- Email: [email protected]
Al–Al3Ti nanocomposites produced in situ by friction stir processing
Acta Materialia 54 (2006) 5241–5249 www.actamat-journals.com
Al–Al3Ti nanocomposites produced in situ by friction stir processing C.J. Hsu, C.Y. Chang, P.W. Kao *, N.J. Ho, C.P. Chang Institute of Materials Science and Engineering, National Sun Yat-Sen University, No. 70 Lian-Hi Road, Kaohsiung 804, Taiwan Received 16 February 2006; received in revised form 25 April 2006; accepted 29 June 2006 Available online 14 September 2006
Abstract Aluminum reinforced with a large amount (up to 50 vol.%) of nanometer-sized Al3Ti particles can be fabricated from Al–Ti elemental powder mixtures via friction stir processing (FSP). This technique combines the hot working nature of FSP and the exothermic reaction between aluminum and titanium. For the production of intermetallic reinforced in situ composites, FSP can provide (a) severe deformation to promote mixing and refining of the constituent phases in the material, (b) elevated temperature to facilitate the formation of intermetallic phase in situ, and (c) hot consolidation to form a fully dense solid. Due to the fine dispersion of Al3Ti particles, the aluminum matrix has a submicrometer-grained structure. As a result of the high Al3Ti content and very fine microstructure, the composites possess enhanced Young’s modulus and strength. 2006 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Metal matrix composites; Friction stir processing; Intermetallic compound
1. Introduction Friction stir welding (FSW) was invented at The Welding Institute (TWI) in the UK in 1991 as a solid-state joining technique [1]. In FSW, a non-consumable rotating tool with a specially designed pin and shoulder is plunged into the interface between two plates to be joined and traversed along the line of the joint. Localized heating is produced by the friction between the rotating tool and the workpiece to raise the local temperature of the material to the range where it can be plastically deformed easily. As the rotating tool traverses along the joint line, metal is essentially extruded around the tool before being forged by the large down pressure [2–5]. The material flow during FSW is quite complex depending on the tool geometry, process parameters, and material to be welded. Reynolds and co-workers [2,3] investigated the material flow behavior in FSW using a marker insert technique. They suggested that there is a secondary, circular motion around the longitudinal axis of the weld, which is a combined result of the stirring
*
Corresponding author. Tel.: +886 7 5252000; fax: +886 7 5254099. E-mail address: [email protected] (P.W. Kao).
action and the extrusion around the pin. The intense plastic deformation around the tool and the friction between tool and workpiece both contribute to the temperature increase in the stirred zone during FSW. However, temperature measurements within the stirred zone are very difficult due to the intense plastic deformation produced by the rotation and translation of the tool. The maximum temperature in the stirred zone during FSW of various aluminum alloys is found to be between 0.6Tm and 0.9Tm, which depends on the ratio of rotation rate to translation speed [5]. Generally, an increase in the ratio of rotation rate to translation speed will increase the peak temperature in FSW. A comprehensive literature review of friction stir welding has been given by Mishra and Ma [5]. Based on the principles of FSW, Mishra et al. [6] developed friction stir processing (FSP) for microstructural modification of materials. FSP has been shown as an effective technique in many applications: (a) to form ultrafinegrained structure in Al alloys [7–10] and Mg alloys [11], (b) to produce a fine-grained microstructure, which exhibits superplasticity [6,12–14], (c) to homogenize the microstructure of nanocomposite aluminum alloys [15], (d) to refine the microstructure of cast aluminum alloys [16,17], and (e) to fabricate a surface composite of Al–SiC on an
1359-6454/$30.00 2006 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.actamat.2006.06.054
5242
C.J. Hsu et al. / Acta Materialia 54 (2006) 5241–5249
aluminum substrate [18]. Recently, FSP has been successfully applied to produce ultrafine-grained Al–Al2Cu in situ composite from an Al–Cu elemental powder mixture [19]. Metal matrix composites (MMCs) are an important class of material for structural and electrical applications [20]. Particulate-reinforced MMCs are of particular interest because of their easy fabrication, low cost, and isotropic properties. In conventionally processed power metallurgy composites, the reinforcing particles are formed prior to their addition to the matrix metal. In this case, the scale of reinforcing phase is limited by the starting powder size, which is typically of the order of several to tens of micrometers and rarely below 1 lm. Other drawbacks for conventionally processed MMCs that have to be overcome are poor interfacial bonding and poor wettability between the reinforcement and the matrix due to surface contamination of the reinforcements. It is widely recognized that the mechanical properties of MMCs are controlled by the size and volume fraction of the reinforcements as well as the nature of the matrix–reinforcement interface [20]. Superior mechanical properties can be achieved when fine and stable reinforcements with good interfacial bonding are dispersed uniformly in the matrix. A possible alternative is to synthesize the reinforcement in situ in the metal matrix [21]. The advantages of in situ MMCs are that they have more homogeneous microstructures and are thermodynamically more stable. Moreover, they also have strong interfacial bonding between the reinforcements and the matrices. In situ aluminum matrix composites may be fabricated by various techniques such as conventional ingot metallurgy, mechanical alloying (MA), rapid solidification processing (RS), combustion synthesis, etc. The microstructure produced by conventional ingot metallurgy is rather coarse. Materials produced by MA, RS, or combustion synthesis must be densified by a hot consolidation process such as hot isostatic pressing, hot pressing or hot extrusion in order to obtain the final product. Aluminum alloys reinforced with aluminide particles (NiAl3, FeAl3, TiAl3, etc.) possess high specific strength, high specific modulus, and excellent properties both at ambient and elevated temperature [22–31]. As compared to most other aluminum-rich intermetallics, Al3Ti is very attractive because it has a higher melting point (1623 K) and relatively low density (3.4 g/cm3). Further, Ti has low diffusivity and solubility in aluminum; hence Al3Ti can be expected to exhibit a low coarsening rate at elevated temperature [32]. In addition, the Young’s modulus of Al3Ti phase has been determined to be 216 GPa [33]. Therefore, the presence of Al3Ti phase is very effective in increasing the stiffness of aluminum alloys. Al–Al3Ti composites have been fabricated by RS [26,27] and MA [28–31] to achieve high strength. The major strengthening mechanisms, which contribute to the high strength of Al–Al3Ti alloys, have been suggested to include Orowan strengthening, grain size strengthening, and load-shearing effects of Al3Ti particles [26,34,35].
The objective of the work reported in this study was to produce fully dense Al–Al3Ti composites with a large amount of nanometer-sized Al3Ti particles as reinforcement. The basic idea for fabricating the Al–Al3Ti composites is to combine the hot working nature of FSP and the exothermic reaction between aluminum and titanium. The FSP is utilized to provide the following functions: (a) severe plastic deformation to promote mixing and refining of the constituent phases in the material, (b) elevated temperature to facilitate the formation of intermetallic phase, and (c) hot consolidation to form a fully dense solid. In this paper, we present the microstructure and mechanical properties of nanometer-sized Al3Ti particle-reinforced aluminum composites produced via FSP. 2. Experimental The starting materials used were aluminum powder (99.7% purity, 325 mesh) and titanium powder (99.1% purity, 325 mesh). Titanium contents of 5, 10, or 15 at.% were pre-mixed with aluminum powder (denoted Al–5Ti, Al–10Ti, and Al–15Ti). The pre-mixed Al–Ti alloy powders were cold compacted to 12 · 20 · 88 mm billets in a steel die set using a pressure of 225 MPa. To improve the billet strength for easier handling in FSP, the green compact was sintered for 20 min in air at 823 K. The tool pin used in FSP is a standard M1.2*6 (metrication, with pitch height and diameter of 1.2 and 6 mm, respectively). Counterclockwise tool rotation with a speed of 700 or 1400 rpm was used, and the rotating tool was traversed at a speed of 45 mm/min along the long axis of the billet. Multiple FSP passes were applied to the billet to enhance the Al–Ti reaction. For multiple FSP, the pin tool was moved along the same line, and the FSP pass was applied after the workpiece had been cooled from the previous FSP pass. A schematic of the FSP setup is shown in Fig. 1. X-ray diffraction (XRD; Cu Ka radiation) was utilized to identify the phases present in the specimens. Scanning electron microscopy (SEM; JSM-6400 and JSM-6330) was used to study the distribution of second-phase particles
Fig. 1. Schematic of the friction stir process.
