Formation of a Self-Interconnected Joint using a Low-Melting-Point Alloy Adhesive

Transcription

1 Materials Transactions, Vol. 45, No. 3 (2004) pp. 799 to 805 Special Issue on Lead-Free Soldering in Electronics #2004 The Japan Institute of Metals Formation of a Self-Interconnected Joint using a Low-Melting-Point Alloy Adhesive Kiyokazu Yasuda, Jong-Min Kim* 1, Masahiro Yasuda* 2 and Kozo Fujimoto Department of Manufacturing Science, Graduate School of Engineering, Osaka University, Suita , Japan The self-interconnection process is one of the promising methodologies for joining novel materials and assembling micro-electronic devices. Basic experiments fabricating micro joints using conductive adhesives with low melting point solders, such as Sn-In eutectic alloy, were demonstrated as an alternative method to conventional soldering or using the usual adhesive joining used in electronic assembly. The joint morphology, the formation of electrical conduction interconnections, and the self-interconnection characteristics were examined by optical and X-ray transmission microscopy. The behaviors of melting fillers such as aggregation, coalescence, and wetting were found to occur during joint formation. Especially on the copper line patterns of a glass-epoxy substrate, selective adhesional wetting of the melting alloy, enhanced by the oxygen-reduction capability of resin and capillary phenomena, was the main driving forces of the self-interconnection of joints. The alignment, joint height, and volume fraction of the filler need to be set correctly for a successful adhesion to be achieved. (Received September 26, 2003; Accepted February 4, 2004) Keywords: self-interconnection, low-melting-point alloy filler, electric conductive adhesive, aggregation, coalescence, wetting, capillary force 1. Introduction * 1 Corresponding author, address: kjm@mapse.eng.osaka-u.ac.jp * 2 Graduate Student, Osaka University Solder alloys have been the main materials used in micro joints in electronic packaging. In particular, lead-free solders would be the first choice for such materials. 1,2) Though it forms an excellent and reliable joints with high electrical conductivity, a joining temperature is higher than about 500 K is required. Therefore, thermal damage would result in devices such as optical modules or opto-electronic packages which require a low heatproof temperature. Recently, electrically-conductive adhesives (ECA) which can be applied at low temperatures have received attention in terms of being a possible solution to this problem. 3) Conductive adhesives in the form of anisotropic conductive film (ACF) 4) and anisotropic conductive paste (ACP) are used for joints in the peripheral area of liquid crystal display panels. For the diversification of the electronics assembly, adhesive joining is promising and its practical use has been developed to realize low-temperature packaging at low cost. Up until now, many kinds of epoxy-system resins mixed with silver fillers have been widely used for the current major electronics assemblies. In these ECAs, the electrical conduction between electrodes is achieved through the conduction path by the contact of the silver fillers, 5) and the joint mechanical strength is obtained by cure hardening, cohesion, and the adhesion of the resin. However, since electrical conduction is obtained through mechanical contact, the consequent low reliability due to the unstable state of the conduction limits the electrical current by concentrating the resistance with the oxide film, and silver migration becomes a serious problem, 6 9) even though a few attempts have been made. 10,11) In this research, we propose the novel ECA process in which the self-interconnected conduction path can be formed with a low-melting-point alloy (LMPA) metallic fillers. It is the process (LMPA-ECA) which controls the joint height after filler melting but before resin curing. Because a conduction path having metallic bonds between the electrodes can be formed, a high electrical conduction can be obtained. 2. Concept of Self-Interconnected Joining by LMPA- ECA In the self-interconnected LMPA-ECA process, the adhesive with an oxygen-reduction capability alone mixed with fusible fillers is used and heated up to the temperature range below the curing peak point of the resin, but above the melting point of fillers. The mechanism of this joining process is the oxygen reduction reaction occurring on the metal surface before filler wetting. The resin has a cross-linking agent with an inherent flux activity represented by the formula R(COOH) x (x is larger than or equal to 2). The fluxing reaction proceeds according to the following reaction formula during the process. 2RCOOH þ CuO! Cu(RCOO) 2 þ H 2 O ð1þ Water as a product of the reaction could be used by the amide hydrolysis for epoxy curing, and then void formation is prevented. The motion of fillers occurs in the joint due to the thermal effect and hydrodynamic resin flow because the viscosity of the resin decreases in the pre-heating period. Then, the adjoining fillers collide, and partly coalesce and even grow to be a huge spherical blob by wetting. Finally, the joint is kept at a temperature of the heat around the temperature range over the curing peak point of the resin until it cures fully, and the joining process is completed. As for the electrical conduction path, when the coalesced filler size exceeds the gap distance between the electrodes, the conduction path forms and gets wet on both sides of the electrode. Though a large conduction path is formed, the direction of the connection is unidirectional, normal to the electrodes and anisotropic. An arrayed conduction path can be achieved only in the direction towards the small electrodes, shown in Fig. 1, by combining the optimal materials.

