasing the cycles to failure. For these BMGs, the lifetime above the fatigue limit demonstrated no large differences. Nevertheless, the fatigue limits (rL), based on the applied stress range, for the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy alloys subjected to tension–tension loading exhibited clear differences, increasing from 537 to 945 MPa with the variation of the Pd content. The fatigue ratios (rL divided by the tensile strength) for these BMGs also improved obviously from 0.28 to 0.50 by changing the Pd content. The tensile strength, fatigue limit, and fatigue ratio, as well as the hardness, for the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy alloys were given in Table 2. It was obviously found that the tensile strength and hardness increased slightly with increasing the Pd content among the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy-alloy system [14]. However, the fatigue limits of the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy alloys showed a distinct trend, as seen in Fig. 4. The fatigue limits of the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy alloys decrease slightly when Pd content increases from 1% to 2% and, then, increase with increasing the Pd Table 2 Hardness, tensile strength, fatigue limit, and fatigue ratio for the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy alloys. Materials
Hardness (HV) [14]
Tensile strength (MPa) [14]
Fatigue limit (MPa)
Fatigue ratioa
Zr50Cu40Al10 Zr50Cu39Al10Pd1 Zr50Cu38Al10Pd2 Zr50Cu37Al10Pd3 Zr50Cu35Al10Pd5 Zr50Cu33Al10Pd7
506 505 504 503 512 528
1851 1909 1911 1899 1929 1952
723 597 537 945 608 776
0.39 0.31 0.28 0.50 0.32 0.40
a
Fatigue ratio = Fatigue limit/tensile strength.
601
Fig. 4. The relationship between the fatigue limit (fatigue ratio) and the Pd content among the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy alloys. Hollow points are from sharp-notched samples and solid points are from taper notched samples.
content. The fatigue ratios of these BMGs have the same trend as the fatigue limits. Why does the addition of Pd have slight effects on the thermal properties and static mechanical properties of Zr50Cu40xAl10Pdx (x: 0–7 at.%) in Tables 1 and 2, while it has obvious effects on the fatigue behavior of Zr50Cu40xAl10Pdx (x: 0–7 at.%) in Table 2? There are some important factors that influence the fatigue behavior. The deformation characteristic of metallic glasses can generally be classified as either inhomogeneous or homogeneous. The homogeneous deformation often occurs at high temperatures (e.g., T > 0.70 Tg), and the metallic glasses can demonstrate significant plasticity, especially in the supercooled liquid region [27]. The inhomogeneous deformation usually occurs when a metallic glass is deformed at low temperatures (e.g., room temperature) and is characterized by the formation of localized shear bands, followed by the rapid propagation of these bands, and sudden fracture. Typically, these shear bands are 10–20 nm in width [28]. A large amount of plastic strain occurs at these localized shear regions. However, the entire plastic deformation of the specimen is generally very low (2–3%) [29]. Two primary hypotheses are presented in the literature to explain the plastic flow and localized shear bands in metallic glasses [30]. The first hypothesis suggests that the viscosity in the shear bands decreases due to the formation of free volumes [31,32]. The second hypothesis suggests that the viscosity in the shear bands decreases due to the generation of the local adiabatic heating [33]. The generally accepted view on the inhomogeneous deformation in metallic glasses is the free-volume theory. Argon thought that the shear transformation at high stresses in metallic glasses narrowed down to a region between two short rows of 4–6 atoms around a free-volume site and resulted in the formation of strong shear localization, giving rise to the shear bands [30]. During cyclic experiments, it was also found that cracks initiated from shear bands [5,9,10,15]. This fact implied that the shear transformation confined into small regions near free-volume sites could produce the shear localization that lead to the formation of the shear bands when the metallic glasses were subjected to a repeated stress. It is suggested that the fatigue behavior could be related to the free volume in metallic glasses. Note that fatigue cracks grow unimpeded once they initiated due to the absence of microstructural barriers in metallic glasses [5]. Since the fatigue limit means that the fatigue sample does not fracture when the applied stress is below this value, it could also suggest that the fatigue sample does not initiate a crack in the case of metallic glasses. Therefore, it is possible that there is a relationship between the fatigue limit and the free volume in metallic glasses.