C.J. Hsu et al. / Acta Materialia 54 (2006) 5241–5249
5243
in the stirred zone (SZ) and the base metal (BM). Thin foils cut from the SZ were studied using analytical electron microscopy (AEM; JEOL 3010, operated at 200 kV). The Vickers microhardness of the SZ was measured with a 300 g load for 15 s. Mechanical properties of specimens machined from the SZ were evaluated. The tests were carried out using an Instron 5582 universal testing machine with an initial strain rate of 1 · 103 s1. The dimensions of the gauge section of tensile specimens were 3 mm in diameter and 14 mm in length. Rectangular specimens with dimensions of 4 · 4 · 6 mm3 were used for compression tests. The stress direction was aligned along the traverse direction of FSP. 3. Results 3.1. Al–Ti reaction during sintering The compacted billet was sintered to provide enough strength to prevent cracking during FSP. After sintering at 823 K for 20 min, only diffraction peaks of Al and Ti appeared as shown in Fig. 2. The formation of Al3Ti phase was observed in an Al–15Ti specimen sintered at 883 K for 20 min as shown in Fig. 2. The 883 K sintered specimen was also examined using SEM/backscattered electron imaging (BEI). The Al3Ti phase (light gray) was found to enclose the Ti particles (white) as shown in Fig. 3(a). In addition, Al3Ti nodules, consisting of faceted particles of a few micrometers in size, were also observed (Fig. 3(b)). In order to avoid the formation of coarse Al3Ti particles, the billets for FSP were sintered at 823 K for 20 min.
Fig. 3. (a) SEM/BEI of Al–15Ti sintered at 883 K. (b) A magnified view of the selected area in (a). The white phase is titanium, the light gray phase is Al3Ti, and the dark matrix is Al.
3.2. Al–Ti reaction during FSP
Fig. 2. XRD patterns for 823 and 883 K sintered Al–15Ti specimens.
XRD was used to identify the phases present in the SZ of specimens during FSP. The diffraction patterns showed that Ti reacted with Al to form Al3Ti, but some unreacted Ti remained after four FSP passes. A typical example of the XRD patterns of the FSP specimen is shown in Fig. 4. The time that material is affected by the rotating pin in FSP can be considered as the processing time. Based upon the pin diameter and the tool traversing speed, the processing time is estimated to be about 8 s. Therefore, the formation of Al3Ti phase is very rapid in FSP. The microstructure of FSP specimens was observed using SEM/BEI as shown in Fig. 5. The white particles of micrometer size are pure Ti which has been verified using energy dispersive spectroscopy (EDS), and the very fine gray particles (<100 nm) are Al3Ti particles which are uniformly dispersed in the Al matrix. After FSP, the average size of Ti particles is refined from 40 lm to about 1–5 lm. Some of the Ti particles after FSP show an irregular morphology (Fig. 5(b)). The appearance of irregular morphology at the interface between Ti and Al + Al3Ti suggests that liquid phase might
5244
C.J. Hsu et al. / Acta Materialia 54 (2006) 5241–5249
Fig. 4. XRD pattern of Al–15Ti alloy produced by four FSP passes with tool rotation at 700 rpm.
strain imposed in FSP. Thus, the reaction can proceed rapidly at the Al–Ti interface. From image analyses of the SEM/BEI images, the amount of residual Ti was quantitatively determined. Since Al3Ti is the only reaction product between Al and Ti observed in the present case, the volume fraction of Al3Ti particles in each specimen can be calculated from the reacted Ti. The quantitative results of unreacted Ti and the volume fraction of Al3Ti particles in each specimen are listed in Table 1. In general, the Al3Ti content for each composition increases with increasing FSP passes. The values of microhardness of the SZ, which are also listed in Table 1, show a general increasing trend with increasing Al3Ti content. For Al–15Ti after 4 FSP passes, the volume fraction of Al3Ti is close to 0.5, which results in a hardness value of 200 Hv. From microscopic observations, the SZ was found to be free of porosity. In addition, the measured density of material cut from the SZ agrees well with the theoretical density. These evidences indicate that the FSP sample is fully dense. 3.3. Microstructure of Al–Al3Ti nanocomposites observed using transmission electron microscopy (TEM) The cross-section of the SZ in specimens produced by FSP was also examined using TEM. Typical microstructures are shown in Fig. 6, which shows a large number of fine Al3Ti particles uniformly dispersed in an ultrafine-grained Al matrix. The fine Al3Ti particles were found within the grain interior as well as along the grain boundaries of the Al matrix (Fig. 7). The Al3Ti particles were identified by the use of electron diffraction as shown in Fig. 8. In addition, the Al3Ti particles were also confirmed by the EDS analysis with TEM, which showed that the Al:Ti atomic ratio is near 75:25. In general, the Al3Ti particles have equiaxed shape and often with facets (Fig. 8).
Table 1 Quantitative results of the volume fraction of residual Ti (VTi), the calculated volume fraction of Al3Ti (V), and the microhardness of Al–Al3Ti composites (Hv) produced by various FSP parameters
Fig. 5. (a) SEM/BEI of Al–5Ti alloy produced by four FSP passes with tool rotation at 700 rpm. (b) SEM/BEI of Al–10Ti alloy produced by four FSP passes with tool rotation at 700 rpm, which shows a Ti particle (white phase) with irregular morphology.
be present at the Al–Ti reaction front. The liquid phase is considered as Al melt, which results from the exothermic Al–Ti reaction. The liquid Al can react rapidly with Ti to form Al3Ti phase. The Al3Ti particles are effectively removed and dispersed in the Al matrix by the large plastic
Sample
Rotation speed (rpm)/passes
Residual Ti (VTi)
Calculated Al3Ti (V)
Al–5Ti
700/2 700/4 1400/2 1400/4
3.75 3.32 3.79 3.51
0.06 0.07 0.06 0.07
86 ± 6 97 ± 3 93 ± 7 98 ± 5
Al–10Ti
700/2 700/4 1400/2 1400/4
5.28 3.97 6.33 3.02
0.20 0.24 0.16 0.28
107 ± 9 148 ± 7 122 ± 2 127 ± 9
Al–15Ti
700/2 700/4 1400/2 1400/4
7.91 3.24 7.16 2.69
0.30 0.47 0.33 0.49
144 ± 9 226 ± 9 148 ± 7 192 ± 12
Microhardness (Hv)
C.J. Hsu et al. / Acta Materialia 54 (2006) 5241–5249
Fig. 6. TEM bright-field image showing the nanometer-size Al3Ti particles uniformly distributed in the Al matrix of an Al–10Ti/700/4 specimen.
5245
Fig. 8. TEM central dark-field image showing a nearly equiaxed Al3Ti particle in an Al–15Ti/700/4 specimen with inset showing the selected area diffraction pattern in [1 0 0] zone axis.
Table 2 Average sizes of Al grains and Al3Ti particles in Al–Al3Ti composites produced by various FSP parameters
Fig. 7. TEM central dark-field image showing the distribution of Al3Ti particles within the grain interior as well as along the grain boundaries of the Al matrix in an Al–5Ti/700/4 specimen.
Quantitative results of the size of Al3Ti particles and Al grains for various alloys and processing conditions are listed in Table 2. In general, the average size of Al grains has been refined to the submicrometer level, except Al– 5Ti processed at 1400 rpm for 4 passes, which has a grain size slightly larger than 1 lm. From Table 2, we also have the following observations: (1) For the same Ti content, the Al3Ti particle size increases with increasing tool rotation speed and increasing FSP passes. This may be attributed to higher temperature associated with the higher tool rotation speed, and longer time exposed at elevated temperature with increasing FSP passes.