2 800 K. Yasuda, J.-M. Kim, M. Yasuda and K. Fujimoto LMPA adhesive Pad mm Chip Substrate 15 mm 50 mm Fig. 1 Schematic illustration of a self-interconnected joint of LMPA adhesive, initial state and final state after joining. 50 mm Table 1 Material properties of resin and filler. Density Viscosity Resistance Gel time (g/cm 3 ) (mpas) (cm) (s) Resin : Density Diameter Melting point Sn In (g/cm 3 ) (mm) (K) (%) (%) Filler (d) 3. Experimental Procedure 3.1 Materials In the experiment, a epoxy thermosetting resin (singleliquid no-filler type) was used which consists of a mixture of both a bisphenol A and a bisphenol F system. This is the epoxy resin normally used as an under-fill material having reductive characteristics against the surface oxide. The physical properties of the resin and filler are shown in Table 1. Sn-In eutectic alloy powder (globular type with a mean diameter 42 mm) was supplied as the metallic filler for which the melting point (390 K) is lower than the curing peak point of the resin. To form the adhesive paste filler and resin were mixed with a spatura. The volume fraction of the filler (V f ) was varied from 10 to 60%. The copper line-patterned glass-epoxy substrate (thickness: 1.6 mm) shown in Fig. 2 was fabricated to observe self-interconnection phenomena. The line width and pitch are mm. The substrates were prepared by cleaning with HCl acid (6%), acetone, and deionized water with ultrasonic cleaner for the self-interconnection model experiment. 3.2 Joining procedure and observation In order to confirm the self-interconnection by the joining process, the LMPA adhesive is supplied on a line-patterned glass-epoxy substrate. The adhesive was dispensed on the substrates with a spatula with an arbitrary quantity sufficient to fill the joint height between the upper and lower substrates. In order to control the joint height, spacers were used, that is, stainless wires (100 mm) or balls (300 mm) of fixed value, and stainless balls and solder balls (300 mm) for the twostep variable height control. For fixing and aligning the upper and lower substrates, two metal binder clips were used to hold the joint and were aligned by confirming the overlapping area of the line pattern in a micro-focus X-ray television system (SMX-160E:Shimadzu Corporation). Fig. 2 Line-patterned substrate and configuration of a substrate setup. the size and shape of the line-patterned substrate, a pair of nonpatterned and patterned substrates, aligned joints with faced pattern substrates, and misaligned joints with faced pattern substrates. The joint specimen was heated in the infrared heating chamber (SMT Scope: Sanyo Seiko Inc.) in air with a temperature profile. As the adhesive is heated at 413 K from the room temperature for 60s, which is about 20 K higher than the Sn-In melting point (390 K), the alloy completely melts and the temperature is maintained for 180 s. It is heated up to the standard curing peak point (453 K), and is afterwards maintained for 3600s in order to cure the resin completely. A non-pattern substrate was used as an upper substrate for observing simple wetting on the line pattern shown in Fig. 2. Two patterned substrates are joined with LMPA adhesive. In order to know the wetting mode (adhesional wetting, penetrating wetting, spreading wetting), aligned and misaligned pairs of substrates were prepared as shown in Figs. 2 and (d). For observing the cross-section, the joint specimen was cut normal to the line pattern near the center of the specimen using a micro-fine cutter. After cutting, the cross-section was polished using emery paper (#2000) and Al 2 O 3 (1 mm) powder. The formation of the conduction path was observed with cross-sectional samples using an optical CCD microscope (VK-8500: Keyence Co. Ltd.). The morphology was observed using the micro-focus X-ray television system. The focus size is 1 mm in diameter at minimum. 4. Results and Discussion 4.1 Aggregation of LMPA fillers in polymeric media The behavior of the LMPA process before melting in the aggregation of solid fillers is shown in Fig. 3. The degree of aggregation proceeds from to (d). The homogenous filler

