- Email: [email protected]
Transient liquid phase diffusion bonding of 6061-15 wt% SiCp in argon environment
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
journal homepage: www.elsevier.com/locate/jmatprotec
Transient liquid phase diffusion bonding of 6061-15 wt% SiCp in argon environment J. Maity a,∗ , T.K. Pal b , R. Maiti c a
Department of Metallurgical and Materials Engineering, National Institute of Technology, Durgapur, Durgapur 713209, West Bengal, India b Welding Technology Center, Department of Metallurgical and Material Engineering, Jadavpur University, Kolkata 700032, West Bengal, India c Central Research Facility, Indian Institute of Technology, Kharagpur 721302, West Bengal, India
a r t i c l e
i n f o
a b s t r a c t
Article history:
Extruded 6061-15 wt% SiCp composite was joined by transient liquid phase diffusion (TLPD)
Received 30 March 2008
bonding process in argon environment using 50-m thick copper foil interlayer. The bonding
Received in revised form
was carried out at 560 ◦ C with two different applied pressures (0.1 and 0.2 MPa) and five differ-
9 August 2008
ent holding times (20 min, 1, 2, 3 and 6 h). Kinetics of the bonding process was significantly
Accepted 22 August 2008
accelerated in the presence of reinforcement (SiC). This acceleration is attributed to the increased solute diffusivity through defect-rich SiC particle/matrix interface and porosity. Adequate bond strength (90% of the original composite strength) was achieved for bonding at
Keywords:
0.2 MPa pressure with 6 h of holding. This is very close to the reported highest bond strength
Transient liquid phase diffusion
achieved (92% of the original composite strength) for joining aluminium-based metal matrix
bonding
composite by TLPD process in vacuum followed by isostatic pressing. The rejection of oxide
6061-SiCp composite
at periphery on completion of isothermal solidification, and elimination of void at bond
Particle segregation
interface through solid state diffusion at higher pressure (0.2 MPa) were the main reasons
Isothermal solidification
of achieving high bond strength.
Oxidation
© 2008 Elsevier B.V. All rights reserved.
Bond strength
1.
Introduction
A major problem in widespread industrial application of aluminium-based metal matrix composite (AlMMC) is the difficulty encountered in joining (Ellis, 1997). Mechanical fastening (bolting or riveting), fusion welding and solid state diffusion bonding of such composites involve several difficulties such as damage of reinforcement (for mechanical fastening); formation of brittle phase (Al4 C3 ), HAZ cracking and weld porosity (for fusion welding); and excessive plastic deformation under high applied pressure (for solidstate diffusion bonding) (Bushby and Scott, 1995; Devletian,
∗
1987; Field, 1989; Gittos and Threadgill, 1991; Hall and Manrique, 1995; Luhman et al., 1983; Shirzadi and Wallach, 1997). The transient liquid phase diffusion (TLPD) bonding process which employs an ‘interlayer’ (often a pure metal) for the formation of low melting point composition (e.g. eutectic), has the advantage of lower bonding temperature, lower bonding pressure and less surface finish requirement than solid-state diffusion bonding. However, completion of the TLPD bonding process requires a long time mainly due to isothermal solidification stage (Natsume et al., 2003). Commercial application of this technique requires adequate bond strength development with regard
Corresponding author. Tel.: +91 343 2755237; fax: +91 343 2547375. E-mail addresses: joydeep [email protected] (J. Maity), [email protected] (T.K. Pal), [email protected] (R. Maiti). 0924-0136/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jmatprotec.2008.08.015
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
to understanding the microstructural variation and process kinetics. Although a number of investigations have been carried out on TLPD bonding of different monolithic metals and alloys, reports on the TLPD bonding of AlMMCs are limited. Among different interlayer used in TLPD bonding of monolithic aluminium-based alloys, the use of copper interlayer has proved to be successful for joining conventional aluminium alloys, and bond strength comparable to that of the parent material has been reported (Dray, 1985). Again, the published literature on TLPD bonding of AlMMCs mainly dealt with the development of bonding conditions using different thickness of copper interlayer in order to achieve adequate bond strength. While studying TLPD bonding of SiC fiber reinforced AlMMC with 10-m thick copper interlayer at 550 ◦ C in air environment, it has been reported by Bushby and Scott (1993) that higher bonding pressure (20 MPa) was necessary in order to limit oxidation of copper and maximize the bonded area to 80%. On the other hand, bond strength of 92% of the parent material strength was achieved by Shirzadi and Wallach (1997) for joining AlMMC by TLPD process at 560 ◦ C, 0.1–0.2 MPa pressure, using a 7-m thick copper interlayer with 20 min bonding time in vacuum followed by isostatic pressing. In other investigation, using mixed powder interlayer (Al–Si–SiC–Ti), bond strength of 50 MPa was achieved by Huang et al. (2007) for joining 6063–SiCp composite by TLPD process in vacuum at 595 ◦ C, 0.003 MPa, with a bonding time of 90 min. Whereas in argon environment, bond strength of 68 MPa was achieved by one of the present authors (Pal, 2005) for joining extruded 606115 wt% SiCp using copper powder interlayer at 560 ◦ C, 2 MPa, with a bonding time of 20 min. In most of these studies of TLPD bonding using copper interlayer, bonding temperature was kept at 550–560 ◦ C, which is slightly above the eutectic temperature of Al–Cu system (548 ◦ C) and below the solidus temperature of AlMMC. However, different pressures were used for bonding performed under different conditions. TLPD bonding in air environment requires very high-pressure application (20 MPa) in order to achieve metal-to-metal contact at bond interface (Bushby and Scott, 1993). The high pressure causes excessive plastic deformation of AlMMC, which is not desired. Therefore, conventional TLPD bonding is performed at lower pressure (0.1–0.2 MPa) in vacuum or inert environment in order to achieve adequate bond strength without plastic deformation. Again, for low pressure conventional TLPD bonding in vacuum with lower bonding time (20 min), presence of void at bond interface has been identified by Shirzadi and Wallach (1997). These voids were responsible for lowering down the bond strength to a certain extent. The low pressure bonding with lower time of holding (20 min) was followed by isostatic pressing to achieve high bond strength. However, the low pressure TLPD bonding with higher bonding time (say, 6 h) has not been studied for AlMMC. Moreover, in all these investigations of TLPD bonding of AlMMC, no explicit correlation was made between bond microstructure and different stages of the process. Also, bonding time was kept low (maximum 2 h) without any correlation with the completion of isothermal solidification or homogenization of bond region, and no comparison of process kinetics was made with monolithic system. Present investigation aims at developing adequate bond strength for extruded 6061-15 wt% SiCp com-
3569
Table 1 – Chemical composition of 6061 alloy (wt%) Mg Si Cu Cr Al
1.0 0.6 0.3 0.2 Rest
posite, by TLPD process in argon environment, with regard to process mechanism and microstructural evaluation, for different bonding times up to 6 h.
2.
Materials and methods
2.1.
Material
As-received material was an extruded rod of AlMMC consisting of 6061 matrix alloy and 15 wt% (12.93 vol%) silicon carbide (SiC) particulate reinforcement of 23 m average size. The nominal composition of 6061 alloy (Anon., 1990) is given in Table 1. In addition it contains some iron (0.6 wt%) as an impurity which was confirmed by chemical analysis in optical emission spectrometer (UNISPEC: 4L/0096). The density of as-received AlMMC was also measured by water displacement method.
2.2.
Specimen preparation for bonding
The extruded rod was machined to produce discs of 15 mm diameter and 10 mm height. As a result the faying surfaces of discs became perpendicular to the extrusion direction. The faying surfaces of discs were polished to 1 m finish. Pure copper (99.97 wt%) foil of 50 m thickness was used as interlayer. The interlayer was punched out to a diameter of 15 mm for bonding. The interlayer and polished faying surfaces of discs were finally rinsed in acetone and dried by a hot air blast just before bonding.
2.3.
TLPD bonding
The interlayer was placed between the polished faying surfaces of the two AlMMC discs. This assembly was then set by an adhesive tape and inserted inside the diffusion bonding unit. The bonding was carried out in a programmable electric furnace keeping bond centerline horizontal. A thermocouple inserted into the drilled hole in one of each pair of discs was used to monitor bonding temperature. The argon gas (99.99% Ar, 3–5 ppm O2 , 3 ppm H2 O, 2 ppm H2 , 1 ppm CO2 , 1 ppm CO) was flown into the bonding chamber at a rate 5 l/min to maintain inert atmosphere. The bonding temperature was kept at 560 ◦ C which is above the eutectic temperature (548 ◦ C) of Al–Cu system (Anon., 1992) and below the solidus temperature (582 ◦ C) of 6061 matrix alloy (Anon., 1990). The specimens were heated to the bonding temperature (560 ◦ C) at a rate of 6 ◦ C/min, held at that temperature for five different lengths of time (bonding time), viz. 20 min, 1, 2, 3 and 6 h, and cooled down to 540 ◦ C at a rate of 5 ◦ C/min inside the furnace. Then the specimens were taken out of the furnace and cooled in still air to the room temperature. Two different pressures, 0.1 and 0.2 MPa, were applied for bonding.