Sample
Rotation speed (rpm)/ passes
Al–5Ti
700/2 700/4 1400/2 1400/4
Al–10Ti Al–15Ti
Al3Ti particle size (nm)
Al grain size (lm)
Zener limiting grain size (lm)
47 ± 22 60 ± 31 51 ± 24 87 ± 33
0.90 ± 0.27 0.86 ± 0.32 0.72 ± 0.21 1.53 ± 0.39
0.52 0.57 0.57 0.83
700/4 1400/4
53 ± 13 83 ± 20
0.30 ± 0.12 0.74 ± 0.24
0.15 0.26
700/2 700/4 1400/2 1400/4
69 ± 21 83 ± 22 78 ± 21 147 ± 36
0.33 ± 0.09 0.41 ± 0.09 0.50 ± 0.20 0.66 ± 0.18
0.15 0.12 0.16 0.2
(2) With the same processing condition, Al–5Ti and Al– 10Ti have similar Al3Ti particle size, while Al–15Ti has larger Al3Ti particle size. The heat release due to the exothermic Al–Ti reaction is expected to be higher for the Al–15Ti system than for Al–5Ti and Al–10Ti systems. As a consequence, a larger Al3Ti particle size results in the Al–15Ti system. 3.4. Mechanical properties Based upon the results of X-ray diffraction and microscopic observations (SEM and TEM), the microstructure of FSP Al–Ti alloys can be characterized as nanometer-sized Al3Ti particles dispersed in an ultrafine-grained Al matrix. Both tensile and compressive properties of the Al–Al3Ti composites produced by 4 FSP passes were determined and the results are summarized in
5246
C.J. Hsu et al. / Acta Materialia 54 (2006) 5241–5249
Table 3 Mechanical properties of Al–Al3Ti composites produced by FSP Sample
Al–5Ti/700/4 Al–5Ti/1400/4 Al–10Ti/700/4 Al–10Ti/1400/4 Al–15Ti/700/4 Al–15Ti/1400/4
Tension
Compression
E (MPa)
ry (MPa)
ru (MPa)
Elongation (%)
ry (MPa)
82 ± 6 80 ± 4 95 ± 4 91 ± 3 108 ± 5 114 ± 1
277 ± 7 250 ± 34 383 ± 5 316 ± 17 471 ± 30 406 ± 28
313 ± 7 300 ± 32 435 ± 17 366 ± 19 518 ± 8 471 ± 21
18 ± 2 18 ± 6 14 ± 2 6 ± 0.2 0.8 ± 0.4 1.0 ± 0.2
315 ± 4 288 ± 3 456 ± 4 432 ± 4 655 ± 66 684 ± 73
E, Young’s modulus; ry, 0.2% offset yield stress; ru, ultimate tensile strength.
Table 3. Several important conclusions can be drawn from Table 3: (a) Both Young’s modulus and strength increase significantly with increasing Ti content. This can be attributed to the increase in Al3Ti volume fraction with increasing Ti content. For an Al–15Ti material, the Young’s modulus reaches 114 GPa. (b) Both Al–5Ti and Al–10Ti exhibit good tensile ductility. However, Al–15Ti shows limited tensile elongation, which may be attributed to the presence of large amount of Al3Ti particles (V = 0.47–0.49) in the composite. (c) Materials produced with higher tool rotation speed (1400 rpm) exhibit lower yield strengths as compared with those produced at 700 rpm. This difference is more evident in tensile tests than in compressive tests. This may be related to the higher temperature introduced by the higher tool rotation speed, which may cause a coarser microstructure. (d) The compressive yield strength is higher than the tensile strength for all samples. 4. Discussion 4.1. In situ Al–Ti reaction during FSP In FSP, the time that the material is subjected to the thermomechanical action is very short, of the order of seconds. The Al–Ti reaction must be very fast. From differential thermal analysis (DTA), the major exothermic reaction between Al and Ti was found to occur after the melting of Al, evidenced by a peak at 1028 K (Fig. 9). Based on the irregular morphology of the interface between Ti and the reaction zone shown in Fig. 5(b), one may suspect that liquid phase might have been present at the Ti–Al interface. It is suggested that the rapid Al–Ti reaction in FSP may be attributed to the following mechanisms: (1) The large plastic strain (40) in FSP [36] can shear the metal powders and break the oxide film surrounding Al and Ti particles, which cause intimate contact between Al and Ti.
(2) The heat provided by the friction stir of the rotating tool can raise the temperature high enough to initiate the exothermic reaction between Al and Ti. (3) The heat of formation for Al3Ti at 298 K is 35.6 kJ/ g atom [37]. As a first approximation, the adiabatic temperature rise due to the formation of Al3Ti is estimated to be 1400 K by taking the heat capacity as 25 J/(K mol) [37]. The heat release due to the exothermic reaction can raise the temperature and accelerate the reaction. The heat release may be high enough to cause local melting of Al at the Al–Ti interface, which can enhance the reaction. The irregular morphology of the interface between Ti and the reaction zone shown in Fig. 5(b) indicates that liquid phase might have been present at the Ti–Al interface. However, the heat generated at the Al–Ti interface may dissipate rapidly into the surrounding Al because of the presence of large amounts of Al. Therefore, only local melting is expected. (4) The large plastic strain imposed in FSP can effectively remove the Al3Ti particles such that direct contact between Al and Ti can be maintained. Thus, the reaction can proceed rapidly at the interface. Since the Al3Ti particles are removed rapidly from the interface, the growth of the particles is limited and nanometer-sized particles result. Assuming a homogeneous distribution of Al3Ti particles, the Zener limiting grain size (dz) can be calculated by the equation dz = 4r/3V, where r and V are the radius and volume fraction of Al3Ti particles, respectively [38]. The calculated Zener limiting grain sizes are listed in Table 2. The measured Al grain size can be related to the Zener limiting grain size. This suggests that the presence of finely dispersed Al3Ti particles can limit the grain growth and result in an ultrafine grain size in the Al matrix. It is also noted that the Al3Ti particle size increases with increasing tool rotation speed and Ti content (Table 2). This can be attributed to a higher temperature resulting from the higher tool rotation speed and the higher Ti content. The higher tool rotation speed can generate more frictional heating, and more reaction heat can be released in the alloy with higher Ti content.
C.J. Hsu et al. / Acta Materialia 54 (2006) 5241–5249
Fig. 9. Differential thermal analysis (DTA) of the green compact of Al–10Ti alloy. The specimen was heated up to 1073 K at 20 K/min followed by cooling at 20 K/min.
4.2. Mechanical behavior Because of the high elastic modulus of Al3Ti, the Young’s modulus of the Al–Al3Ti composites increases significantly with increasing the volume fraction of Al3Ti as shown in Fig. 10. The Young’s modulus of particlereinforced composites can be predicted from the Halpin– Tsai equation [39] Em ð1 þ gqV Þ ð1Þ Ec ¼ 1 qV where q¼
ðEp =Em Þ 1 ðEp =Em Þ þ g
ð2Þ
Ec, Em, and Ep are the Young’s moduli of the composite, matrix, and particle, respectively, g is an adjustable parameter, and V is the volume fraction of the particles. For Al–Al3Ti composites, Em = 70 GPa and Ep = 216 GPa [33]. By taking g = 1, the calculated Young’s moduli of Al–Al3Ti composites are shown as a function of the volume fraction of Al3Ti particles in Fig. 10. The measured Young’s moduli of Al–Al3Ti composites agree well with the prediction of the Halpin–Tsai equation. This suggests that there is good bonding strength between Al3Ti particles and the Al matrix so that Al3Ti particles can contribute effectively to load sharing in the composites. As shown in Table 3, the compressive yield strength is always higher than the tensile strength for all materials studied. It is also noted that the difference between compressive and tensile yield strength increases with the Ti content as well as the tool rotation speed as shown in Fig. 11. This phenomenon may be attributed to the presence of residual stress in the composites. There are two possible causes for the residual stress in the composites produced by FSP. (1) The temperature gradient in FSW may result in significant tensile residual stress along the welding direction [40]. This residual stress is supported by the
5247
Fig. 10. Young’s modulus of Al–Al3Ti composites as a function of the volume fraction of Al3Ti (V). The prediction of the Halpin–Tsai equation is shown as a dotted line.
material surrounding the SZ. As the testing specimens were machined from the SZ, the residual stress must be relaxed considerably. Therefore, this could not be the major cause of the tension–compression strength difference. (2) The residual stress may result from the difference in the coefficient of thermal expansion (CTE) between the Al matrix and the Al3Ti particles. The CTE of Al3Ti is not known, but it is believed that the intermetallic compound has a lower CTE than Al. When the Al–Al3Ti composite is cooled from the processing temperature, tensile residual stresses are expected in the Al matrix. This means that when the composite is loaded, plastic flow occurs earlier in tension than in compression. The magnitude of the residual stress due to differential thermal contraction is expected to increase with the processing temperature, which increases with the Ti content and the tool rotation speed. Thus, the differential thermal contraction
Fig. 11. Difference between compressive and tensile yield strength of Al–Al3Ti composites as a function of both Ti content and tool rotation speed.