602
G.Y. Wang et al. / International Journal of Fatigue 32 (2010) 599–604
Fig. 5. The relationship between the fatigue limit (fatigue ratio) and volume change among the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy alloys. Hollow points are from sharp-notched samples and solid points are from taper notched samples.
Metallic Glasses are solidified at high cooling rates below the glass-transition temperature. Thus, the free volume in a glassy alloy may change with the cooling rate, and the quantity of free volumes may also be changed by the alloy composition [24]. However, it is difficult to determine the absolute quantity of the free volume in a metallic glass. Therefore, the volume change is employed to evaluate the free-volume variation in the metallic glass [24]. Fig. 5 exhibited the relationship of the fatigue limit and the fatigue ratio with the volume change in the Zr50Cu40xAl10Pdx (x: 0–7 at.%)
glassy alloys. It seems that there is a possible trend between the fatigue limit (or the fatigue ratio) and the volume change, in which the fatigue limit increases with increasing the volume change. The larger value of the volume change could suggest more free-volume sites in metallic glasses and the existence of more free-volume sites could allow atoms to move more easily in metallic glasses. Thus, more shear bands could form and propagate easily. As a result, these shear bands disperse the localized stress. This trend could reduce the probability of the stress concentration during cyclic loading. Thus, it is difficult to initiate a fatigue crack, and the metallic glass could show a higher fatigue limit with increasing the volume change (Fig. 5). Yokoyama et al. [14] thought that the volume change indicated thermally unstable volumes existed below Tg in the amorphous structure, and it was generally assumed that the amorphous structure was composed of clusters and free volumes. Thermally unstable free volumes were probably effective to relax the localized stress because the atoms surrounding the void moved more easily than others, and the stress relaxation was beneficial to fracture and fatigue behavior [14]. In addition, Suh and Dauskardt [34] found that the plane-strain fracture toughness of a Zr–Ti–Ni–Cu–Be bulk-metallic glass was significantly decreased after annealing at 300 °C. They thought the anneal-out of free volumes resulted in the retarded atomic arrangement process for viscous flows, which led to the loss of the stress relief by viscous flows at the crack tip. As a result, the annealing embrittlement was observed [34]. In fact, it could be possible that the decrease in free volumes made it difficult for shear bands to produce and propagate so that the significant stress concentration occurred at the crack tip, which could generally result in lower fatigue limits with decreasing volume change (Fig. 5).
Fig. 6. (a) Overall fatigue fractography of the Zr50Cu38Al10Pd2 specimen tested at rmax = 663 MPa; (b) crack-initiation site; (c) boundary between the crack-growth region and final fast fracture regions and (d) local melting phenomena.