3570
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
2.4.
Optical metallography
Bonded cylindrical samples of 15 mm diameter were sectioned perpendicular to the bonding plane using a low speed diamond cutter. The section was polished to 1 m finish and etched with Keller’s reagent. Both qualitative and quantitative study of microstructure around bond interface was carried out for these metallographic specimens using optical microscope with digital photomicrography (ZEISS, Imager.A1m). The interface width was measured at ‘periphery’ and ‘central zone’ of the bond interface. The bond centerline had 15 mm length. The two edges of bond centerline, with each edge of 3.5 mm length, were considered as ‘periphery’. The remaining part at the middle, of 8 mm length, was considered as ‘central zone’.
2.5.
Fig. 1 – Specimen loaded in jig: the arrangement for shear strength (bond strength) determination.
Mechanical testing
Bonded cylindrical samples were machined to 10 mm diameter to eliminate edge effects. These specimens of approximately 20 mm length and 10 mm diameter and asreceived AlMMC of similar dimension were loaded in a specially prepared jig, which is schematically shown in Fig. 1. The grips of the jig were pulled in tension in a
Fig. 2 – Microstructure of as-received 6061-SiCp composite—(a) optical image: bands of SiC along extrusion direction; (b) SEM secondary electron image: SiC bands and porosity and (c) SEM back-scattered electron image: presence of FeAlX.
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
3571
Fig. 3 – Typical optical microstructures of bonded specimens at central zone of bond interface at 0.2 MPa pressure: (a) 20 min; (b) 1 h; (c) 2 h; (d) 3 h and (e) 6 h.
100-kN capacity universal testing machine (INSTRON-8801) at a cross-head speed of 0.5 mm/min in position control mode such that the specimen experienced pure shear stress across the bond interface. The maximum load was divided by the bond area in order to calculate shear strength. For each bonding condition three specimens were tested and the average value is considered as shear strength (bond strength).
2.6.
SEM and EDS analysis
As-polished section of the metallographic specimens was examined around bond interface under scanning electron microscope (JEOL, JSM-5800). Microstructures were studied with both back-scattered electron image mode and secondary electron image mode. Accordingly different phases were identified by EDS spot analysis. Also the progress
3572
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
Fig. 4 – SEM back-scattered electron image of bonded specimen (20 min, 0.2 MPa).
of diffusion was studied by line scan. The line scan for concentration variation of copper was carried out along a line of 600 m length lying perpendicular to the bond interface keeping bond centerline approximately at the middle. The fractured surfaces of shear-tested specimens were also studied under the scanning electron microscope with both back-scattered electron image mode and secondary electron image mode. Different phases present on fractured surface were identified with EDS spot analysis.
3.
Results and discussion
3.1.
As-received composite
The microstructure of as-received AlMMC, as shown in Fig. 2, clearly reveals bands of SiC-rich areas along the extrusion direction as well as some porosity near the particle cluster (band). The presence of porosity is also indicated in the measured density of the composite (2.72 gm cm−3 ) that shows lower value than the theoretical density of com-
Fig. 5 – Schematic representation of bond region during isothermal solidification.
posite (2.77 gm cm−3 ) based on the calculation from the reported density of 6061 matrix alloy (2.70 gm cm−3 ) and SiC (3.21 gm cm−3 ) (Anon., 1990, 1991). SEM back-scattered electron image (Fig. 2(c)) and EDS spot analysis reveal the presence of intermetallic phases of iron and aluminium (FeAlX in general) in as-received AlMMC. Both FeAl3 (containing about 61 wt% Al) and Fe2 Al5 (containing 56 wt% Al) are identified. Chemical analysis by optical emission spectrometer indicates the presence of about 0.6 wt% iron impurity in the composite. The presence of iron impurity (about 0.7 wt% Fe) in 6061 alloy was reported (Rohatgi et al., 1986). Also, the presence of intermetallic phases of Fe and Al in AlMMC containing iron impurity was observed (Bushby and Scott, 1993). This is due to very little solubility of iron
Table 2 – Interface width at central zone and periphery Bonding pressure (MPa)
Bonding time
Interface width (m) At central zone
At periphery
0.1
20 min 1h 2h 3h 6h
230 107 93 Negligible Negligible
185 177 183 163 143
0.2
20 min 1h 2h 3h 6h
223 86 73 Negligible Negligible
205 173 171 196 134
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
3573
Fig. 6 – Line scan result representing concentration variation of Cu around bond interface for 6 h of holding at 0.1 MPa pressure (X-axis length = 600 m).
in aluminium as indicated in Fe–Al phase diagram (Anon., 1992).
3.2.
Microstructural evaluation and mechanism
Based on the study of monolithic system, the TLPD bonding process was described as consisting of four stages namely, (I) dissolution of interlayer, (II) homogenization of liquid (widening of liquid to its maximum width), (III) isothermal solidification, and (IV) homogenization of bond region (Duvall et al., 1974; Nakao et al., 1989; Natsume et al., 2003; Tuah-Poku et al., 1988). The first two stages together take few minutes. However, duration of last two stages may extend to several hours. In present study, heating from eutectic temperature (548 ◦ C) to the bonding temperature (560 ◦ C) at a rate 6 ◦ C/min takes 2 min. It is possible that the first two stages (dissolution of interlayer and widening of liquid) are completed during heating to the bonding temperature. Optical microstructures of TLPD bonded specimens in etched condition representing the bond interface at central zone is shown in Fig. 3. SEM backscattered electron image of the bond interface is presented in Fig. 4. The specimens with lowest bonding time (20 min) exhibit isothermally solidified grains (mainly of primary ˛) adjacent to bond interface (Fig. 3(a)). Furthermore, presence of CuAl2 phase (containing about 54 wt% Cu) is observed adjacent to the bond interface (Fig. 4). It is to be noted that as-received AlMMC (unbonded) does not contain any CuAl2 intermetallic phase. In bonded composites CuAl2 phase is likely to precipitate out in isothermally solidified zone during cooling from bonding temperature as solubility of Cu in primary ˛ decreases with decreasing temperature following solvus curve. Therefore, liquid widening process completes and isothermal solidification starts before 20 min of holding.
Fig. 7 – Segregation of FeAlX at bond interface (1 h, 0.2 MPa).
The presence of 15 wt% SiC particle in 6061 aluminium matrix is expected to have significant effect on TLPD bonding process as compared to pure monolithic system. Microstructural study reveals the segregation of SiC particles at bond
Table 3 – Result of shear test Bonding condition As-received composite (no bonding) 20 min, 0.1 MPa 1 h, 0.1 MPa 2 h, 0.1 MPa 3 h, 0.1 MPa 6 h, 0.1 MPa 20 min, 0.2 MPa 1 h, 0.2 MPa 2 h, 0.2 MPa 3 h, 0.2 MPa 6 h, 0.2 MPa
Shear strength (MPa) 105 55 50 34 75 76 69 54 49 77 95
Fig. 8 – Fractured surface of as-received composite: (a) secondary electron image and (b) back-scattered electron image.
3574
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
Fig. 9 – Secondary electron images of the fractured surfaces of bonded composites: (a) 20 min, 0.1 MPa; (b) 20 min, 0.2 MPa; (c) 1 h, 0.1 MPa; (d) 1 h, 0.2 MPa; (e) 2 h, 0.1 MPa and (f) 2 h, 0.2 MPa.
centerline with isothermally solidified zones on both sides. During bonding the bond centerline was kept horizontal. Since no preferential segregation occurs towards the lower AlMMC disc, this is not the case of gravity segregation. According to published literature on general solidification characteristic of SiC reinforced AlMMC, the primary ˛ is very efficient at rejecting SiC, and pushing the particles ahead of the solid/liquid
interface (Gallerneault et al., 1991). In this regard a critical velocity of solid/liquid interface has been reported, below which the SiC particles are pushed by the moving interface and above which they are engulfed (Rohatgi et al., 1986). Since the first two stages of TLPD bonding (dissolution of interlayer and widening of liquid) are very fast, the velocity of solid/liquid interface is also very high. As a result, SiC par-
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
3575
Fig. 10 – Back-scattered electron images of the fractured surfaces of bonded composites: (a) 1 h, 0.1 MPa; (b) 1 h, 0.2 MPa; (c) 3 h, 0.1 MPa; (d) 3 h, 0.2 MPa; (e) 6 h, 0.1 MPa and (f) 6 h, 0.2 MPa.
ticles are not pushed by the solid/liquid interface away from the bond centerline during widening of liquid. The next stage, isothermal solidification, is slow due to solid-state diffusion controlling the process and takes several hours for completion. During isothermal solidification, due to low velocity of solid/liquid interface, most of the SiC particles are pushed by the moving solid/liquid interface, as evident in the present work. As a result, particles segregate at bond centerline along with liquid phase and the residual liquid gets solidified during
cooling. This aggregate of residual liquid and segregated SiC particles may be called as ‘Segregation Zone’ and the width of this segregation zone may be termed as ‘interface width’. This is schematically shown in Fig. 5. The measured interface width at central zone and at periphery of bond interface is presented in Table 2. At 20 min bonding time, the interface width at central zone is somewhat similar (slightly greater) to that at periphery. However, with increasing bonding time interface width at central zone decreases continuously and always
3576
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
Fig. 11 – Secondary electron images of the fractured surfaces of bonded composites: (a) 3 h, 0.1 MPa; (b) 3 h, 0.2 MPa; (c) 6 h, 0.1 MPa and (d) 6 h, 0.2 MPa.
remains lower than that at periphery. This phenomenon indicates that during TLPD bonding under pressure (0.1 and 0.2 MPa) liquid–particle aggregate at interface moves towards periphery and flows out. The interface width at periphery exhibits decrease and occasional increase depending on the flow of liquid–particle aggregate from center to periphery and from periphery to outside. Also, with increasing bonding time reduction of interface width at central zone is more at higher pressure (0.2 MPa) than at lower pressure (0.1 MPa). This again indicates more mass flow towards periphery under higher pressure. It is important to note that specimens with 3 and 6 h bonding time (for both the pressures 0.1 and 0.2 MPa) exhibit negligible interface width at central zone with least segregation of SiC particles as shown in Fig. 3(d) and (e). Furthermore, the bond interface is hardly discernible and grain continuity exists across the interface indicating the completion of isothermal solidification in 3 h of bonding time. The EDS spot analysis of matrix at bond interface, for the specimen with 3 h bonding time and 0.2 MPa pressure, exhibits the presence of 2.11 wt% Cu. According to Al–Cu phase diagram, at bonding temperature (560 ◦ C) the maximum solubility of copper in primary ˛ is 4.35 wt% (Anon., 1992). Therefore, the copper content of 2.11 wt% indicates the presence of primary ˛ grains. This further confirms the completion of isothermal solidification.
Natsume et al. (2003) investigated the mechanism of TLPD bonding for joining pure Al by 50 m Cu interlayer at 570 ◦ C up to a maximum bonding time of 1 h. According to their experimental data, melting of interlayer took about 15 sec. The liquid widened to its maximum width of 460 m for 60 s of holding. It is to be noted that the specimen was not held at bonding temperature till the completion of isothermal solidification. Up to 1 h of maximum holding liquid width decreased to 406 m through isothermal solidification. Whereas, in the present investigation interface width at central zone has reduced to 107 m for 0.1 MPa pressure and 86 m for 0.2 MPa pressure after 1 h of holding. This indicates much faster kinetics of isothermal solidification in case of AlMMC as compared to the pure monolithic system. In the present study, AlMMC contains substantial amount (15 wt%) of SiC particles. The presence of SiC particles in the metallic matrix leads to the formation of defect-rich interfacial region of high dislocation density, mainly due to the difference in coefficient of thermal expansion between metallic matrix and SiC particles (Papazian, 1988). The coefficient of thermal expansion of 6061 matrix alloy and SiC particle are 23.6 × 10−6 K−1 and 5.5 × 10−6 K−1 , respectively (Anon., 1990, 1991). Thus, it is likely that the particle/matrix interface of composite is associated with high dislocation density. In addition, as discussed earlier, the com-
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
3577
Fig. 12 – Secondary electron images of metallographic specimens near the edge of periphery of bond interface: (a) 3 h, 0.1 MPa; (b) 3 h, 0.2 MPa; (c) 6 h, 0.1 MPa and (d) 6 h, 0.2 MPa.
posite contains porosity. These defects provide short circuit paths for diffusion. Therefore, at the bonding temperature of 560 ◦ C (which is nearer to the solidus temperature, 582 ◦ C, of 6061 alloy), along with lattice diffusion, short circuit diffusion is also expected to be operative. It is reported that short circuit diffusivities are larger than lattice diffusivities by a factor of 103 near the melting point (Gjostein, 1972). As a result, diffusion becomes faster. Moreover, due to application of pressure some part of liquid flows out of periphery reducing the amount of liquid to be solidified isothermally. The EDS line scan for Cu concentration variation across the bond interface indicates diffusion of Cu away from the interface. An occasional rise in copper concentration is due to the presence of copper enriched phase CuAl2 . However, on a gross scale, distribution of copper in the matrix for 6 h of bonding is found to be uniform across the bond centerline (Fig. 6). This indicates homogenization of bond region. The CuAl2 phase is found to be present at bond interface (segregation zone) as well as in isothermally solidified zone. At the bond interface CuAl2 forms through eutectic solidification of residual liquid during cooling. At the bonding temperature this residual liquid remains intermixed with segregated SiC particles. SEM backscattered electron image (Fig. 4) clearly reveals that during cooling CuAl2 preferentially nucleates on SiC par-
ticles at bond interface. On the other hand, at the bonding temperature, isothermally solidified zone contains the grains of primary ˛. During cooling from bonding temperature, as the solubility of Cu decreases, CuAl2 precipitates out of primary ˛. In bonded composite the particles of intermetallic phases of iron impurity (FeAlX ) is found to be segregated around bond centerline (Fig. 7). It is likely that during isothermal solidification these particles are pushed by solid/liquid interface and thereby segregate at bond interface (segregation zone) along with SiC particles.
3.3.
Bond strength and fractured surface
The result of shear test is given in Table 3. In general, the as-received composite exhibits higher shear strength as compared to bonded specimens. However, the specimen bonded at 0.2 MPa pressure with 6 h holding time shows the highest shear strength (bond strength) of 95 MPa which about 90% of the as-received composite strength (105 MPa). The fractured surface of shear tested as-received composite exhibits dimples and tear ridges (Fig. 8(a)), characteristic of a typical ductile fracture of Al/SiCp composite (Ge and Schmauder, 1995). The presence of FeAlX is revealed in back scattered electron image (Fig. 8(b)). However, fractured surface does not contain CuAl2
3578
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
Fig. 13 – Secondary electron images of metallographic specimens at the central zone of bond Interface (higher magnification): (a) 3 h, 0.1 MPa; (b) 6 h, 0.1 MPa; (c) 3 h, 0.2 MPa and (d) 6 h, 0.2 MPa.