5248
C.J. Hsu et al. / Acta Materialia 54 (2006) 5241–5249
Table 4 Calculated contributions of Orowan strengthening (rOR) and grain size strengthening (rgs) to the yield strength of Al–Al3Ti composites Al–5Ti/700/4 Al–5Ti/1400/4 Al–10Ti/700/4 Al–10Ti/1400/4 Al–15Ti/700/4 Al–15Ti/1400/4
rOR (MPa)
rgs (MPa)
rOR + rgs (MPa)
ryt (MPa)
ryc (MPa)
117 89 372 320 738 522
93 73 148 99 129 104
210 161 520 419 867 626
277 ± 7 250 ± 34 383 ± 5 316 ± 17 471 ± 30 406 ± 28
315 ± 4 288 ± 3 456 ± 4 432 ± 4 655 ± 66 684 ± 73
ryt, measured tensile yield strength; ryc, measured compressive yield strength.
between Al and Al3Ti is believed to be the major cause for the difference between compressive and tensile yield strength. The possible strengthening mechanisms which may operate in particle-reinforced metal matrix composites are [41]: (1) Orowan strengthening. (2) Grain and substructure strengthening. (3) Quench hardening resulting from the dislocations generated to accommodate the differential thermal contraction between the reinforcing particles and the matrix. (4) Work hardening, due to the strain misfit between the elastic reinforcing particles and the plastic matrix. According to the characteristics of the microstructure, the major contributions to the yield strength of the Al– Al3Ti composites fabricated by FSP are: (1) the fine grain size of the Al matrix, and (2) the Orowan strengthening due to the fine dispersion of a large amount of Al3Ti particles. The contribution from the fine grain size can be calculated from the Hall–Petch equation pffiffiffi rgs ¼ r0 þ k= d ð3Þ where d is the grain size, and r0 = 13 MPa and k = 74 MPa lm1/2 for pure Al [42]. To estimate the contribution of Orowan strengthening by Al3Ti particles, it is assumed that the Al3Ti particles are spherical and uniformly distributed. The interparticle spacing, k, of Al3Ti particles can be calculated from [43] rffiffiffiffi rffiffiffi! p 2 k¼ 2 r ð4Þ V 3 where r is the radius of Al3Ti particles. The shear strength contributed from the Al3Ti particles can then be calculated using the modified Orowan equation given by Martin [43] pffiffiffiffiffiffiffiffi 0:81Gb sOR ¼ ð5Þ ln 2 2=3r=r0 1=2 2pð1 tÞ k where G (=26.2 GPa) is the matrix shear modulus, b (=0.286 nm) is the Burgers vector, t (=0.345) is Poisson’s ratio, and r0 (=4b) is the dislocation core radius. The contribution of the Orowan strengthening to tensile strength can be calculated as rOR = MsOR, where M is the Taylor
factor and M = 3 for a face-centered cubic polycrystal. Based upon the microstructural parameters listed in Tables 1 and 2, the contributions of grain size strengthening and Orowan strengthening to yield strength are calculated. As shown in Table 4, the high strength of these FSP Al–Ti alloys can be mainly attributed to (1) the grain size strengthening (rgs) which results from the ultrafine Al grain size, and (2) the Orowan strengthening (rOR) which is a result of the presence of a large volume fraction of nanometersized Al3Ti particles. 5. Conclusions (1) The rapid Al–Ti reaction in FSP is a combined result of the large plastic strain and high temperature introduced in FSP and the exothermic reaction between Al and Ti. The Al3Ti particle size is affected by both the Ti content and the FSP parameters. It increases with increasing tool rotation speed and Ti content, since a higher temperature can result from a higher tool rotation speed and higher Ti content. In addition, the Al3Ti particle size also increases with increasing number of FSP passes, which may be a result of a longer time exposed at elevated temperature with increasing FSP passes. (2) The Young’s modulus of the Al–Al3Ti composites increases significantly with increasing volume fraction of Al3Ti. For an Al–15Ti material, the Young’s modulus can reach 114 GPa, which is 63% higher than that of Al. The measured Young’s moduli of Al– Al3Ti composites agree well with the prediction of the Halpin–Tsai equation. (3) The high strength of these FSP Al–Al3Ti composites can be attributed to the presence of a large volume fraction of nanometer-sized Al3Ti particles, which contribute significantly to the strength through the Orowan mechanism, as well as the ultrafine grain size of the Al matrix. There is considerable difference between tensile and compressive yield strength, which increases with tool rotation speed and Ti content. This strength difference may be attributed to the presence of tensile residual stress in the aluminum matrix of Al–Al3Ti composites as a result of differential thermal contraction between Al and Al3Ti during cooling from the processing temperature.
C.J. Hsu et al. / Acta Materialia 54 (2006) 5241–5249
Acknowledgement This research was supported by the National Science Council of ROC (NSC93-2216-E-110-020). References [1] Thomas WM, Nicoholas ED, Needham JC, Church MG, Templesmith P, Dawes CJ. GB patent application 9125978.8, December 1991; US patent 5460317, October 1995. [2] Colligan K. Weld J 1999;78:229S. [3] Seidel TU, Reynolds AP. Metall Mater Trans A 2001;32A:2879. [4] Arbegast WJ. In: Jin Z et al., editors. Hot deformation of aluminum alloys III. Warrendale (PA): TMS; 2003. p. 313. [5] Mishra RS, Ma ZY. Mater Sci Eng R 2005;50:1. [6] Mishra RS, Mahoney MW, McFadden SX, Mara NA, Mukherjee AK. Scripta Mater 2000;42:163. [7] Lee WB, Yeon YM, Jung SB. Mater Sci Eng A 2003;355:154. [8] Kwon YJ, Shigematsu I, Saito N. Scripta Mater 2003;49:785. [9] Rhodes CG, Mahoney MW, Bingel WH, Calabrese M. Scripta Mater 2003;48:1451. [10] Su JQ, Nelson TW, Sterling CJ. Scripta Mater 2005;52:135. [11] Chang CI, Lee CJ, Huang JC. Scripta Mater 2004;51:509. [12] Ma ZY, Mishra RS, Mahoney MW. Acta Mater 2002;50:4419. [13] Ma ZY, Mishra RS. Acta Mater 2003;51:3551. [14] Charit I, Mishra RS. Mater Sci Eng A 2003;359:290. [15] Berbon PB, Bingel WH, Mishra RS, Bampton CC, Mahoney MW. Scripta Mater 2001;44:61. [16] Ma ZY, Mishra RS, Mahoney MW. Scripta Mater 2004;50:931. [17] Sharma SR, Ma ZY, Mishra RS. Scripta Mater 2004;51:237. [18] Mishra RS, Ma ZY, Charit I. Mater Sci Eng A 2003;341:307. [19] Hsu CJ, Kao PW, Ho NJ. Scripta Mater 2005;53:341. [20] Clyne TW, Withers PJ. An introduction to metal matrix composites. Cambridge: Cambridge University Press; 1993. [21] Tjong SC, Ma ZY. Mater Sci Eng R 2000;29:49.
5249
[22] Srinivasan S, Chen SR, Schwarz RB. Mater Sci Eng A 1992;153: 691. [23] Cheng YC, Wang SH, Kao PW, Chang CP. Mater Sci Forum 1996;217:1891. [24] Varin RA. Metall Mater Trans A 2002;33A:193. [25] Senkov ON, Froes FH, Stolyarov VV, Valiev RZ, Liu J. Nanostruct Mater 1998;10:691. [26] Wu JM, Zheng SL, Li ZZ. Mater Sci Eng A 2000;289:246. [27] Tong XC, Fang HS. Metall Mater Trans A 1998;29A:893. [28] Wang SH, Kao PW. Acta Mater 1998;46:2675. [29] Lerf R, Morris DG. Mater Sci Eng A 1990;128:119. [30] Mirchandani PK, Benn RC, Heck KA. In: Lee WE et al., editors. Light-weight alloys for aerospace applications. Warrendale (PA): TMS; 1989. p. 33. [31] Frazier WE, Koczak MJ. In: Kim YW et al., editors. Dispersion strengthened aluminum alloys. Warrendale (PA): TMS; 1988. p. 573. [32] Das SK, Davis LA. Mater Sci Eng 1988;98:1. [33] Nakamura M, Kimura K. J Mater Sci 1991;26:2208. [34] Wang SH, Kao PW, Chang CP. Scripta Mater 1999;40:289. [35] Feng CF, Froyen L. Acta Mater 1999;47:4571. [36] Heurtier P, Desrayaud C, Montheillet F. Mater Sci Forum 2002;396– 402:1537. [37] Gale WF, Totemeier TC. Smithells metals reference book. 8th ed. Elsevier; 2004. [38] Martin JW, Doherty RD. Stability of microstructures in metallic systems. Cambridge: Cambridge University Press; 1976. p. 234. [39] Hull D, Clyne TW. An introduction to composite materials. 2nd ed. Cambridge: Cambridge University Press; 1996. p. 66. [40] Peel M, Steuwer A, Preuss M, Withers PJ. Acta Mater 2003;51: 4791. [41] Lloyd DJ. Int Mater Rev 1994;39:1. [42] Yu CY, Kao PW, Chang CP. Acta Mater 2005;53:4019. [43] Martin JW. Micromechanisms in particle hardened alloys. Cambridge: Cambridge University Press; 1980. p. 60.