3 Formation of a Self-Interconnected Joint using a Low-Melting-Point Alloy Adhesive µm (d) 250 µm substrate when the fillers is still solidus. The filler distribution has some tendency of local inhomogenity with the fluctuation of the filler packing density. The light gray zone represents the areas lacking in filler. In contrast, the distribution of filler on top of the copper line patterns is much more homogenous with a high density. This is thought to originate from the difference in the surface characteristics, such as the electrostatic and thermal conditions. Even after melting, the fillers would never have been trapped by wetting if they were on the glass-epoxy surface, although they interact with each other. Consequently, the number of collisions between them is enhanced, and, as a result, the morphology changes. 600 µm 600 µm Fig. 3 Micro-focus X-ray photographs showing filler aggregation proceeding from to (d). Homogenous dispersion, packing density fluctuation, large-scale low-density-domain appearance, and (d) network structure. distribution (Fig. 3) becomes inhomogeneous with a packing density fluctuation in. The domain with the low filler density (light gray) grows to hundreds of microns in diameter in. Then, the filler distribution was very inhomogeneous in (d). Eventually there existed no fillers in the low-density domains. Most of the fillers tend to accumulate with network-like structure. The surface-acting force is larger than the force of gravity if the filler size is sufficiently small, in the order of several tens of microns. From these results, it was thought that the surface state has changed to become very sticky as the surface energy of the metal is larger than that of the oxide. Figure 4 shows the filler distribution on the patterned Fig. 4 Micro-focus X-ray photograph showing filler density fluctuation on the copper patterned glass-epoxy surface. The white box in the figure is the copper patterned area. 4.2 Selective wetting of LMPA on Cu pattern There are two stages of self-interconnection processes after filler melting; the first stage is the lateral wetting growth, and the second is the vertical bridging stage between the metal pads. The cross-section of a pair of non-pattern and pattern substrates shown in Fig. 5 depicts the wetting phenomena on the copper surface in the first stage. In Fig. 5, the neighboring filler to the line pattern adheres on the surface of copper. It is observed that a copper line is partially covered by the alloy film. It clearly shows that adhesional wetting is the driving force of filler coalescence and is dominant in this stage. In, the copper line (center one) is globally covered by the single wetting sessile drop, although most fillers around are still small. In, the neighboring fillers coalesced and grew to larger alloy balls of about a hundred microns in diameter. This phenomenon is independent of the joint height. According to these results, it is necessary for drop growth to occur for the metal surface to have good wettability, and that the grass-epoxy surface has no wetting characteristics. As a consequence, the fillers on the grass-epoxy keep their spherical forms which results in good mobility in the liquid resin. At the latter stage of the self-interconnection of the joint formation shown in Fig. 6, much coalescence of the fillers has occurred, and even large fillers have accumulated in the wetting drop on the pattern. Filler coalescence is a kind of survival with size priority for the sake of surface energy minimization. Larger fillers can eat smaller ones as a result of the pressure difference caused by surface tension. The wetting drop on the metal surface essentially has a much larger diameter since the contact angle is small. Consequently, the drop grows further. The apparent influence of gravity on the filler distribution is observed, since the lower pattern tends to be covered initially by the liquid filler. 4.3 Effect of alignment on the wetting of the LMPA interconnection Figure 7 shows micro-focused X-ray transparent photographs and optical cross-sectional photographs of a line pattern area of the misaligned joints and aligned joints. In the matched area between the upper copper line and the lower line of the misaligned pair, the filler alloy is continuously wetting because of strong capillary action, while both the unmatched areas aside showed partial wetting in which isolated wetting domains form. Between the glassepoxy surfaces, fillers remain as spherical in appearance although their size is larger than they were initially.

4 802 K. Yasuda, J.-M. Kim, M. Yasuda and K. Fujimoto Fig. 5 Cross-sectional optical photographs showing self-interconnection phenomenon by adhesional wetting of the in-between patterned substrate (lower) and the no-patterned substrate (upper). Partial wetting, sessile drop formation, and growth of filler size. This is also good evidence to suggest that fillers show frequent collisions and coalescences between them, but poor wetting capability on the glass-epoxy surface, maintaining their high mobility in the polymeric media with a minimum of viscous resistance. With the misaligned substrates, the alloy melt formed the meniscus between the patterns. The contact angle is quite small since the copper surface has excellent wetting characteristics. The spherical fillers remain between the glass-epoxy surfaces. From these results, it is thought that fillers exhibit partial wetting if a misalignment of the substrates exists. This is due to the enlargement of the Cu-mismatched surface area. In Fig. 7(b-1) of the aligned pair of substrates, the melting alloy generally accumulates between the copper lines. With the help of the capillary force and the hydrodynamic flow of the polymeric media, the alloy completely covers the copper Fig. 6 Cross-sectional optical photographs showing self-interconnection phenomena for in-between paired copper line-patterned substrates. Sessile drop growth on the lower surface at first because of gravitational effects, followed by that on the upper surface, and finally the bridge between them to form the joint. line. However, residual fillers still exist on the glass-epoxy surface on the right hand of Fig. 7(b-1). These fillers are thought to flow along the epoxy channels by resin flow. In Fig. 7(b-2) the alloy penetrated between the copper lines. Even with a misaligned copper line, meniscus formation is possible. With an aligned copper line, capillary action is also the driving force of the self-interconnection, but the filler resides at the glass-epoxy region because of the excessive volume. From this observation, it is concluded that holding joint height is necessary to keep the melt between the leads, otherwise it spreads away along the gap to form a nonwetting chain of the filler melt domains shown in Fig. 7(b-2). With low joint height, penetrating wetting or capillary motion is important for enhancing self-interconnections.

5 Formation of a Self-Interconnected Joint using a Low-Melting-Point Alloy Adhesive 803 (a-1) X-ray photograph (a-2) Cross-sectional photograph Pad Fillet Matched line Mismatched line 300µm (b-1) X-ray photograph (b-2) Cross-sectional photograph Copper line Glass-epoxy line 300µm LMPA filler Fig. 7 Morphology and cross-sectional structures of joints by self-interconnection with misaligned and aligned substrates. 4.4 Influence of volume fraction and joint height The parameters such as the volume fraction of the filler and the joint height are the key parameters in the LMPA-ECA process for obtaining a good ordered joint. In Fig. 8, with a volume fraction of 10% and a constant joint height of 100 mm, there exist many non-wetting zones on the copper line and isolated fillers on the glass-epoxy. While with V f ¼ 30%, most fillers accumulate between the copper lines (Fig. 8). With H ¼ 300 mm, however, the wetting boundary is not straight. In contrast, with the height controlled from 300 to 100 mm (Fig. 8(d)), the wetting boundary is straight, though much coalesced filler still remains in the glass-epoxy surface region. Since the line patterned area is also filled with LMPA, the total volume of fillers is excessive for this condition. This means that not only the volume fraction of fillers, but also the total supply volume in the joint should be controlled to realize optimal self-interconnected joints. To see the influence of the alloy filler volume on the wetting geometry by self-interconnected joints, the difference in the cross-sectional observation result between the sample with constant height and one with height control is shown in Fig. 9. In Fig. 9, a concave surface of alloy was observed, while a convex surface was observed in the sample with height control. For the latter one, it is thought that the dynamic height change pushed the alloy melt between the copper line. This caused the internal pressure of the alloy liquid to increase. Then, in order to maintain the balance between the pressure and the surface tension, the curvature of the alloy surface became convex. For this reason, not only the surplus filler volume but also the pressure would prevent the melt of fillers from self-interconnecting. 5. Conclusion Micro joints formed with conductive adhesive with the low melting point solder of Sn-In eutectic alloy were demonstrated using optical and X-ray transmission microscopy. The joints were achieved using self-interconnection mechanisms, mainly derived by the selective aggregation and coalescence of solid filler and the subsequent wetting on a metal surface enhanced by the capillary driving force. The main results are summarized as follows: (1) The aggregation of solid alloy fillers occurred at the initial stage of the self-interconnection. The filler density fluctuation was enhanced on the glass-epoxy substrate, while not observed on copper surfaces. (2) Adhesional wetting on copper surfaces and bridge formation were achieved by the oxygen reduction capability of resin. The behaviors of liquid-state fillers such as coalescence and wetting were found to occur selectively in the joint area by the rule of surface energy minimization. (3) A misalignment of the pattern inhibits interconnections,

6 804 K. Yasuda, J.-M. Kim, M. Yasuda and K. Fujimoto (d) Fig. 8 Micro-focus X-ray photographs showing the influence of volume fraction (V f ) and stand-off height (H). V f ¼ 10%, H ¼ 100 mm, V f ¼ 30%, H ¼ 100 mm, V f ¼ 30%, H ¼ 300 mm, (d) V f ¼ 30%, H ¼ 300 mm! 100 mm. 50µm 50µm Fig. 9 Cross-sectional optical photographs of joints with constant height (100 mm), and with height control (300 to 100 mm). The filler alloy forms concave and convex surfaces at the sides of the joints. though capillary phenomena can help in the formation of joint patterns. (4) To utilize such an effect, the process parameters such as joint height and adhesive volume should be controlled to form optimal self-interconnected joints. Acknowledgments The authors would like to thank Sanyu Rec and Sunstar Engineering for supplying the resins, and Senjyu Metal Industry for supplying the alloy fillers. Part of this work was

7 Formation of a Self-Interconnected Joint using a Low-Melting-Point Alloy Adhesive 805 supported by the Priority Assistance of the Formation of Worldwide Renowned Centers of Research The 21st Century COE Program (Project: Center of Excellence for Advanced Structural and Functional Materials Design) from the Ministry of Education, Sports, Culture, Science and Technology of Japan. REFERENCES 1) K. Suganuma: Solid State Mater. Sci. 5 (2001) ) M. Abtew and G. Selvadurray: Mater. Sci. Eng. 27 (2000) ) Y. Ohta, K. Yasuda, K. Fujimoto, K. Nishikawa, