纳米电子封装19

Chapter 19 Properties of 63Sn-37Pb and Sn-3.8Ag-0.7Cu Solders Reinforced With Single-Wall Carbon Nanotubes K. Mohan Kumar(*ü ), V. Kripesh, and Andrew A.O. Tay 19.1 Introduction As integrated circuit (IC) technology continues to advance, there will be increasing demands on I/O counts and power requirements, leading to decreasing solder pitch and increasing current density for solder balls in high-density wafer-level packages [1] . As the electronics industry continues to push for miniaturization, reliability becomes a vital issue. The demand for more and smaller solder bumps, while increasing the current, has also resulted in a significant increase in current density [2] , which can cause the failure of solder interconnects due to electromigration [3] . Solders are extensively used in IC technology as mechanical and electrical inter- connects because of their ease of processing and lower cost. However, because of their relatively low melting temperatures, creep is a major concern. When elec- tronic devices are switched on and off, the electronic packages experience cyclic changes in temperature. Because of differences between the packages and the sub- strate, cyclic changes in thermomechanical stresses are induced in the package-to- board solder joints. Such cyclic stresses in the solder joints eventually lead to failure of the solder joints through thermomechanical fatigue [4 , 5] . With the relentless trend toward very fine pitch IC packages, the cyclic stresses experienced by flip chip-to-board interconnects are increasing greatly, resulting in a drastic drop in fatigue life of solder joints. One way of overcoming this problem is to use new materials, which can provide enhanced mechanical, electrical, and thermal properties. Composite solders can offer improved properties [6] . Although a few researchers have investigated the influence of nanoparticles and nanotubes on the properties of solder [7 – 9] , these investigators were mainly focused on the mechanical properties of the solders. In this study, the influence of nanotube addi- tion on microstructural, mechanical, electrical, wetting, and thermal properties has been investigated. In addition to this, efforts have been made to evaluate the joint strength and creep strength of the composite solder joints. K.M. Kumar Nano/Microsystems Integration Laboratory, Department of Mechanical Engineering, National University of Singapore, Singapore J.E. Morris (ed.) Nanopackaging: Nanotechnologies and Electronics Packaging, 415 DOI: 10.1007/978-0-387-47326-0_19, © Springer Science + Business Media, LLC 2008 Morris_Ch19.indd 415Morris_Ch19.indd 415 9/29/2008 8:09:25 PM9/29/2008 8:09:25 PM 416 K.M. Kumar et al. Owing to their fascinating physical properties and unique structures, carbon nano- tubes (CNTs) are receiving steadily increasing attention since their discovery [10] . Intense interest from researchers has been generated in utilizing these unique struc- tures and outstanding properties, for example, in hydrogen storage, supercapacitors, biosensors, electromechanical actuators, and nanoprobes for high-resolution imaging and so on [11 , 12] . In recent years, there has been a steadily increasing interest in the development of CNT-reinforced composites due to their remarkable mechanical, electrical, and thermal properties [13 – 16] . Depending on their length, diameter, chirality, and orientations, CNTs show almost five times the elastic modulus (1 TPa) and nearly 100 times the tensile strength (150 GPa) of high-strength steels [17] . The motive is to transfer the exceptional mechanical and physical properties of CNTs to the bulk engineering materials. Polymers, ceramics, and metals are favorable as matrix materials. CNT-reinforced polymer-based composites were widely synthe- sized by surfactant-assisted processing, repeated stirring, solution evaporation with high-energy sonication, and interfacial covalent functionalization [18 – 20] . Much of the research in nanotube-based composites has been on polymer or ceramic matrix materials and less on metal–matrix composites [21 – 23] . This is mainly due to the fact that uniform dispersion of CNTs in a metal matrix is quite difficult. Nai et al. [9] demonstrated that the dispersion and homogenous mixing between MWCNTs and a lead-free solder matrix could be obtained by mixing nanosized matrix powders with CNTs. They showed that the powder metallurgy process was a very promising technique for full densification of CNT/lead-free solder nanocom- posites, which showed remarkable enhancement of yield strength compared with that of unreinforced lead-free solders. The current work provides an insight into the usage of SWCNTs as a reinforcing material for the enhancement of the properties of the solder material to be used in wafer-level chip-scale packages (WLCSP). The aim of this work is to fabricate and characterize CNT-reinforced nanocomposite solders and show their improved physical, thermal, electrical, mechanical, and wetting properties compared with the original Sn–Pb and Sn–Ag–Cu solders. 