Al–Al3Ti nanocomposites produced in situ by friction stir processing C.J. Hsu, C.Y. Chang, P.W. Kao *, N.J. Ho, C.P. Chang Institute of Materials Science and Engineering, National Sun Yat-Sen University, No. 70 Lian-Hi Road, Kaohsiung 804, Taiwan Received 16 February 2006; received in revised form 25 April 2006; accepted 29 June 2006 Available online 14 September 2006
Abstract Aluminum reinforced with a large amount (up to 50 vol.%) of nanometer-sized Al3Ti particles can be fabricated from Al–Ti elemental powder mixtures via friction stir processing (FSP). This technique combines the hot working nature of FSP and the exothermic reaction between aluminum and titanium. For the production of intermetallic reinforced in situ composites, FSP can provide (a) severe deformation to promote mixing and refining of the constituent phases in the material, (b) elevated temperature to facilitate the formation of intermetallic phase in situ, and (c) hot consolidation to form a fully dense solid. Due to the fine dispersion of Al3Ti particles, the aluminum matrix has a submicrometer-grained structure. As a result of the high Al3Ti content and very fine microstructure, the composites possess enhanced Young’s modulus and strength. 2006 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Metal matrix composites; Friction stir processing; Intermetallic compound
1. Introduction Friction stir welding (FSW) was invented at The Welding Institute (TWI) in the UK in 1991 as a solid-state joining technique [1]. In FSW, a non-consumable rotating tool with a specially designed pin and shoulder is plunged into the interface between two plates to be joined and traversed along the line of the joint. Localized heating is produced by the friction between the rotating tool and the workpiece to raise the local temperature of the material to the range where it can be plastically deformed easily. As the rotating tool traverses along the joint line, metal is essentially extruded around the tool before being forged by the large down pressure [2–5]. The material flow during FSW is quite complex depending on the tool geometry, process parameters, and material to be welded. Reynolds and co-workers [2,3] investigated the material flow behavior in FSW using a marker insert technique. They suggested that there is a secondary, circular motion around the longitudinal axis of the weld, which is a combined result of the stirring
*
Corresponding author. Tel.: +886 7 5252000; fax: +886 7 5254099. E-mail address: [email protected] (P.W. Kao).
action and the extrusion around the pin. The intense plastic deformation around the tool and the friction between tool and workpiece both contribute to the temperature increase in the stirred zone during FSW. However, temperature measurements within the stirred zone are very difficult due to the intense plastic deformation produced by the rotation and translation of the tool. The maximum temperature in the stirred zone during FSW of various aluminum alloys is found to be between 0.6Tm and 0.9Tm, which depends on the ratio of rotation rate to translation speed [5]. Generally, an increase in the ratio of rotation rate to translation speed will increase the peak temperature in FSW. A comprehensive literature review of friction stir welding has been given by Mishra and Ma [5]. Based on the principles of FSW, Mishra et al. [6] developed friction stir processing (FSP) for microstructural modification of materials. FSP has been shown as an effective technique in many applications: (a) to form ultrafinegrained structure in Al alloys [7–10] and Mg alloys [11], (b) to produce a fine-grained microstructure, which exhibits superplasticity [6,12–14], (c) to homogenize the microstructure of nanocomposite aluminum alloys [15], (d) to refine the microstructure of cast aluminum alloys [16,17], and (e) to fabricate a surface composite of Al–SiC on an
1359-6454/$30.00 2006 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.actamat.2006.06.054
5242
C.J. Hsu et al. / Acta Materialia 54 (2006) 5241–5249
aluminum substrate [18]. Recently, FSP has been successfully applied to produce ultrafine-grained Al–Al2Cu in situ composite from an Al–Cu elemental powder mixture [19]. Metal matrix composites (MMCs) are an important class of material for structural and electrical applications [20]. Particulate-reinforced MMCs are of particular interest because of their easy fabrication, low cost, and isotropic properties. In conventionally processed power metallurgy composites, the reinforcing particles are formed prior to their addition to the matrix metal. In this case, the scale of reinforcing phase is limited by the starting powder size, which is typically of the order of several to tens of micrometers and rarely below 1 lm. Other drawbacks for conventionally processed MMCs that have to be overcome are poor interfacial bonding and poor wettability between the reinforcement and the matrix due to surface contamination of the reinforcements. It is widely recognized that the mechanical properties of MMCs are controlled by the size and volume fraction of the reinforcements as well as the nature of the matrix–reinforcement interface [20]. Superior mechanical properties can be achieved when fine and stable reinforcements with good interfacial bonding are dispersed uniformly in the matrix. A possible alternative is to synthesize the reinforcement in situ in the metal matrix [21]. The advantages of in situ MMCs are that they have more homogeneous microstructures and are thermodynamically more stable. Moreover, they also have strong interfacial bonding between the reinforcements and the matrices. In situ aluminum matrix composites may be fabricated by various techniques such as conventional ingot metallurgy, mechanical alloying (MA), rapid solidification processing (RS), combustion synthesis, etc. The microstructure produced by conventional ingot metallurgy is rather coarse. Materials produced by MA, RS, or combustion synthesis must be densified by a hot consolidation process such as hot isostatic pressing, hot pressing or hot extrusion in order to obtain the final product. Aluminum alloys reinforced with aluminide particles (NiAl3, FeAl3, TiAl3, etc.) possess high specific strength, high specific modulus, and excellent properties both at ambient and elevated temperature [22–31]. As compared to most other aluminum-rich intermetallics, Al3Ti is very attractive because it has a higher melting point (1623 K) and relatively low density (3.4 g/cm3). Further, Ti has low diffusivity and solubility in aluminum; hence Al3Ti can be expected to exhibit a low coarsening rate at elevated temperature [32]. In addition, the Young’s modulus of Al3Ti phase has been determined to be 216 GPa [33]. Therefore, the presence of Al3Ti phase is very effective in increasing the stiffness of aluminum alloys. Al–Al3Ti composites have been fabricated by RS [26,27] and MA [28–31] to achieve high strength. The major strengthening mechanisms, which contribute to the high strength of Al–Al3Ti alloys, have been suggested to include Orowan strengthening, grain size strengthening, and load-shearing effects of Al3Ti particles [26,34,35].
The objective of the work reported in this study was to produce fully dense Al–Al3Ti composites with a large amount of nanometer-sized Al3Ti particles as reinforcement. The basic idea for fabricating the Al–Al3Ti composites is to combine the hot working nature of FSP and the exothermic reaction between aluminum and titanium. The FSP is utilized to provide the following functions: (a) severe plastic deformation to promote mixing and refining of the constituent phases in the material, (b) elevated temperature to facilitate the formation of intermetallic phase, and (c) hot consolidation to form a fully dense solid. In this paper, we present the microstructure and mechanical properties of nanometer-sized Al3Ti particle-reinforced aluminum composites produced via FSP. 2. Experimental The starting materials used were aluminum powder (99.7% purity, 325 mesh) and titanium powder (99.1% purity, 325 mesh). Titanium contents of 5, 10, or 15 at.% were pre-mixed with aluminum powder (denoted Al–5Ti, Al–10Ti, and Al–15Ti). The pre-mixed Al–Ti alloy powders were cold compacted to 12 · 20 · 88 mm billets in a steel die set using a pressure of 225 MPa. To improve the billet strength for easier handling in FSP, the green compact was sintered for 20 min in air at 823 K. The tool pin used in FSP is a standard M1.2*6 (metrication, with pitch height and diameter of 1.2 and 6 mm, respectively). Counterclockwise tool rotation with a speed of 700 or 1400 rpm was used, and the rotating tool was traversed at a speed of 45 mm/min along the long axis of the billet. Multiple FSP passes were applied to the billet to enhance the Al–Ti reaction. For multiple FSP, the pin tool was moved along the same line, and the FSP pass was applied after the workpiece had been cooled from the previous FSP pass. A schematic of the FSP setup is shown in Fig. 1. X-ray diffraction (XRD; Cu Ka radiation) was utilized to identify the phases present in the specimens. Scanning electron microscopy (SEM; JSM-6400 and JSM-6330) was used to study the distribution of second-phase particles
Fig. 1. Schematic of the friction stir process.
C.J. Hsu et al. / Acta Materialia 54 (2006) 5241–5249
5243
in the stirred zone (SZ) and the base metal (BM). Thin foils cut from the SZ were studied using analytical electron microscopy (AEM; JEOL 3010, operated at 200 kV). The Vickers microhardness of the SZ was measured with a 300 g load for 15 s. Mechanical properties of specimens machined from the SZ were evaluated. The tests were carried out using an Instron 5582 universal testing machine with an initial strain rate of 1 · 103 s1. The dimensions of the gauge section of tensile specimens were 3 mm in diameter and 14 mm in length. Rectangular specimens with dimensions of 4 · 4 · 6 mm3 were used for compression tests. The stress direction was aligned along the traverse direction of FSP. 3. Results 3.1. Al–Ti reaction during sintering The compacted billet was sintered to provide enough strength to prevent cracking during FSP. After sintering at 823 K for 20 min, only diffraction peaks of Al and Ti appeared as shown in Fig. 2. The formation of Al3Ti phase was observed in an Al–15Ti specimen sintered at 883 K for 20 min as shown in Fig. 2. The 883 K sintered specimen was also examined using SEM/backscattered electron imaging (BEI). The Al3Ti phase (light gray) was found to enclose the Ti particles (white) as shown in Fig. 3(a). In addition, Al3Ti nodules, consisting of faceted particles of a few micrometers in size, were also observed (Fig. 3(b)). In order to avoid the formation of coarse Al3Ti particles, the billets for FSP were sintered at 823 K for 20 min.
Fig. 3. (a) SEM/BEI of Al–15Ti sintered at 883 K. (b) A magnified view of the selected area in (a). The white phase is titanium, the light gray phase is Al3Ti, and the dark matrix is Al.
3.2. Al–Ti reaction during FSP
Fig. 2. XRD patterns for 823 and 883 K sintered Al–15Ti specimens.
XRD was used to identify the phases present in the SZ of specimens during FSP. The diffraction patterns showed that Ti reacted with Al to form Al3Ti, but some unreacted Ti remained after four FSP passes. A typical example of the XRD patterns of the FSP specimen is shown in Fig. 4. The time that material is affected by the rotating pin in FSP can be considered as the processing time. Based upon the pin diameter and the tool traversing speed, the processing time is estimated to be about 8 s. Therefore, the formation of Al3Ti phase is very rapid in FSP. The microstructure of FSP specimens was observed using SEM/BEI as shown in Fig. 5. The white particles of micrometer size are pure Ti which has been verified using energy dispersive spectroscopy (EDS), and the very fine gray particles (<100 nm) are Al3Ti particles which are uniformly dispersed in the Al matrix. After FSP, the average size of Ti particles is refined from 40 lm to about 1–5 lm. Some of the Ti particles after FSP show an irregular morphology (Fig. 5(b)). The appearance of irregular morphology at the interface between Ti and Al + Al3Ti suggests that liquid phase might
5244
C.J. Hsu et al. / Acta Materialia 54 (2006) 5241–5249
Fig. 4. XRD pattern of Al–15Ti alloy produced by four FSP passes with tool rotation at 700 rpm.
strain imposed in FSP. Thus, the reaction can proceed rapidly at the Al–Ti interface. From image analyses of the SEM/BEI images, the amount of residual Ti was quantitatively determined. Since Al3Ti is the only reaction product between Al and Ti observed in the present case, the volume fraction of Al3Ti particles in each specimen can be calculated from the reacted Ti. The quantitative results of unreacted Ti and the volume fraction of Al3Ti particles in each specimen are listed in Table 1. In general, the Al3Ti content for each composition increases with increasing FSP passes. The values of microhardness of the SZ, which are also listed in Table 1, show a general increasing trend with increasing Al3Ti content. For Al–15Ti after 4 FSP passes, the volume fraction of Al3Ti is close to 0.5, which results in a hardness value of 200 Hv. From microscopic observations, the SZ was found to be free of porosity. In addition, the measured density of material cut from the SZ agrees well with the theoretical density. These evidences indicate that the FSP sample is fully dense. 3.3. Microstructure of Al–Al3Ti nanocomposites observed using transmission electron microscopy (TEM) The cross-section of the SZ in specimens produced by FSP was also examined using TEM. Typical microstructures are shown in Fig. 6, which shows a large number of fine Al3Ti particles uniformly dispersed in an ultrafine-grained Al matrix. The fine Al3Ti particles were found within the grain interior as well as along the grain boundaries of the Al matrix (Fig. 7). The Al3Ti particles were identified by the use of electron diffraction as shown in Fig. 8. In addition, the Al3Ti particles were also confirmed by the EDS analysis with TEM, which showed that the Al:Ti atomic ratio is near 75:25. In general, the Al3Ti particles have equiaxed shape and often with facets (Fig. 8).
Table 1 Quantitative results of the volume fraction of residual Ti (VTi), the calculated volume fraction of Al3Ti (V), and the microhardness of Al–Al3Ti composites (Hv) produced by various FSP parameters
Fig. 5. (a) SEM/BEI of Al–5Ti alloy produced by four FSP passes with tool rotation at 700 rpm. (b) SEM/BEI of Al–10Ti alloy produced by four FSP passes with tool rotation at 700 rpm, which shows a Ti particle (white phase) with irregular morphology.
be present at the Al–Ti reaction front. The liquid phase is considered as Al melt, which results from the exothermic Al–Ti reaction. The liquid Al can react rapidly with Ti to form Al3Ti phase. The Al3Ti particles are effectively removed and dispersed in the Al matrix by the large plastic
Sample
Rotation speed (rpm)/passes
Residual Ti (VTi)
Calculated Al3Ti (V)
Al–5Ti
700/2 700/4 1400/2 1400/4
3.75 3.32 3.79 3.51
0.06 0.07 0.06 0.07
86 ± 6 97 ± 3 93 ± 7 98 ± 5
Al–10Ti
700/2 700/4 1400/2 1400/4
5.28 3.97 6.33 3.02
0.20 0.24 0.16 0.28
107 ± 9 148 ± 7 122 ± 2 127 ± 9
Al–15Ti
700/2 700/4 1400/2 1400/4
7.91 3.24 7.16 2.69
0.30 0.47 0.33 0.49
144 ± 9 226 ± 9 148 ± 7 192 ± 12
Microhardness (Hv)
C.J. Hsu et al. / Acta Materialia 54 (2006) 5241–5249
Fig. 6. TEM bright-field image showing the nanometer-size Al3Ti particles uniformly distributed in the Al matrix of an Al–10Ti/700/4 specimen.
5245
Fig. 8. TEM central dark-field image showing a nearly equiaxed Al3Ti particle in an Al–15Ti/700/4 specimen with inset showing the selected area diffraction pattern in [1 0 0] zone axis.
Table 2 Average sizes of Al grains and Al3Ti particles in Al–Al3Ti composites produced by various FSP parameters
Fig. 7. TEM central dark-field image showing the distribution of Al3Ti particles within the grain interior as well as along the grain boundaries of the Al matrix in an Al–5Ti/700/4 specimen.
Quantitative results of the size of Al3Ti particles and Al grains for various alloys and processing conditions are listed in Table 2. In general, the average size of Al grains has been refined to the submicrometer level, except Al– 5Ti processed at 1400 rpm for 4 passes, which has a grain size slightly larger than 1 lm. From Table 2, we also have the following observations: (1) For the same Ti content, the Al3Ti particle size increases with increasing tool rotation speed and increasing FSP passes. This may be attributed to higher temperature associated with the higher tool rotation speed, and longer time exposed at elevated temperature with increasing FSP passes.
Sample
Rotation speed (rpm)/ passes
Al–5Ti
700/2 700/4 1400/2 1400/4
Al–10Ti Al–15Ti
Al3Ti particle size (nm)
Al grain size (lm)
Zener limiting grain size (lm)
47 ± 22 60 ± 31 51 ± 24 87 ± 33
0.90 ± 0.27 0.86 ± 0.32 0.72 ± 0.21 1.53 ± 0.39
0.52 0.57 0.57 0.83
700/4 1400/4
53 ± 13 83 ± 20
0.30 ± 0.12 0.74 ± 0.24
0.15 0.26
700/2 700/4 1400/2 1400/4
69 ± 21 83 ± 22 78 ± 21 147 ± 36
0.33 ± 0.09 0.41 ± 0.09 0.50 ± 0.20 0.66 ± 0.18
0.15 0.12 0.16 0.2
(2) With the same processing condition, Al–5Ti and Al– 10Ti have similar Al3Ti particle size, while Al–15Ti has larger Al3Ti particle size. The heat release due to the exothermic Al–Ti reaction is expected to be higher for the Al–15Ti system than for Al–5Ti and Al–10Ti systems. As a consequence, a larger Al3Ti particle size results in the Al–15Ti system. 3.4. Mechanical properties Based upon the results of X-ray diffraction and microscopic observations (SEM and TEM), the microstructure of FSP Al–Ti alloys can be characterized as nanometer-sized Al3Ti particles dispersed in an ultrafine-grained Al matrix. Both tensile and compressive properties of the Al–Al3Ti composites produced by 4 FSP passes were determined and the results are summarized in
5246
C.J. Hsu et al. / Acta Materialia 54 (2006) 5241–5249
Table 3 Mechanical properties of Al–Al3Ti composites produced by FSP Sample
Al–5Ti/700/4 Al–5Ti/1400/4 Al–10Ti/700/4 Al–10Ti/1400/4 Al–15Ti/700/4 Al–15Ti/1400/4
Tension
Compression
E (MPa)
ry (MPa)
ru (MPa)
Elongation (%)
ry (MPa)
82 ± 6 80 ± 4 95 ± 4 91 ± 3 108 ± 5 114 ± 1
277 ± 7 250 ± 34 383 ± 5 316 ± 17 471 ± 30 406 ± 28
313 ± 7 300 ± 32 435 ± 17 366 ± 19 518 ± 8 471 ± 21
18 ± 2 18 ± 6 14 ± 2 6 ± 0.2 0.8 ± 0.4 1.0 ± 0.2
315 ± 4 288 ± 3 456 ± 4 432 ± 4 655 ± 66 684 ± 73
E, Young’s modulus; ry, 0.2% offset yield stress; ru, ultimate tensile strength.
Table 3. Several important conclusions can be drawn from Table 3: (a) Both Young’s modulus and strength increase significantly with increasing Ti content. This can be attributed to the increase in Al3Ti volume fraction with increasing Ti content. For an Al–15Ti material, the Young’s modulus reaches 114 GPa. (b) Both Al–5Ti and Al–10Ti exhibit good tensile ductility. However, Al–15Ti shows limited tensile elongation, which may be attributed to the presence of large amount of Al3Ti particles (V = 0.47–0.49) in the composite. (c) Materials produced with higher tool rotation speed (1400 rpm) exhibit lower yield strengths as compared with those produced at 700 rpm. This difference is more evident in tensile tests than in compressive tests. This may be related to the higher temperature introduced by the higher tool rotation speed, which may cause a coarser microstructure. (d) The compressive yield strength is higher than the tensile strength for all samples. 4. Discussion 4.1. In situ Al–Ti reaction during FSP In FSP, the time that the material is subjected to the thermomechanical action is very short, of the order of seconds. The Al–Ti reaction must be very fast. From differential thermal analysis (DTA), the major exothermic reaction between Al and Ti was found to occur after the melting of Al, evidenced by a peak at 1028 K (Fig. 9). Based on the irregular morphology of the interface between Ti and the reaction zone shown in Fig. 5(b), one may suspect that liquid phase might have been present at the Ti–Al interface. It is suggested that the rapid Al–Ti reaction in FSP may be attributed to the following mechanisms: (1) The large plastic strain (40) in FSP [36] can shear the metal powders and break the oxide film surrounding Al and Ti particles, which cause intimate contact between Al and Ti.
(2) The heat provided by the friction stir of the rotating tool can raise the temperature high enough to initiate the exothermic reaction between Al and Ti. (3) The heat of formation for Al3Ti at 298 K is 35.6 kJ/ g atom [37]. As a first approximation, the adiabatic temperature rise due to the formation of Al3Ti is estimated to be 1400 K by taking the heat capacity as 25 J/(K mol) [37]. The heat release due to the exothermic reaction can raise the temperature and accelerate the reaction. The heat release may be high enough to cause local melting of Al at the Al–Ti interface, which can enhance the reaction. The irregular morphology of the interface between Ti and the reaction zone shown in Fig. 5(b) indicates that liquid phase might have been present at the Ti–Al interface. However, the heat generated at the Al–Ti interface may dissipate rapidly into the surrounding Al because of the presence of large amounts of Al. Therefore, only local melting is expected. (4) The large plastic strain imposed in FSP can effectively remove the Al3Ti particles such that direct contact between Al and Ti can be maintained. Thus, the reaction can proceed rapidly at the interface. Since the Al3Ti particles are removed rapidly from the interface, the growth of the particles is limited and nanometer-sized particles result. Assuming a homogeneous distribution of Al3Ti particles, the Zener limiting grain size (dz) can be calculated by the equation dz = 4r/3V, where r and V are the radius and volume fraction of Al3Ti particles, respectively [38]. The calculated Zener limiting grain sizes are listed in Table 2. The measured Al grain size can be related to the Zener limiting grain size. This suggests that the presence of finely dispersed Al3Ti particles can limit the grain growth and result in an ultrafine grain size in the Al matrix. It is also noted that the Al3Ti particle size increases with increasing tool rotation speed and Ti content (Table 2). This can be attributed to a higher temperature resulting from the higher tool rotation speed and the higher Ti content. The higher tool rotation speed can generate more frictional heating, and more reaction heat can be released in the alloy with higher Ti content.
C.J. Hsu et al. / Acta Materialia 54 (2006) 5241–5249
Fig. 9. Differential thermal analysis (DTA) of the green compact of Al–10Ti alloy. The specimen was heated up to 1073 K at 20 K/min followed by cooling at 20 K/min.
4.2. Mechanical behavior Because of the high elastic modulus of Al3Ti, the Young’s modulus of the Al–Al3Ti composites increases significantly with increasing the volume fraction of Al3Ti as shown in Fig. 10. The Young’s modulus of particlereinforced composites can be predicted from the Halpin– Tsai equation [39] Em ð1 þ gqV Þ ð1Þ Ec ¼ 1 qV where q¼
ðEp =Em Þ 1 ðEp =Em Þ þ g
ð2Þ
Ec, Em, and Ep are the Young’s moduli of the composite, matrix, and particle, respectively, g is an adjustable parameter, and V is the volume fraction of the particles. For Al–Al3Ti composites, Em = 70 GPa and Ep = 216 GPa [33]. By taking g = 1, the calculated Young’s moduli of Al–Al3Ti composites are shown as a function of the volume fraction of Al3Ti particles in Fig. 10. The measured Young’s moduli of Al–Al3Ti composites agree well with the prediction of the Halpin–Tsai equation. This suggests that there is good bonding strength between Al3Ti particles and the Al matrix so that Al3Ti particles can contribute effectively to load sharing in the composites. As shown in Table 3, the compressive yield strength is always higher than the tensile strength for all materials studied. It is also noted that the difference between compressive and tensile yield strength increases with the Ti content as well as the tool rotation speed as shown in Fig. 11. This phenomenon may be attributed to the presence of residual stress in the composites. There are two possible causes for the residual stress in the composites produced by FSP. (1) The temperature gradient in FSW may result in significant tensile residual stress along the welding direction [40]. This residual stress is supported by the
5247
Fig. 10. Young’s modulus of Al–Al3Ti composites as a function of the volume fraction of Al3Ti (V). The prediction of the Halpin–Tsai equation is shown as a dotted line.
material surrounding the SZ. As the testing specimens were machined from the SZ, the residual stress must be relaxed considerably. Therefore, this could not be the major cause of the tension–compression strength difference. (2) The residual stress may result from the difference in the coefficient of thermal expansion (CTE) between the Al matrix and the Al3Ti particles. The CTE of Al3Ti is not known, but it is believed that the intermetallic compound has a lower CTE than Al. When the Al–Al3Ti composite is cooled from the processing temperature, tensile residual stresses are expected in the Al matrix. This means that when the composite is loaded, plastic flow occurs earlier in tension than in compression. The magnitude of the residual stress due to differential thermal contraction is expected to increase with the processing temperature, which increases with the Ti content and the tool rotation speed. Thus, the differential thermal contraction
Fig. 11. Difference between compressive and tensile yield strength of Al–Al3Ti composites as a function of both Ti content and tool rotation speed.
5248
C.J. Hsu et al. / Acta Materialia 54 (2006) 5241–5249
Table 4 Calculated contributions of Orowan strengthening (rOR) and grain size strengthening (rgs) to the yield strength of Al–Al3Ti composites Al–5Ti/700/4 Al–5Ti/1400/4 Al–10Ti/700/4 Al–10Ti/1400/4 Al–15Ti/700/4 Al–15Ti/1400/4
rOR (MPa)
rgs (MPa)
rOR + rgs (MPa)
ryt (MPa)
ryc (MPa)
117 89 372 320 738 522
93 73 148 99 129 104
210 161 520 419 867 626
277 ± 7 250 ± 34 383 ± 5 316 ± 17 471 ± 30 406 ± 28
315 ± 4 288 ± 3 456 ± 4 432 ± 4 655 ± 66 684 ± 73
ryt, measured tensile yield strength; ryc, measured compressive yield strength.
between Al and Al3Ti is believed to be the major cause for the difference between compressive and tensile yield strength. The possible strengthening mechanisms which may operate in particle-reinforced metal matrix composites are [41]: (1) Orowan strengthening. (2) Grain and substructure strengthening. (3) Quench hardening resulting from the dislocations generated to accommodate the differential thermal contraction between the reinforcing particles and the matrix. (4) Work hardening, due to the strain misfit between the elastic reinforcing particles and the plastic matrix. According to the characteristics of the microstructure, the major contributions to the yield strength of the Al– Al3Ti composites fabricated by FSP are: (1) the fine grain size of the Al matrix, and (2) the Orowan strengthening due to the fine dispersion of a large amount of Al3Ti particles. The contribution from the fine grain size can be calculated from the Hall–Petch equation pffiffiffi rgs ¼ r0 þ k= d ð3Þ where d is the grain size, and r0 = 13 MPa and k = 74 MPa lm1/2 for pure Al [42]. To estimate the contribution of Orowan strengthening by Al3Ti particles, it is assumed that the Al3Ti particles are spherical and uniformly distributed. The interparticle spacing, k, of Al3Ti particles can be calculated from [43] rffiffiffiffi rffiffiffi! p 2 k¼ 2 r ð4Þ V 3 where r is the radius of Al3Ti particles. The shear strength contributed from the Al3Ti particles can then be calculated using the modified Orowan equation given by Martin [43] pffiffiffiffiffiffiffiffi 0:81Gb sOR ¼ ð5Þ ln 2 2=3r=r0 1=2 2pð1 tÞ k where G (=26.2 GPa) is the matrix shear modulus, b (=0.286 nm) is the Burgers vector, t (=0.345) is Poisson’s ratio, and r0 (=4b) is the dislocation core radius. The contribution of the Orowan strengthening to tensile strength can be calculated as rOR = MsOR, where M is the Taylor
factor and M = 3 for a face-centered cubic polycrystal. Based upon the microstructural parameters listed in Tables 1 and 2, the contributions of grain size strengthening and Orowan strengthening to yield strength are calculated. As shown in Table 4, the high strength of these FSP Al–Ti alloys can be mainly attributed to (1) the grain size strengthening (rgs) which results from the ultrafine Al grain size, and (2) the Orowan strengthening (rOR) which is a result of the presence of a large volume fraction of nanometersized Al3Ti particles. 5. Conclusions (1) The rapid Al–Ti reaction in FSP is a combined result of the large plastic strain and high temperature introduced in FSP and the exothermic reaction between Al and Ti. The Al3Ti particle size is affected by both the Ti content and the FSP parameters. It increases with increasing tool rotation speed and Ti content, since a higher temperature can result from a higher tool rotation speed and higher Ti content. In addition, the Al3Ti particle size also increases with increasing number of FSP passes, which may be a result of a longer time exposed at elevated temperature with increasing FSP passes. (2) The Young’s modulus of the Al–Al3Ti composites increases significantly with increasing volume fraction of Al3Ti. For an Al–15Ti material, the Young’s modulus can reach 114 GPa, which is 63% higher than that of Al. The measured Young’s moduli of Al– Al3Ti composites agree well with the prediction of the Halpin–Tsai equation. (3) The high strength of these FSP Al–Al3Ti composites can be attributed to the presence of a large volume fraction of nanometer-sized Al3Ti particles, which contribute significantly to the strength through the Orowan mechanism, as well as the ultrafine grain size of the Al matrix. There is considerable difference between tensile and compressive yield strength, which increases with tool rotation speed and Ti content. This strength difference may be attributed to the presence of tensile residual stress in the aluminum matrix of Al–Al3Ti composites as a result of differential thermal contraction between Al and Al3Ti during cooling from the processing temperature.
C.J. Hsu et al. / Acta Materialia 54 (2006) 5241–5249
Acknowledgement This research was supported by the National Science Council of ROC (NSC93-2216-E-110-020). References [1] Thomas WM, Nicoholas ED, Needham JC, Church MG, Templesmith P, Dawes CJ. GB patent application 9125978.8, December 1991; US patent 5460317, October 1995. [2] Colligan K. Weld J 1999;78:229S. [3] Seidel TU, Reynolds AP. Metall Mater Trans A 2001;32A:2879. [4] Arbegast WJ. In: Jin Z et al., editors. Hot deformation of aluminum alloys III. Warrendale (PA): TMS; 2003. p. 313. [5] Mishra RS, Ma ZY. Mater Sci Eng R 2005;50:1. [6] Mishra RS, Mahoney MW, McFadden SX, Mara NA, Mukherjee AK. Scripta Mater 2000;42:163. [7] Lee WB, Yeon YM, Jung SB. Mater Sci Eng A 2003;355:154. [8] Kwon YJ, Shigematsu I, Saito N. Scripta Mater 2003;49:785. [9] Rhodes CG, Mahoney MW, Bingel WH, Calabrese M. Scripta Mater 2003;48:1451. [10] Su JQ, Nelson TW, Sterling CJ. Scripta Mater 2005;52:135. [11] Chang CI, Lee CJ, Huang JC. Scripta Mater 2004;51:509. [12] Ma ZY, Mishra RS, Mahoney MW. Acta Mater 2002;50:4419. [13] Ma ZY, Mishra RS. Acta Mater 2003;51:3551. [14] Charit I, Mishra RS. Mater Sci Eng A 2003;359:290. [15] Berbon PB, Bingel WH, Mishra RS, Bampton CC, Mahoney MW. Scripta Mater 2001;44:61. [16] Ma ZY, Mishra RS, Mahoney MW. Scripta Mater 2004;50:931. [17] Sharma SR, Ma ZY, Mishra RS. Scripta Mater 2004;51:237. [18] Mishra RS, Ma ZY, Charit I. Mater Sci Eng A 2003;341:307. [19] Hsu CJ, Kao PW, Ho NJ. Scripta Mater 2005;53:341. [20] Clyne TW, Withers PJ. An introduction to metal matrix composites. Cambridge: Cambridge University Press; 1993. [21] Tjong SC, Ma ZY. Mater Sci Eng R 2000;29:49.
5249
[22] Srinivasan S, Chen SR, Schwarz RB. Mater Sci Eng A 1992;153: 691. [23] Cheng YC, Wang SH, Kao PW, Chang CP. Mater Sci Forum 1996;217:1891. [24] Varin RA. Metall Mater Trans A 2002;33A:193. [25] Senkov ON, Froes FH, Stolyarov VV, Valiev RZ, Liu J. Nanostruct Mater 1998;10:691. [26] Wu JM, Zheng SL, Li ZZ. Mater Sci Eng A 2000;289:246. [27] Tong XC, Fang HS. Metall Mater Trans A 1998;29A:893. [28] Wang SH, Kao PW. Acta Mater 1998;46:2675. [29] Lerf R, Morris DG. Mater Sci Eng A 1990;128:119. [30] Mirchandani PK, Benn RC, Heck KA. In: Lee WE et al., editors. Light-weight alloys for aerospace applications. Warrendale (PA): TMS; 1989. p. 33. [31] Frazier WE, Koczak MJ. In: Kim YW et al., editors. Dispersion strengthened aluminum alloys. Warrendale (PA): TMS; 1988. p. 573. [32] Das SK, Davis LA. Mater Sci Eng 1988;98:1. [33] Nakamura M, Kimura K. J Mater Sci 1991;26:2208. [34] Wang SH, Kao PW, Chang CP. Scripta Mater 1999;40:289. [35] Feng CF, Froyen L. Acta Mater 1999;47:4571. [36] Heurtier P, Desrayaud C, Montheillet F. Mater Sci Forum 2002;396– 402:1537. [37] Gale WF, Totemeier TC. Smithells metals reference book. 8th ed. Elsevier; 2004. [38] Martin JW, Doherty RD. Stability of microstructures in metallic systems. Cambridge: Cambridge University Press; 1976. p. 234. [39] Hull D, Clyne TW. An introduction to composite materials. 2nd ed. Cambridge: Cambridge University Press; 1996. p. 66. [40] Peel M, Steuwer A, Preuss M, Withers PJ. Acta Mater 2003;51: 4791. [41] Lloyd DJ. Int Mater Rev 1994;39:1. [42] Yu CY, Kao PW, Chang CP. Acta Mater 2005;53:4019. [43] Martin JW. Micromechanisms in particle hardened alloys. Cambridge: Cambridge University Press; 1980. p. 60.