phase since this phase is absent in as-received composite. In bonded composite, failure is found to occur through the segregation zone. The specimens with 20 min, 1 h and 2 h bonding time, at both the pressures (0.1 and 0.2 MPa) show the presence of oxide on fractured surface as indicated in secondary electron images (Fig. 9). Decohesion between oxide and metallic matrix is clearly evident. The presence of oxide indicates that during bonding the residual oxygen present in the system oxidizes the transient liquid phase. EDS analysis reveals that these oxides are the oxides of Al, Mg and Si. Apart from oxides, presence of brittle phases such as FeAlX and CuAl2 are also revealed in back scattered electron images (Fig. 10). Furthermore, in these bonding conditions, segregation of SiC particles at bond interface is observed in metallographic study (Fig. 3(a–c)). Accordingly, the fractured surface (Fig. 9) appears to be flat (devoid of dimples), characteristic of brittle fracture. In general, these specimens exhibit poor bond strength. Out of these, the specimen bonded at higher pressure (0.2 MPa) with lowest bonding time (20 min) possesses relatively higher bond strength (69 MPa). Similar bond strength (68 MPa) was achieved by one of the present authors (Pal, 2005) for joining same composite using copper powder interlayer of 0.8 mm thickness at 560 ◦ C with 20 min of holding under 2 MPa pressure. Fractured surface of this specimen contains relatively lesser extent of
oxide (Fig. 9(b)). Oxidation is expected to be less for lower time of holding at higher pressure. It has been reported that higher pressure lowers the extent of oxidation due to more liquid expulsion (Bushby and Scott, 1993). However, specimens with 3 h and 6 h bonding time, for both the pressures, show least presence of oxide on fractured surface (Fig. 11). In these bonding conditions, segregation of SiC particles at bond interface is negligible as revealed in metallographic study (Fig. 3(d and e)). Therefore, the bond strength achieved is relatively higher. However, presence of brittle phases (CuAl2 , FeAlX ) is still observed (Fig. 10). In general, these specimens exhibit presence of dimples along with cleavage facets indicating mixed mode of failure. While investigating metallographic specimens with 3 h and 6 h bonding time (prepared by sectioning bonded specimens of 15 mm diameter along axial length) in SEM secondary electron image mode, the presence of oxide near the edge of periphery is observed (Fig. 12). Therefore, it appears that on completion of isothermal solidification (3 h of holding) the solid/liquid interfaces from both sides approach each other and merge together and eventually the residual liquid along with oxide is rejected at the edge of periphery. Once the isothermal solidification completes, the entire bond region becomes solid and no further oxide phase can form at the bond interface through oxidation
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
of liquid for longer time of holding (3 h or more). Moreover, bonded composites of 15 mm diameter were machined to 10 mm diameter in order to prepare specimens for shear test. As a result, the edge of periphery was machined out. Therefore, for 3 h and 6 h bonding time, the shear tested specimens of 10 mm diameter do not exhibit any significant presence of oxides on fractured surface. Before completion of isothermal solidification, the liquid phase has maximum exposure to environment for 2 h of holding that results in severe oxidation. Thus specimen bonded with 2 h of holding at lower pressure (0.1 MPa) exhibit the lowest bond strength (34 MPa). In general, before completion of isothermal solidification, higher the bonding time (more exposure of liquid) and lower the pressure, greater is the extent of oxidation of liquid and lower is the bond strength. It is also interesting to note that at 0.1 MPa pressure, bonded composites with 3 h and 6 h bonding time show similar bond strength (75 MPa and 76 MPa, respectively). The SEM study in secondary electron image mode on metallographic specimens at higher magnification reveals the presence of voids at bond interface for both the bonded composites (Fig. 13(a and b)). The voids are likely to be generated due to the volumetric shrinkage of metal during isothermal solidification. The density of liquid aluminium (2400 kg m−3 ) is lower than that of solid aluminium (2700 kg m−3 ) (Natsume et al., 2003). On the other hand, at 0.2 MPa pressure, the composite bonded with 6 h of holding shows much higher bond strength (95 MPa) than that with 3 h bonding time (77 MPa). The SEM study shows that the bond interface of the former (0.2 MPa, 6 h) is almost free from void (Fig. 13(d)), while the later (0.2 MPa, 3 h) contains void at bond interface (Fig. 13(c)). Therefore, under higher pressure (0.2 MPa), on completion of isothermal solidification, solid state diffusion eliminates void at bond interface, thereby improving bond strength. The fractured surface (Fig. 11(d)) of this composite bonded at 0.2 MPa pressure with 6 h of holding appears somewhat similar to that of as-received composite (Fig. 8(a)) and bond strength achieved is also closer (90% of the shear strength of as-received composite). It is important to note that, while joining AlMMCs by TLPD process, Shirzadi and Wallach (1997) achieved the highest bond strength, 92% of the parent material strength, using 7-m thick copper interlayer at 560 ◦ C, 0.1–0.2 MPa, with 20 min of holding in vacuum followed by isostatic pressing (costlier process). Therefore, in present study, bond strength very close to the highest achieved bond strength is obtained in argon environment using thicker interlayer (50 m) with higher bonding time (6 h).
4.
Conclusion
(i) TLPD bonding process of 6061-SiCp composite occurs much faster than that of pure aluminium. Isothermal solidification takes about 3 h and bond region gets homogenized in 6 h. (ii) In composite, the presence of defect-rich particle/matrix interface and porosity makes diffusion process faster. In addition, liquid expulsion under applied pressure reduces the amount of residual liquid to be solidified. As a result the duration of isothermal solidification is reduced.
3579
(iii) The CuAl2 phase is found to be present at bond interface (segregation zone) as well as in the isothermally solidified zone. The intermetallic phases of iron and aluminium (FeAlX ) segregate mainly at the bond interface. (iv) While bonding in commercial argon environment, the transient liquid gets oxidized. In general, the presence of oxide and other brittle phases (CuAl2 , FeAlX ) lowers down the bond strength. (v) On completion of isothermal solidification oxides get rejected to the periphery along with residual liquid, thereby reducing the amount of oxides at bond interface. Furthermore, longer time holding (6 h) under higher pressure (0.2 MPa) removes the void at bond interface through solid-state diffusion. As a result, bond strength is improved. (vi) TLPD bonding at 560 ◦ C under 0.2 MPa pressure with 6 h of holding in commercial argon environment produces a bond strength of 90% of the parent AlMMC strength, which is very close to the bond strength achieved (92% of the parent AlMMC strength) for TLPD process carried out at 560 ◦ C under 0.1–0.2 MPa pressure with 20 min of holding in vacuum followed by isostatic pressing.
references
Anon., 1990. Properties and Selection: Non-ferrous Alloys and Special-Purpose Materials Metals Handbook, vol. 2., 10th ed. ASM International, The Materials Information Society, pp. 102–103. Anon., 1991. Ceramics and Glasses. Engineering Materials Handbook, vol. 4. ASM international, The Materials Information Society, p. 30. Anon., 1992. Alloy Phase Diagrams, ASM Handbook, vol. 3. ASM International, The Materials Information Society, 2.44-3, p. 10. Bushby, R.S., Scott, V.D., 1993. Liquid Phase bonding of aluminium and aluminium/Nicalon composite using copper interlayers. Mater. Sci. Technol. 9, 417–423. Bushby, R.S., Scott, V.D., 1995. Liquid phase bonding of aluminium and aluminium/Nicalon composite using interlayers of Cu–Ag alloy. Mater. Sci. Technol. 11, 643–649. Devletian, J.H., 1987. SiC/Al metal matrix composite welding by a capacitor discharge process. Weld. J. 66, 33–39. Dray, A.E., 1985. Diffusion bonding of aluminium. PhD Thesis. University of Cambridge. Duvall, D.S., Owczarski, W.A., Paulonis, D.F., 1974. TLP bonding: a new method for joining heat resistant alloys. Weld. J. 53, 203–214. Ellis, M.B.D., 1997. Joining of metal matrix composites: a review. TWI J. 6, 69–128. Field, D.J., 1989. Aluminium Alloys-Contemporary Research and Applications. London Academic Press, pp. 523–537. Gallerneault, M., Kaya, M., Smith, R.W., Dellamore, G.W., 1991. Ultimate sediment concentration in directionally solidified Al–Si/SiC particulate metal matrix composites. In: Proceedings Conference on Extraction, Refining and Fabrication of Light Metals. Pergamon Press Inc., New York, USA, pp. 69–81. Ge, X., Schmauder, S., 1995. Micromechanism of fracture in Al/SiC composites. J. Mater. Sci. 30, 173–178. Gittos, M.F., Threadgill, P.L., 1991. Preliminary studies of joining particulate reinforced aluminium alloy metal matrix composites. In: Proceedings of the International Symposium on Metal Matrix Composites III. Exploiting the Investment, Inst. of Metals, December, pp. 10–11.
3580
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
GJOSTEIN, N.A., 1972. Short circuit diffusion. Diffusion, Papers Presented at a Seminar of the American Society for Metals, American Society for Metals, Metals Park, OH, October 14–15, p. 241. Hall, W.I., Manrique, F., 1995. Surface treatment of carbon fiber for aluminium alloy matrix composites. Scripta Met. 33, 2037–2043. Huang, J., Dong, Y., Wan, Y., Zhao, X., Zhang, H., 2007. Investigation on reactive diffusion bonding of SiCp/6063 MMC by using mixed powders as interlayers. J. Mater. Process. Technol. 190, 312–316. Luhman, T.S., Williams, R.L., Das, K.B., 1983. Development of joint and joining techniques for metal matrix composites. In: Fourth Quarterly Progress Report, Army Materials and Mechanics Research Center, pp. 1–58. Nakao, Y., Nishimoto, K., Shinozaki, K., Kang, C., 1989. Theoretical research on transient liquid insert metal diffusion bonding of nickel base alloys. Trans. Jpn. Weld. Soc. 20, 60–65.
Natsume, Y., Ohsasa, K., Tayu, Y., Momono, T., Narita, T., 2003. Numerical modeling of transient liquid-phase diffusion bonding process of al using cu filler metal. ISIJ Int. 43, 1976–1982. Pal, T.K., 2005. Joining of aluminium metal matrix composites. Mater. Manuf. Process. 20, 717–726. Papazian, J.M., 1988. Effects of SiC Whiskers and particles on precipitation in aluminium matrix composites. Metall. Trans. A 19A, 2945–2953. Rohatgi, P.K., Asthana, Das, R.S., 1986. Solidification, structures, and properties of cast metal–ceramic particle composites. Int. Met. Rev. 31, 115–136. Shirzadi, A.A., Wallach, E.R., 1997. New approaches for transient liquid phase diffusion bonding of aluminium based metal matrix composites. Mater. Sci. Technol. 13, 135–142. Tuah-Poku, I., Dollar, M., Massalski, T.B., 1988. A study of the Transient liquid phase bonding process applied to a Ag/Cu/Ag sandwich joint. Metall. Trans. A 19A, 675–686.
journal homepage: www.elsevier.com/locate/jmatprotec
Transient liquid phase diffusion bonding of 6061-15 wt% SiCp in argon environment J. Maity a,∗ , T.K. Pal b , R. Maiti c a
Department of Metallurgical and Materials Engineering, National Institute of Technology, Durgapur, Durgapur 713209, West Bengal, India b Welding Technology Center, Department of Metallurgical and Material Engineering, Jadavpur University, Kolkata 700032, West Bengal, India c Central Research Facility, Indian Institute of Technology, Kharagpur 721302, West Bengal, India
a r t i c l e
i n f o
a b s t r a c t
Article history:
Extruded 6061-15 wt% SiCp composite was joined by transient liquid phase diffusion (TLPD)
Received 30 March 2008
bonding process in argon environment using 50-m thick copper foil interlayer. The bonding
Received in revised form
was carried out at 560 ◦ C with two different applied pressures (0.1 and 0.2 MPa) and five differ-
9 August 2008
ent holding times (20 min, 1, 2, 3 and 6 h). Kinetics of the bonding process was significantly
Accepted 22 August 2008
accelerated in the presence of reinforcement (SiC). This acceleration is attributed to the increased solute diffusivity through defect-rich SiC particle/matrix interface and porosity. Adequate bond strength (90% of the original composite strength) was achieved for bonding at
Keywords:
0.2 MPa pressure with 6 h of holding. This is very close to the reported highest bond strength
Transient liquid phase diffusion
achieved (92% of the original composite strength) for joining aluminium-based metal matrix
bonding
composite by TLPD process in vacuum followed by isostatic pressing. The rejection of oxide
6061-SiCp composite
at periphery on completion of isothermal solidification, and elimination of void at bond
Particle segregation
interface through solid state diffusion at higher pressure (0.2 MPa) were the main reasons
Isothermal solidification
of achieving high bond strength.
Oxidation
© 2008 Elsevier B.V. All rights reserved.
Bond strength
1.
Introduction
A major problem in widespread industrial application of aluminium-based metal matrix composite (AlMMC) is the difficulty encountered in joining (Ellis, 1997). Mechanical fastening (bolting or riveting), fusion welding and solid state diffusion bonding of such composites involve several difficulties such as damage of reinforcement (for mechanical fastening); formation of brittle phase (Al4 C3 ), HAZ cracking and weld porosity (for fusion welding); and excessive plastic deformation under high applied pressure (for solidstate diffusion bonding) (Bushby and Scott, 1995; Devletian,
∗
1987; Field, 1989; Gittos and Threadgill, 1991; Hall and Manrique, 1995; Luhman et al., 1983; Shirzadi and Wallach, 1997). The transient liquid phase diffusion (TLPD) bonding process which employs an ‘interlayer’ (often a pure metal) for the formation of low melting point composition (e.g. eutectic), has the advantage of lower bonding temperature, lower bonding pressure and less surface finish requirement than solid-state diffusion bonding. However, completion of the TLPD bonding process requires a long time mainly due to isothermal solidification stage (Natsume et al., 2003). Commercial application of this technique requires adequate bond strength development with regard
Corresponding author. Tel.: +91 343 2755237; fax: +91 343 2547375. E-mail addresses: joydeep [email protected] (J. Maity), [email protected] (T.K. Pal), [email protected] (R. Maiti). 0924-0136/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jmatprotec.2008.08.015
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
to understanding the microstructural variation and process kinetics. Although a number of investigations have been carried out on TLPD bonding of different monolithic metals and alloys, reports on the TLPD bonding of AlMMCs are limited. Among different interlayer used in TLPD bonding of monolithic aluminium-based alloys, the use of copper interlayer has proved to be successful for joining conventional aluminium alloys, and bond strength comparable to that of the parent material has been reported (Dray, 1985). Again, the published literature on TLPD bonding of AlMMCs mainly dealt with the development of bonding conditions using different thickness of copper interlayer in order to achieve adequate bond strength. While studying TLPD bonding of SiC fiber reinforced AlMMC with 10-m thick copper interlayer at 550 ◦ C in air environment, it has been reported by Bushby and Scott (1993) that higher bonding pressure (20 MPa) was necessary in order to limit oxidation of copper and maximize the bonded area to 80%. On the other hand, bond strength of 92% of the parent material strength was achieved by Shirzadi and Wallach (1997) for joining AlMMC by TLPD process at 560 ◦ C, 0.1–0.2 MPa pressure, using a 7-m thick copper interlayer with 20 min bonding time in vacuum followed by isostatic pressing. In other investigation, using mixed powder interlayer (Al–Si–SiC–Ti), bond strength of 50 MPa was achieved by Huang et al. (2007) for joining 6063–SiCp composite by TLPD process in vacuum at 595 ◦ C, 0.003 MPa, with a bonding time of 90 min. Whereas in argon environment, bond strength of 68 MPa was achieved by one of the present authors (Pal, 2005) for joining extruded 606115 wt% SiCp using copper powder interlayer at 560 ◦ C, 2 MPa, with a bonding time of 20 min. In most of these studies of TLPD bonding using copper interlayer, bonding temperature was kept at 550–560 ◦ C, which is slightly above the eutectic temperature of Al–Cu system (548 ◦ C) and below the solidus temperature of AlMMC. However, different pressures were used for bonding performed under different conditions. TLPD bonding in air environment requires very high-pressure application (20 MPa) in order to achieve metal-to-metal contact at bond interface (Bushby and Scott, 1993). The high pressure causes excessive plastic deformation of AlMMC, which is not desired. Therefore, conventional TLPD bonding is performed at lower pressure (0.1–0.2 MPa) in vacuum or inert environment in order to achieve adequate bond strength without plastic deformation. Again, for low pressure conventional TLPD bonding in vacuum with lower bonding time (20 min), presence of void at bond interface has been identified by Shirzadi and Wallach (1997). These voids were responsible for lowering down the bond strength to a certain extent. The low pressure bonding with lower time of holding (20 min) was followed by isostatic pressing to achieve high bond strength. However, the low pressure TLPD bonding with higher bonding time (say, 6 h) has not been studied for AlMMC. Moreover, in all these investigations of TLPD bonding of AlMMC, no explicit correlation was made between bond microstructure and different stages of the process. Also, bonding time was kept low (maximum 2 h) without any correlation with the completion of isothermal solidification or homogenization of bond region, and no comparison of process kinetics was made with monolithic system. Present investigation aims at developing adequate bond strength for extruded 6061-15 wt% SiCp com-
3569
Table 1 – Chemical composition of 6061 alloy (wt%) Mg Si Cu Cr Al
1.0 0.6 0.3 0.2 Rest
posite, by TLPD process in argon environment, with regard to process mechanism and microstructural evaluation, for different bonding times up to 6 h.
2.
Materials and methods
2.1.
Material
As-received material was an extruded rod of AlMMC consisting of 6061 matrix alloy and 15 wt% (12.93 vol%) silicon carbide (SiC) particulate reinforcement of 23 m average size. The nominal composition of 6061 alloy (Anon., 1990) is given in Table 1. In addition it contains some iron (0.6 wt%) as an impurity which was confirmed by chemical analysis in optical emission spectrometer (UNISPEC: 4L/0096). The density of as-received AlMMC was also measured by water displacement method.
2.2.
Specimen preparation for bonding
The extruded rod was machined to produce discs of 15 mm diameter and 10 mm height. As a result the faying surfaces of discs became perpendicular to the extrusion direction. The faying surfaces of discs were polished to 1 m finish. Pure copper (99.97 wt%) foil of 50 m thickness was used as interlayer. The interlayer was punched out to a diameter of 15 mm for bonding. The interlayer and polished faying surfaces of discs were finally rinsed in acetone and dried by a hot air blast just before bonding.
2.3.
TLPD bonding
The interlayer was placed between the polished faying surfaces of the two AlMMC discs. This assembly was then set by an adhesive tape and inserted inside the diffusion bonding unit. The bonding was carried out in a programmable electric furnace keeping bond centerline horizontal. A thermocouple inserted into the drilled hole in one of each pair of discs was used to monitor bonding temperature. The argon gas (99.99% Ar, 3–5 ppm O2 , 3 ppm H2 O, 2 ppm H2 , 1 ppm CO2 , 1 ppm CO) was flown into the bonding chamber at a rate 5 l/min to maintain inert atmosphere. The bonding temperature was kept at 560 ◦ C which is above the eutectic temperature (548 ◦ C) of Al–Cu system (Anon., 1992) and below the solidus temperature (582 ◦ C) of 6061 matrix alloy (Anon., 1990). The specimens were heated to the bonding temperature (560 ◦ C) at a rate of 6 ◦ C/min, held at that temperature for five different lengths of time (bonding time), viz. 20 min, 1, 2, 3 and 6 h, and cooled down to 540 ◦ C at a rate of 5 ◦ C/min inside the furnace. Then the specimens were taken out of the furnace and cooled in still air to the room temperature. Two different pressures, 0.1 and 0.2 MPa, were applied for bonding.
3570
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
2.4.
Optical metallography
Bonded cylindrical samples of 15 mm diameter were sectioned perpendicular to the bonding plane using a low speed diamond cutter. The section was polished to 1 m finish and etched with Keller’s reagent. Both qualitative and quantitative study of microstructure around bond interface was carried out for these metallographic specimens using optical microscope with digital photomicrography (ZEISS, Imager.A1m). The interface width was measured at ‘periphery’ and ‘central zone’ of the bond interface. The bond centerline had 15 mm length. The two edges of bond centerline, with each edge of 3.5 mm length, were considered as ‘periphery’. The remaining part at the middle, of 8 mm length, was considered as ‘central zone’.
2.5.
Fig. 1 – Specimen loaded in jig: the arrangement for shear strength (bond strength) determination.
Mechanical testing
Bonded cylindrical samples were machined to 10 mm diameter to eliminate edge effects. These specimens of approximately 20 mm length and 10 mm diameter and asreceived AlMMC of similar dimension were loaded in a specially prepared jig, which is schematically shown in Fig. 1. The grips of the jig were pulled in tension in a
Fig. 2 – Microstructure of as-received 6061-SiCp composite—(a) optical image: bands of SiC along extrusion direction; (b) SEM secondary electron image: SiC bands and porosity and (c) SEM back-scattered electron image: presence of FeAlX.
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
3571
Fig. 3 – Typical optical microstructures of bonded specimens at central zone of bond interface at 0.2 MPa pressure: (a) 20 min; (b) 1 h; (c) 2 h; (d) 3 h and (e) 6 h.
100-kN capacity universal testing machine (INSTRON-8801) at a cross-head speed of 0.5 mm/min in position control mode such that the specimen experienced pure shear stress across the bond interface. The maximum load was divided by the bond area in order to calculate shear strength. For each bonding condition three specimens were tested and the average value is considered as shear strength (bond strength).
2.6.
SEM and EDS analysis
As-polished section of the metallographic specimens was examined around bond interface under scanning electron microscope (JEOL, JSM-5800). Microstructures were studied with both back-scattered electron image mode and secondary electron image mode. Accordingly different phases were identified by EDS spot analysis. Also the progress
3572
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
Fig. 4 – SEM back-scattered electron image of bonded specimen (20 min, 0.2 MPa).
of diffusion was studied by line scan. The line scan for concentration variation of copper was carried out along a line of 600 m length lying perpendicular to the bond interface keeping bond centerline approximately at the middle. The fractured surfaces of shear-tested specimens were also studied under the scanning electron microscope with both back-scattered electron image mode and secondary electron image mode. Different phases present on fractured surface were identified with EDS spot analysis.
3.
Results and discussion
3.1.
As-received composite
The microstructure of as-received AlMMC, as shown in Fig. 2, clearly reveals bands of SiC-rich areas along the extrusion direction as well as some porosity near the particle cluster (band). The presence of porosity is also indicated in the measured density of the composite (2.72 gm cm−3 ) that shows lower value than the theoretical density of com-
Fig. 5 – Schematic representation of bond region during isothermal solidification.
posite (2.77 gm cm−3 ) based on the calculation from the reported density of 6061 matrix alloy (2.70 gm cm−3 ) and SiC (3.21 gm cm−3 ) (Anon., 1990, 1991). SEM back-scattered electron image (Fig. 2(c)) and EDS spot analysis reveal the presence of intermetallic phases of iron and aluminium (FeAlX in general) in as-received AlMMC. Both FeAl3 (containing about 61 wt% Al) and Fe2 Al5 (containing 56 wt% Al) are identified. Chemical analysis by optical emission spectrometer indicates the presence of about 0.6 wt% iron impurity in the composite. The presence of iron impurity (about 0.7 wt% Fe) in 6061 alloy was reported (Rohatgi et al., 1986). Also, the presence of intermetallic phases of Fe and Al in AlMMC containing iron impurity was observed (Bushby and Scott, 1993). This is due to very little solubility of iron
Table 2 – Interface width at central zone and periphery Bonding pressure (MPa)
Bonding time
Interface width (m) At central zone
At periphery
0.1
20 min 1h 2h 3h 6h
230 107 93 Negligible Negligible
185 177 183 163 143
0.2
20 min 1h 2h 3h 6h
223 86 73 Negligible Negligible
205 173 171 196 134
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
3573
Fig. 6 – Line scan result representing concentration variation of Cu around bond interface for 6 h of holding at 0.1 MPa pressure (X-axis length = 600 m).
in aluminium as indicated in Fe–Al phase diagram (Anon., 1992).
3.2.
Microstructural evaluation and mechanism
Based on the study of monolithic system, the TLPD bonding process was described as consisting of four stages namely, (I) dissolution of interlayer, (II) homogenization of liquid (widening of liquid to its maximum width), (III) isothermal solidification, and (IV) homogenization of bond region (Duvall et al., 1974; Nakao et al., 1989; Natsume et al., 2003; Tuah-Poku et al., 1988). The first two stages together take few minutes. However, duration of last two stages may extend to several hours. In present study, heating from eutectic temperature (548 ◦ C) to the bonding temperature (560 ◦ C) at a rate 6 ◦ C/min takes 2 min. It is possible that the first two stages (dissolution of interlayer and widening of liquid) are completed during heating to the bonding temperature. Optical microstructures of TLPD bonded specimens in etched condition representing the bond interface at central zone is shown in Fig. 3. SEM backscattered electron image of the bond interface is presented in Fig. 4. The specimens with lowest bonding time (20 min) exhibit isothermally solidified grains (mainly of primary ˛) adjacent to bond interface (Fig. 3(a)). Furthermore, presence of CuAl2 phase (containing about 54 wt% Cu) is observed adjacent to the bond interface (Fig. 4). It is to be noted that as-received AlMMC (unbonded) does not contain any CuAl2 intermetallic phase. In bonded composites CuAl2 phase is likely to precipitate out in isothermally solidified zone during cooling from bonding temperature as solubility of Cu in primary ˛ decreases with decreasing temperature following solvus curve. Therefore, liquid widening process completes and isothermal solidification starts before 20 min of holding.
Fig. 7 – Segregation of FeAlX at bond interface (1 h, 0.2 MPa).
The presence of 15 wt% SiC particle in 6061 aluminium matrix is expected to have significant effect on TLPD bonding process as compared to pure monolithic system. Microstructural study reveals the segregation of SiC particles at bond
Table 3 – Result of shear test Bonding condition As-received composite (no bonding) 20 min, 0.1 MPa 1 h, 0.1 MPa 2 h, 0.1 MPa 3 h, 0.1 MPa 6 h, 0.1 MPa 20 min, 0.2 MPa 1 h, 0.2 MPa 2 h, 0.2 MPa 3 h, 0.2 MPa 6 h, 0.2 MPa
Shear strength (MPa) 105 55 50 34 75 76 69 54 49 77 95
Fig. 8 – Fractured surface of as-received composite: (a) secondary electron image and (b) back-scattered electron image.
3574
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
Fig. 9 – Secondary electron images of the fractured surfaces of bonded composites: (a) 20 min, 0.1 MPa; (b) 20 min, 0.2 MPa; (c) 1 h, 0.1 MPa; (d) 1 h, 0.2 MPa; (e) 2 h, 0.1 MPa and (f) 2 h, 0.2 MPa.
centerline with isothermally solidified zones on both sides. During bonding the bond centerline was kept horizontal. Since no preferential segregation occurs towards the lower AlMMC disc, this is not the case of gravity segregation. According to published literature on general solidification characteristic of SiC reinforced AlMMC, the primary ˛ is very efficient at rejecting SiC, and pushing the particles ahead of the solid/liquid
interface (Gallerneault et al., 1991). In this regard a critical velocity of solid/liquid interface has been reported, below which the SiC particles are pushed by the moving interface and above which they are engulfed (Rohatgi et al., 1986). Since the first two stages of TLPD bonding (dissolution of interlayer and widening of liquid) are very fast, the velocity of solid/liquid interface is also very high. As a result, SiC par-
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
3575
Fig. 10 – Back-scattered electron images of the fractured surfaces of bonded composites: (a) 1 h, 0.1 MPa; (b) 1 h, 0.2 MPa; (c) 3 h, 0.1 MPa; (d) 3 h, 0.2 MPa; (e) 6 h, 0.1 MPa and (f) 6 h, 0.2 MPa.
ticles are not pushed by the solid/liquid interface away from the bond centerline during widening of liquid. The next stage, isothermal solidification, is slow due to solid-state diffusion controlling the process and takes several hours for completion. During isothermal solidification, due to low velocity of solid/liquid interface, most of the SiC particles are pushed by the moving solid/liquid interface, as evident in the present work. As a result, particles segregate at bond centerline along with liquid phase and the residual liquid gets solidified during
cooling. This aggregate of residual liquid and segregated SiC particles may be called as ‘Segregation Zone’ and the width of this segregation zone may be termed as ‘interface width’. This is schematically shown in Fig. 5. The measured interface width at central zone and at periphery of bond interface is presented in Table 2. At 20 min bonding time, the interface width at central zone is somewhat similar (slightly greater) to that at periphery. However, with increasing bonding time interface width at central zone decreases continuously and always
3576
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
Fig. 11 – Secondary electron images of the fractured surfaces of bonded composites: (a) 3 h, 0.1 MPa; (b) 3 h, 0.2 MPa; (c) 6 h, 0.1 MPa and (d) 6 h, 0.2 MPa.
remains lower than that at periphery. This phenomenon indicates that during TLPD bonding under pressure (0.1 and 0.2 MPa) liquid–particle aggregate at interface moves towards periphery and flows out. The interface width at periphery exhibits decrease and occasional increase depending on the flow of liquid–particle aggregate from center to periphery and from periphery to outside. Also, with increasing bonding time reduction of interface width at central zone is more at higher pressure (0.2 MPa) than at lower pressure (0.1 MPa). This again indicates more mass flow towards periphery under higher pressure. It is important to note that specimens with 3 and 6 h bonding time (for both the pressures 0.1 and 0.2 MPa) exhibit negligible interface width at central zone with least segregation of SiC particles as shown in Fig. 3(d) and (e). Furthermore, the bond interface is hardly discernible and grain continuity exists across the interface indicating the completion of isothermal solidification in 3 h of bonding time. The EDS spot analysis of matrix at bond interface, for the specimen with 3 h bonding time and 0.2 MPa pressure, exhibits the presence of 2.11 wt% Cu. According to Al–Cu phase diagram, at bonding temperature (560 ◦ C) the maximum solubility of copper in primary ˛ is 4.35 wt% (Anon., 1992). Therefore, the copper content of 2.11 wt% indicates the presence of primary ˛ grains. This further confirms the completion of isothermal solidification.
Natsume et al. (2003) investigated the mechanism of TLPD bonding for joining pure Al by 50 m Cu interlayer at 570 ◦ C up to a maximum bonding time of 1 h. According to their experimental data, melting of interlayer took about 15 sec. The liquid widened to its maximum width of 460 m for 60 s of holding. It is to be noted that the specimen was not held at bonding temperature till the completion of isothermal solidification. Up to 1 h of maximum holding liquid width decreased to 406 m through isothermal solidification. Whereas, in the present investigation interface width at central zone has reduced to 107 m for 0.1 MPa pressure and 86 m for 0.2 MPa pressure after 1 h of holding. This indicates much faster kinetics of isothermal solidification in case of AlMMC as compared to the pure monolithic system. In the present study, AlMMC contains substantial amount (15 wt%) of SiC particles. The presence of SiC particles in the metallic matrix leads to the formation of defect-rich interfacial region of high dislocation density, mainly due to the difference in coefficient of thermal expansion between metallic matrix and SiC particles (Papazian, 1988). The coefficient of thermal expansion of 6061 matrix alloy and SiC particle are 23.6 × 10−6 K−1 and 5.5 × 10−6 K−1 , respectively (Anon., 1990, 1991). Thus, it is likely that the particle/matrix interface of composite is associated with high dislocation density. In addition, as discussed earlier, the com-
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
3577
Fig. 12 – Secondary electron images of metallographic specimens near the edge of periphery of bond interface: (a) 3 h, 0.1 MPa; (b) 3 h, 0.2 MPa; (c) 6 h, 0.1 MPa and (d) 6 h, 0.2 MPa.
posite contains porosity. These defects provide short circuit paths for diffusion. Therefore, at the bonding temperature of 560 ◦ C (which is nearer to the solidus temperature, 582 ◦ C, of 6061 alloy), along with lattice diffusion, short circuit diffusion is also expected to be operative. It is reported that short circuit diffusivities are larger than lattice diffusivities by a factor of 103 near the melting point (Gjostein, 1972). As a result, diffusion becomes faster. Moreover, due to application of pressure some part of liquid flows out of periphery reducing the amount of liquid to be solidified isothermally. The EDS line scan for Cu concentration variation across the bond interface indicates diffusion of Cu away from the interface. An occasional rise in copper concentration is due to the presence of copper enriched phase CuAl2 . However, on a gross scale, distribution of copper in the matrix for 6 h of bonding is found to be uniform across the bond centerline (Fig. 6). This indicates homogenization of bond region. The CuAl2 phase is found to be present at bond interface (segregation zone) as well as in isothermally solidified zone. At the bond interface CuAl2 forms through eutectic solidification of residual liquid during cooling. At the bonding temperature this residual liquid remains intermixed with segregated SiC particles. SEM backscattered electron image (Fig. 4) clearly reveals that during cooling CuAl2 preferentially nucleates on SiC par-
ticles at bond interface. On the other hand, at the bonding temperature, isothermally solidified zone contains the grains of primary ˛. During cooling from bonding temperature, as the solubility of Cu decreases, CuAl2 precipitates out of primary ˛. In bonded composite the particles of intermetallic phases of iron impurity (FeAlX ) is found to be segregated around bond centerline (Fig. 7). It is likely that during isothermal solidification these particles are pushed by solid/liquid interface and thereby segregate at bond interface (segregation zone) along with SiC particles.
3.3.
Bond strength and fractured surface
The result of shear test is given in Table 3. In general, the as-received composite exhibits higher shear strength as compared to bonded specimens. However, the specimen bonded at 0.2 MPa pressure with 6 h holding time shows the highest shear strength (bond strength) of 95 MPa which about 90% of the as-received composite strength (105 MPa). The fractured surface of shear tested as-received composite exhibits dimples and tear ridges (Fig. 8(a)), characteristic of a typical ductile fracture of Al/SiCp composite (Ge and Schmauder, 1995). The presence of FeAlX is revealed in back scattered electron image (Fig. 8(b)). However, fractured surface does not contain CuAl2
3578
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
Fig. 13 – Secondary electron images of metallographic specimens at the central zone of bond Interface (higher magnification): (a) 3 h, 0.1 MPa; (b) 6 h, 0.1 MPa; (c) 3 h, 0.2 MPa and (d) 6 h, 0.2 MPa.
phase since this phase is absent in as-received composite. In bonded composite, failure is found to occur through the segregation zone. The specimens with 20 min, 1 h and 2 h bonding time, at both the pressures (0.1 and 0.2 MPa) show the presence of oxide on fractured surface as indicated in secondary electron images (Fig. 9). Decohesion between oxide and metallic matrix is clearly evident. The presence of oxide indicates that during bonding the residual oxygen present in the system oxidizes the transient liquid phase. EDS analysis reveals that these oxides are the oxides of Al, Mg and Si. Apart from oxides, presence of brittle phases such as FeAlX and CuAl2 are also revealed in back scattered electron images (Fig. 10). Furthermore, in these bonding conditions, segregation of SiC particles at bond interface is observed in metallographic study (Fig. 3(a–c)). Accordingly, the fractured surface (Fig. 9) appears to be flat (devoid of dimples), characteristic of brittle fracture. In general, these specimens exhibit poor bond strength. Out of these, the specimen bonded at higher pressure (0.2 MPa) with lowest bonding time (20 min) possesses relatively higher bond strength (69 MPa). Similar bond strength (68 MPa) was achieved by one of the present authors (Pal, 2005) for joining same composite using copper powder interlayer of 0.8 mm thickness at 560 ◦ C with 20 min of holding under 2 MPa pressure. Fractured surface of this specimen contains relatively lesser extent of
oxide (Fig. 9(b)). Oxidation is expected to be less for lower time of holding at higher pressure. It has been reported that higher pressure lowers the extent of oxidation due to more liquid expulsion (Bushby and Scott, 1993). However, specimens with 3 h and 6 h bonding time, for both the pressures, show least presence of oxide on fractured surface (Fig. 11). In these bonding conditions, segregation of SiC particles at bond interface is negligible as revealed in metallographic study (Fig. 3(d and e)). Therefore, the bond strength achieved is relatively higher. However, presence of brittle phases (CuAl2 , FeAlX ) is still observed (Fig. 10). In general, these specimens exhibit presence of dimples along with cleavage facets indicating mixed mode of failure. While investigating metallographic specimens with 3 h and 6 h bonding time (prepared by sectioning bonded specimens of 15 mm diameter along axial length) in SEM secondary electron image mode, the presence of oxide near the edge of periphery is observed (Fig. 12). Therefore, it appears that on completion of isothermal solidification (3 h of holding) the solid/liquid interfaces from both sides approach each other and merge together and eventually the residual liquid along with oxide is rejected at the edge of periphery. Once the isothermal solidification completes, the entire bond region becomes solid and no further oxide phase can form at the bond interface through oxidation
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
of liquid for longer time of holding (3 h or more). Moreover, bonded composites of 15 mm diameter were machined to 10 mm diameter in order to prepare specimens for shear test. As a result, the edge of periphery was machined out. Therefore, for 3 h and 6 h bonding time, the shear tested specimens of 10 mm diameter do not exhibit any significant presence of oxides on fractured surface. Before completion of isothermal solidification, the liquid phase has maximum exposure to environment for 2 h of holding that results in severe oxidation. Thus specimen bonded with 2 h of holding at lower pressure (0.1 MPa) exhibit the lowest bond strength (34 MPa). In general, before completion of isothermal solidification, higher the bonding time (more exposure of liquid) and lower the pressure, greater is the extent of oxidation of liquid and lower is the bond strength. It is also interesting to note that at 0.1 MPa pressure, bonded composites with 3 h and 6 h bonding time show similar bond strength (75 MPa and 76 MPa, respectively). The SEM study in secondary electron image mode on metallographic specimens at higher magnification reveals the presence of voids at bond interface for both the bonded composites (Fig. 13(a and b)). The voids are likely to be generated due to the volumetric shrinkage of metal during isothermal solidification. The density of liquid aluminium (2400 kg m−3 ) is lower than that of solid aluminium (2700 kg m−3 ) (Natsume et al., 2003). On the other hand, at 0.2 MPa pressure, the composite bonded with 6 h of holding shows much higher bond strength (95 MPa) than that with 3 h bonding time (77 MPa). The SEM study shows that the bond interface of the former (0.2 MPa, 6 h) is almost free from void (Fig. 13(d)), while the later (0.2 MPa, 3 h) contains void at bond interface (Fig. 13(c)). Therefore, under higher pressure (0.2 MPa), on completion of isothermal solidification, solid state diffusion eliminates void at bond interface, thereby improving bond strength. The fractured surface (Fig. 11(d)) of this composite bonded at 0.2 MPa pressure with 6 h of holding appears somewhat similar to that of as-received composite (Fig. 8(a)) and bond strength achieved is also closer (90% of the shear strength of as-received composite). It is important to note that, while joining AlMMCs by TLPD process, Shirzadi and Wallach (1997) achieved the highest bond strength, 92% of the parent material strength, using 7-m thick copper interlayer at 560 ◦ C, 0.1–0.2 MPa, with 20 min of holding in vacuum followed by isostatic pressing (costlier process). Therefore, in present study, bond strength very close to the highest achieved bond strength is obtained in argon environment using thicker interlayer (50 m) with higher bonding time (6 h).
4.
Conclusion
(i) TLPD bonding process of 6061-SiCp composite occurs much faster than that of pure aluminium. Isothermal solidification takes about 3 h and bond region gets homogenized in 6 h. (ii) In composite, the presence of defect-rich particle/matrix interface and porosity makes diffusion process faster. In addition, liquid expulsion under applied pressure reduces the amount of residual liquid to be solidified. As a result the duration of isothermal solidification is reduced.
3579
(iii) The CuAl2 phase is found to be present at bond interface (segregation zone) as well as in the isothermally solidified zone. The intermetallic phases of iron and aluminium (FeAlX ) segregate mainly at the bond interface. (iv) While bonding in commercial argon environment, the transient liquid gets oxidized. In general, the presence of oxide and other brittle phases (CuAl2 , FeAlX ) lowers down the bond strength. (v) On completion of isothermal solidification oxides get rejected to the periphery along with residual liquid, thereby reducing the amount of oxides at bond interface. Furthermore, longer time holding (6 h) under higher pressure (0.2 MPa) removes the void at bond interface through solid-state diffusion. As a result, bond strength is improved. (vi) TLPD bonding at 560 ◦ C under 0.2 MPa pressure with 6 h of holding in commercial argon environment produces a bond strength of 90% of the parent AlMMC strength, which is very close to the bond strength achieved (92% of the parent AlMMC strength) for TLPD process carried out at 560 ◦ C under 0.1–0.2 MPa pressure with 20 min of holding in vacuum followed by isostatic pressing.
references
Anon., 1990. Properties and Selection: Non-ferrous Alloys and Special-Purpose Materials Metals Handbook, vol. 2., 10th ed. ASM International, The Materials Information Society, pp. 102–103. Anon., 1991. Ceramics and Glasses. Engineering Materials Handbook, vol. 4. ASM international, The Materials Information Society, p. 30. Anon., 1992. Alloy Phase Diagrams, ASM Handbook, vol. 3. ASM International, The Materials Information Society, 2.44-3, p. 10. Bushby, R.S., Scott, V.D., 1993. Liquid Phase bonding of aluminium and aluminium/Nicalon composite using copper interlayers. Mater. Sci. Technol. 9, 417–423. Bushby, R.S., Scott, V.D., 1995. Liquid phase bonding of aluminium and aluminium/Nicalon composite using interlayers of Cu–Ag alloy. Mater. Sci. Technol. 11, 643–649. Devletian, J.H., 1987. SiC/Al metal matrix composite welding by a capacitor discharge process. Weld. J. 66, 33–39. Dray, A.E., 1985. Diffusion bonding of aluminium. PhD Thesis. University of Cambridge. Duvall, D.S., Owczarski, W.A., Paulonis, D.F., 1974. TLP bonding: a new method for joining heat resistant alloys. Weld. J. 53, 203–214. Ellis, M.B.D., 1997. Joining of metal matrix composites: a review. TWI J. 6, 69–128. Field, D.J., 1989. Aluminium Alloys-Contemporary Research and Applications. London Academic Press, pp. 523–537. Gallerneault, M., Kaya, M., Smith, R.W., Dellamore, G.W., 1991. Ultimate sediment concentration in directionally solidified Al–Si/SiC particulate metal matrix composites. In: Proceedings Conference on Extraction, Refining and Fabrication of Light Metals. Pergamon Press Inc., New York, USA, pp. 69–81. Ge, X., Schmauder, S., 1995. Micromechanism of fracture in Al/SiC composites. J. Mater. Sci. 30, 173–178. Gittos, M.F., Threadgill, P.L., 1991. Preliminary studies of joining particulate reinforced aluminium alloy metal matrix composites. In: Proceedings of the International Symposium on Metal Matrix Composites III. Exploiting the Investment, Inst. of Metals, December, pp. 10–11.
3580
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 9 ( 2 0 0 9 ) 3568–3580
GJOSTEIN, N.A., 1972. Short circuit diffusion. Diffusion, Papers Presented at a Seminar of the American Society for Metals, American Society for Metals, Metals Park, OH, October 14–15, p. 241. Hall, W.I., Manrique, F., 1995. Surface treatment of carbon fiber for aluminium alloy matrix composites. Scripta Met. 33, 2037–2043. Huang, J., Dong, Y., Wan, Y., Zhao, X., Zhang, H., 2007. Investigation on reactive diffusion bonding of SiCp/6063 MMC by using mixed powders as interlayers. J. Mater. Process. Technol. 190, 312–316. Luhman, T.S., Williams, R.L., Das, K.B., 1983. Development of joint and joining techniques for metal matrix composites. In: Fourth Quarterly Progress Report, Army Materials and Mechanics Research Center, pp. 1–58. Nakao, Y., Nishimoto, K., Shinozaki, K., Kang, C., 1989. Theoretical research on transient liquid insert metal diffusion bonding of nickel base alloys. Trans. Jpn. Weld. Soc. 20, 60–65.
Natsume, Y., Ohsasa, K., Tayu, Y., Momono, T., Narita, T., 2003. Numerical modeling of transient liquid-phase diffusion bonding process of al using cu filler metal. ISIJ Int. 43, 1976–1982. Pal, T.K., 2005. Joining of aluminium metal matrix composites. Mater. Manuf. Process. 20, 717–726. Papazian, J.M., 1988. Effects of SiC Whiskers and particles on precipitation in aluminium matrix composites. Metall. Trans. A 19A, 2945–2953. Rohatgi, P.K., Asthana, Das, R.S., 1986. Solidification, structures, and properties of cast metal–ceramic particle composites. Int. Met. Rev. 31, 115–136. Shirzadi, A.A., Wallach, E.R., 1997. New approaches for transient liquid phase diffusion bonding of aluminium based metal matrix composites. Mater. Sci. Technol. 13, 135–142. Tuah-Poku, I., Dollar, M., Massalski, T.B., 1988. A study of the Transient liquid phase bonding process applied to a Ag/Cu/Ag sandwich joint. Metall. Trans. A 19A, 675–686.