19.2 Experimental Aspects 19.2.1 Materials The starting materials used in this study were Sn–Pb and Sn–Ag–Cu solder pow- ders of type 7 (2–11 µm). The SWCNTs employed in this study were prepared using the chemical vapor deposition (CVD) technique and typically have an average diameter of 1.2 nm and lengths between 5 and 10 µm. 19.2.2 Preparation of Composite Solders The solder powder and SWCNTs were weighed to the approximate weight percent ratio. Different compositions were prepared with varying SWCNT content ranging Morris_Ch19.indd 416Morris_Ch19.indd 416 9/29/2008 8:09:25 PM9/29/2008 8:09:25 PM 19 Properties of 63Sn-37Pb and Sn-3.8Ag-0.7Cu Solders Reinforced 417 from 0.01 to 1 wt%. The preweighed SWCNTs and solder powders were blended homo- geneously using a V-cone blender operated at a speed of 50 rpm. The homogenously blended composite solder powders were consolidated by uniaxial cold pressing with a pressure of 110 bar in the case of Sn–Pb composite solders while Sn–Ag–Cu composite solders were compacted at a pressure of 120 bar. The consolidated green composite solder compacts of diameter 35 mm were sintered at 150°C for Sn–Pb composite solders and at 180°C for Sn–Ag–Cu composite solders. Sn–Ag–Cu composite solders were sintered at 180°C to approach a reasonable rate of solid-state sintering. The sintered compacts were finally extruded at room temperature with an extrusion ratio of 20:1. 19.2.3 Scanning Electron Microscopy Samples were cut from the extruded solder bars with a diamond saw, and mechanically polished with diamond pastes after cutting, and finishing with 0.02-µm grade. Micro- structural observations were performed by scanning electron microscopy (SEM) using a Hitachi FE-SEM 4100 operated at 10 kV. The elemental analysis of the phases was carried out using energy-dispersive X-ray spectroscopy (EDX) equipped with FE-SEM. 19.2.4 Thermomechanical Analysis (TMA) The linear thermal expansion coefficient of composite solders was measured using a Perkin-Elmer TMA-7 thermal mechanical analyzer operated in expansion mode. Cylindrical samples of diameter 8 mm were employed. TMA data were obtained in the heating range of 25–125°C in the case of Sn–Pb composite solders, while a 25–150°C heating range was employed for the Sn–Ag–Cu composite solders at a rate of 5°C/min. All TMA experiments were performed with a small loading force of 5 g to avoid defor- mation of the samples during testing. The CTEs of the composite solder specimens were obtained from the slope of the curve over a linear temperature range. 19.2.5 Differential Scanning Calorimetry (DSC) The melting behaviors of the composite solder specimens were examined by a Perkin- Elmer DSC-7 system. DSC experiments were carried out at a heating rate of 10°C/min from 25 to 250°C. The heat flow as a function of temperature was recorded and ana- lyzed. The entire scanning was carried out under an inert nitrogen atmosphere. 19.2.6 Electrical Properties Electrical conductivity was measured on strips having dimensions of 50 mm × 10 mm cut from rolled composite solder preforms with a thickness of ∼0.13 mm using a four-point probe technique. Morris_Ch19.indd 417Morris_Ch19.indd 417 9/29/2008 8:09:25 PM9/29/2008 8:09:25 PM 418 K.M. Kumar et al. 19.2.7 Wettability Solder alloys were cold rolled to preforms of thickness 1 and 0.13 mm for the joint tensile testing, wetting, and creep-rupture analyses. The solder preforms were remelted four times to get a uniform structure and composition. Approximately 0.2 g of the remelted solder preforms were weighed using an electronic balance. The weighed solder preforms were cleaned with acetone in an ultrasonic bath. The sub- strate used was a thin copper plate of 99.9% purity and dimensions of 25 mm × 25 mm × 0.1 mm. These small substrates were polished sequentially with silicon car- bide sandpaper of up to 800 abrasive number, and then cleaned ultrasonically in acetone for 10 min to achieve an ultraclean substrate for wetting experiments. The measurement of contact angle was performed using the following tech- nique. First wetting was carried out on a hot plate. Rosin mildly activated (RMA) flux was applied on a copper substrate. Some flux was then applied on the surface of the preweighed solder preform before placing it on the copper substrate. In preparation for the reflow, the substrate containing the solder and the flux was first preheated to 100°C, and then to the reflow temperature of 240°C. After the time of reflow, the specimen was quickly removed, allowed to solidify, and later quenched to room temperature. The solder after reflow on the copper substrate was cleaned with alcohol for 10 min to remove the flux residues. After each test, the solder drop was cut perpendicular to the interface, mounted in resin, and polished to examine the morphology and contact angle of solder on copper substrate. Then the photo- graph of the specimen was taken and analyzed with the help of commercially avail- able software for measuring spreading area. 19.2.8 Microhardness Testing The sintered samples were polished to a mirror finish prior to the microhardness indentation tests. Microhardness of the composite solder specimens was measured using a Digital Micro-Hardness Tester with a Vicker’s indenter. The samples were indented with a load of 10 g, and an average of seven indentations was made at dif- ferent locations of the composite solder specimens for further analysis. 19.2.9 Tensile Testing The samples for tensile testing were machined from the extruded bars. Dog-bone- shaped specimens of gauge length 25 mm and diameter 5 mm were prepared. Tensile experiments were carried out at room temperature on the specimens using an Instron 5569 tensile tester at a constant cross-head displacement of 1 mm/min. Five samples of each composite solder were tested. All samples were tested to failure. Morris_Ch19.indd 418Morris_Ch19.indd 418 9/29/2008 8:09:26 PM9/29/2008 8:09:26 PM 19 Properties of 63Sn-37Pb and Sn-3.8Ag-0.7Cu Solders Reinforced 419 19.2.10 Tensile Strength of Solder Joint Cu samples of length 45 mm were cut from a 99.9% pure, half-hardened Cu bar with rectangular cross section (10 mm × 1 mm). These were etched in 50% sulfuric acid to get rid of the surface oxide layer. The mating surfaces were fluxed immedi- ately with commercial RMA flux and the rest of the surfaces were coated with sol- der resist to prevent them from being wetted by molten solder. Solder alloys were rolled into thin sheets of thickness 1 mm and sliced into pieces that approximately covered the mating surface area of the Cu samples. Then, the sliced solder pieces were placed between the mating surfaces of two Cu samples in an aluminum mold and were heated in a furnace to a temperature 50°C above the liquidus of the solder. After holding in the molten state for 2 min, the samples were gently soldered with the help of a screw-driven mold to maintain a joint thickness of 500 µm by adjust- ing the screws placed at each end of the mold to obtain a good joint, and were cooled in the furnace. It was found that the tensile testing specimens thus prepared resulted in joints with solder of thickness between 300 and 400 µm. 19.2.11. Creep Rupture Analysis The creep rupture tests were conducted using composite solder lap joints between two dog-bone-shaped copper pieces, which were fabricated as follows. Two 99.9% pure, 0.1-mm thin copper sheets were first wire-cut into the shape of a dog bone. The composite solder alloys were cold rolled to obtain performs of thickness 0.13 mm and cut into square specimens of dimensions 1 mm × 1 mm. The dog-bone- shaped copper substrates were cleaned with dilute sulfuric acid and rinsed with acetone. The narrow ends of the copper substrates were coated with solder resist to obtain a cross-sectional area of 1 mm 2 . Then, RMA flux was applied to each narrow end of the substrate and the composite solder preform was sandwiched between the two copper substrates. Reflow soldering was performed in a programmable oven. The creep-rupture life tests were performed at room temperature with a dead load stress of ∼10.4 MPa. 19.3 Results and Discussion 19.3.1. Microstructural Observation The SEM and TEM microstructures of the as-received SWCNTs used in the present study are shown in Fig. 19.1 . The FE-SEM microstructure of the original Sn–Pb solder is shown in Fig. 19.2a showing white contrast for tin grains and dark contrast for lead grains. The average grain size of the as-cast Sn–Pb solder was 5.12 µm. Morris_Ch19.indd 419Morris_Ch19.indd 419 9/29/2008 8:09:26 PM9/29/2008 8:09:26 PM 420 K.M. Kumar et al. Figure 19.2b shows the highly refined microstructure of 0.3 wt% SWCNT-doped Sn–Pb composite solder, which is a consequence of homogenous dispersion of the nanotubes. The average grain size of the composite is measured to be 1.08 µm by employing image analysis software. An obvious difference between the microstruc- tures of the solder alloys with and without addition of nanotubes can be observed. There is some porosity observed in the solder matrix. This is mainly attributed to the sintering process. During the sintering process, the matter of the solder matrix flows and the SWCNTs act as solid impurities [24] . The van der Waal forces cause the SWCNTs to get entangled with one another. Because of this phenomenon, it is very difficult to achieve a higher degree of homogeneous dispersion of the SWCNTs throughout the solder matrix. In this manner, the entangled SWCNTs may have resulted in the forma- tion of pores in the solder matrix, which is being observed in the micrographs. Fig. 19.1 Images of SWCNTs: (a) scanning electron microscopy (SEM) image of SWCNT, (b) TEM micrograph of SWCNT produced by a chemical vapor deposition (CVD) process ba Fig. 19.2 FE-SEM micrographs of 63Sn-37Pb solder with ( a ) 0 wt% SWCNT, ( b ) 0.3 wt% SWCNT P Solder matrix PbPb ba Pb Sn CNT distribution Sn 10µm10µm Morris_Ch19.indd 420Morris_Ch19.indd 420 9/29/2008 8:09:26 PM9/29/2008 8:09:26 PM 19 Properties of 63Sn-37Pb and Sn-3.8Ag-0.7Cu Solders Reinforced 421 Figure 19.3 compares the microstructures of the Sn–Ag–Cu + SWCNT compos- ite solders and pure Sn–Ag–Cu solder. The higher magnification micrographs in Fig. 19.3a reveal that the microstructure of Sn–Ag–Cu solder is composed of a dark-gray phase (Cu 6 Sn 5 ) and brighter light-gray grains (Ag 3 Sn) dispersed evenly in the β-Sn solder matrix. For the Sn–Ag–Cu pure solder sample, the average grain size of the secondary phase varied between 3.75 and 4.25 µm. The average grain size of the secondary phase was found to be 0.5–0.8 µm with 1 wt% addition of nanotubes to the Sn–Ag–Cu solder as shown in Fig. 19.3b . In the SWCNT-rein- forced solder samples, the SWCNTs are distributed at the boundaries of the Ag 3 Sn equiaxed grains. They can be identified by the difference in contrast, which is mainly associated with the different atomic numbers of the individual phases under consideration. Brighter regions correspond to the higher atomic numbers while darker phases correspond to the lower atomic numbers. The elemental analysis obtained by EDX is shown in Fig. 19.4 . The intense “C” peak represents the presence Fig. 19.3 FE-SEM micrographs of Sn-3.8Ag-0.7Cu with ( a ) 0 wt% SWCNT, ( b ) 1 wt% SWCNT 10µm Sn a b Ag3Sn Ag3Sn Cu6Sn5 Cu6Sn5 Sn 10µm Fig. 19.4 Phase identification of SWCNT at the grain boundary of Ag 3 Sn in sintered Sn–Ag–Cu/ CNT composite: EDS of the white region showing the presence of carbon 3µm Morris_Ch19.indd 421Morris_Ch19.indd 421 9/29/2008 8:09:26 PM9/29/2008 8:09:26 PM 422 K.M. Kumar et al. Fig. 19.5 Variation of CTE of both Sn–Pb and Sn–Ag–Cu composites with weight percent of SWCNTs Wt% of SWCNTs CT E (10 ^- 6/ K) Sn-Pb Sn-Ag-Cu 10 0 0.01 0.03 0.05 0.08 0.1 0.3 0.5 0.8 1 15 20 25 30 of SWCNTs at the boundaries of the Ag 3 Sn grains. This shows that the SWCNTs remained inside the solder matrix after sintering, but were concentrated at the boundaries of the Ag 3 Sn grains. The possible reason for the size refinement is as follows. SWCNT is a ceramic material. While processing the composite solder specimens, the surface diffusion of the Ag 3 Sn can be suppressed by the extremely quick translations of ceramic materi- als through the temperatures that exist during the sintering process [25] . The rein- forcement of the microstructure, as shown in Fig. 19.2 with the varying content of SWCNT, demonstrates a strong dependence of the sintered microstructure of the composite solders on the initial composition and morphology of the starting materials. 19.3.2 Coefficient of Thermal Expansion (CTE) The CTE was measured using TMA and was obtained from the initial linear slope of the thermal strain–temperature plot. The CTEs of pure Sn–Pb and Sn–Ag–Cu were found to be 25.8 × 10 −6 and 18.7 × 10 −6 /°C, respectively, which are comparable with those in the literature [26 , 27] . The variation of CTE with weight percent of SWCNT addition for the Sn–Pb and Sn–Ag–Cu composites is shown in Fig. 19.5 . The composite solders exhibit lower CTE values than the parent alloys. It was observed that the CTE of both the solders decreases with increasing content of SWCNT. In general, the lower CTE can be attributed to the rigidity of the nano- tubes and the fine dispersion of nanotubes in the solder matrix, which can obstruct Morris_Ch19.indd 422Morris_Ch19.indd 422 9/29/2008 8:09:27 PM9/29/2008 8:09:27 PM 19 Properties of 63Sn-37Pb and Sn-3.8Ag-0.7Cu Solders Reinforced 423 the expansion of the solder matrix at elevated temperatures. However, factors such as the adhesion of nanotube-matrix interfaces at testing temperatures, the apparent lack of orientation of the nanotubes, and the inevitable agglomeration at higher nanotube loads might affect the CTE values of nanocomposite solders and need to be confirmed by further studies and analysis. 19.3.3 DSC Analysis DSC measurements were carried out to determine the thermal properties such as melting point and onset temperature of both Sn–Pb and Sn–Ag–Cu composite sol- ders containing varying amounts of SWCNTs. The results are given in Table 19.1 . Typical DSC thermograms of the Sn–Pb and Sn–Ag–Cu solders and their compos- ites with SWCNTs are shown in Figs. 19.6 and 19.7 . The shapes of the thermo- grams closely resemble