G.Y. Wang et al. / International Journal of Fatigue 32 (2010) 599–604
The volume change was assumed to be the degree of volume expansion from the fully relaxed state and could only indicate the relative quantity of free-volume sites. This situation could be one of the reasons why the linear relationship between the fatigue limit (the fatigue ratio) and the volume change is not so strong in metallic glasses (Fig. 5). Other factors, such as the surface condition, sample quality, and residual stress, could also affect the fatigue limit. Therefore, complex factors need to be considered during studying the fatigue behavior of BMGs. The fractured surface of a Zr50Cu38Al10Pd2 BMG sample with SEM observations after tension–tension fatigue experiments is shown in Fig. 6. It is found that the whole fatigue-fracture surface clearly includes the fatigue–crack initiation, crack propagation, final fast fracture, and apparent melting areas. The fracture surface is basically perpendicular to the loading direction. In general, the fatigue crack initiates from the outer surface of the BMG sample. The area near the crack-initiation site with shear slips is relatively rough, which could result from the shear transformation (Fig. 6a and b). Then, the fatigue crack propagates towards the inside of the specimen. The fatigue-fractured crack-propagation region shows a thumb-nail shape and exhibits a striation-type fracture, which could be due to a repeated resharpening and blunting process during cyclic loading (Fig. 6a and c). The final fast fracture region (Fig. 6a and c) was similar to the tensile facture surface, which forms because the crack grows very fast and the sample breaks immediately. There is a distinct boundary between the crack-propagation and fast fracture regions (Fig. 6c), which reveals that the fatigue and tensile fracture are probably controlled by different fracture mechanisms. The distinct melting mark, droplet, and vein pattern can be observed in the apparent melting region (Fig. 6d). There is no melting mark, droplet, or vein pattern observed in the crack-propagation region (Fig. 6c), which indicates that the melting phenomenon did not occur at the tip of the fatigue crack during the crack-propagation stage. It suggests that the released elastic energy due to crack propagation is too low to melt the metallic glass locally. However, in the melting region the vein-like structure and droplets appeared (Fig. 6d), which suggests that most of the elastic and plastic energies were released, and the temperature increased quickly at the moment of the fracture so that a part of the BMG sample was melted. The same fracture morphology was also found in other Zr-based BMG fatigue-fractured samples. Based on the above discussion, the fatigue and fracture process of BMGs during tension–tension cyclic loading could be summarized: First, shear bands initiate due to shear localization that resulted from the shear transformation confined into small regions near free-volume sites. Then, some shear bands grow into a crack under cyclic stress. This step forms the relative rough crack-initiation region. Secondly, the crack will propagate via a blunting and resharpening process under cyclic loading. This step forms the relative flat crack-propagation region with a striation-type structure. Lastly, the fatigue crack advances very quickly, and the sample fractures shortly. This process forms the fast fracture region with the typical overload feature and the melting region with the melting mark, droplet, and vein pattern.
There is a possible trend between the fatigue limit (or the fatigue ratio) and the volume change. The volume change could indicate the relative quantity of free-volume sites. In general, the fatigue limit (or the fatigue ratio) increased when the volume change became larger. The fatigue crack could initiate from shear bands, which resulted from the shear localization. The crack advanced because of the blunting and resharpening under cyclic loading, which formed the relative flat crack-propagation region with a striation-type structure. At last, the sample fractures quickly, which produces the fast fracture region with the typical overload feature and the melting region with the melting mark, droplet, and vein pattern because most of the elastic and plastic energies were released and the temperature increased quickly at the moment of the fracture. Acknowledgements We would like to acknowledge the financial support of the National Science Foundation: (1) the Division of the Design, Manufacture, and Industrial Innovation Program, under Grant No. DMI9724476, (2) the Combined Research-Curriculum Development (CRCD) Programs, under EEC-9527527 and EEC-0203415, (3) the Integrative Graduate Education and Research Training (IGERT) Program, under DGE-9987548, (4) the International Materials Institutes (IMI) Program, under DMR-0231320, and (5) the Major Research Instrumentation (MRI) Program, under DMR-0421219, to the University of Tennessee, Knoxville, with Dr. D. Durham, Ms. M. Poats, Dr. C.J. Van Hartesveldt, Dr. D. Dutta, Dr. W. Jennings, Dr. L. Goldberg, Dr. C. Huber, and Dr. C.R. Bouldin as Program Directors, respectively. We are also thankful to Mr. Douglas E. Fielden for helping us on machining samples. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20]
4. Conclusions According to the fatigue experiments and the above discussion, the following results could be summarized: In the Zr50Cu40xAl10Pdx (x: 0–7 at.%) glassy-alloy systems, the additions of Pd have only slight effects on the thermal properties and static mechanical properties. However, it has obvious effects on the fatigue behavior.
603
[21] [22] [23] [24] [25] [26]
Johnson WL. MRS Bull 1999;10:42. Inoue A. Acta Mater 2000;48:279. Gilbert CJ, Lippmann JM, Ritchie RO. Scripta Mater 1998;38:537. Gilbert CJ, Schroeder V, Ritchie RO. Metal Mater Trans A 1999;30A:1739. Menzel BC, Dauskardt RH. Acta Mater 2006;54:935. Wang GY, Liaw PK, Peker A, Freels M, Peter WH, Buchanan RA, et al. Intermetallics 2006;14:1091. Peter WH, Liaw PK, Buchanan RA, Liu CT, Brooks CR, Horton Jr JA, et al. Intermetallics 2002;10:1125. Peter WH, Buchanan RA, Liu CT, Liaw PK. J Non-Cryst Solids 2003;317:187. Wang GY, Liaw PK, Peter WH, Yang B, Yokoyama Y, Benson ML, et al. Intermetallics 2004;12:885. Wang GY, Liaw PK, Peter WH, Yang B, Freels M, Yokoyama Y, et al. Intermetallics 2004;12:1219. Wang GY, Liaw PK, Peker A, Yang B, Benson ML, Yuan W, et al. Intermetallics 2005;13:429. Wang GY, Liaw PK, Yokoyama Y, Peter WH, Yang B, Freels M, et al. J Alloy Compd 2007;434–435:68. Wang GY, Liaw PK, Yokoyama Y, Peker A, Peter WH, Yang B, et al. Intermetallics 2007;15:663. Yokoyama Y, Liaw PK, Nishijima M, Hiraga K, Buchanan RA, Inoue A. Mater Trans JIM 2006;47:1286. Nakai Y, Hosomi S. Mater Trans JIM 2007;48:1770. Flores KM, Johnson WL, Dauskardt RH. Scripta Mater 2003;49:1181. Hess PA, Menzel BC, Dauskardt RH. Scripta Mater 2006;54:355. Freels M, Liaw PK, Wang GY, Zhang QS, Hu ZQ. J Mater Res 2007;22:374. Wang GY, Liaw PK, Yokoyama Y, Freels M, Buchanan RA, Inoue A, et al. J Mater Res 2007;22:493. Qiao DC, Wang GY, Liaw PK, Ponnambalam V, Poon SJ, Shiflet GJ. J Mater Res 2007;22:544. Yokoyama Y, Inoue K, Fukaura K. Mater Trans JIM 2002;43:2316. Yokoyama Y, Akeno Y, Yamasaki T, Liaw PK, Buchanan RA, Inoue A. Mater Trans JIM 2005;46:2755. Flores KM, Suh D, Asoka-Kumar P, Sterne PA, Howell R, Dauskardt RH. J Mater Res 2002;17:1153. Yokoyama Y, Yamasaki T, Liaw PK, Buchanan RA, Inoue A. J Alloy Comd 2007;434–435:160. Inoue A. Mater Trans JIM 1995;36:866. Wang GY, Liaw PK, Smyth A, Denda M, Peker A, Freels M, Buchanan RA, Brooks CR. Rev Adv Mater Sci 2008;18:18.
604
G.Y. Wang et al. / International Journal of Fatigue 32 (2010) 599–604
[27] Nieh TG, Mukai T, Liu CT, Wadsworth J. Scripta Mater 1999;40:1021. [28] Pekarskaya E, Kim CP, Johnson WL. J Mater Res 2001;16:2513. [29] Liu CT, Heatherly L, Easton DS, Carmichael CA, Schneibel JH, Chen CH, et al. Metall Mater Trans A 1998;29A:1811. [30] Wright WJ, Saha R, Nix WD. Mater Trans JIM 2001;42:642.
[31] [32] [33] [34]
Spaepen F. Acta Mater 1977;25:407. Argon AS. Acta Mater 1979;27:47. Leamy HJ, Chen HS, Wang TT. Metall Trans 1972;3:699. Suh D, Dauskardt RH. J Non-Cryst Solids 2003;317:181.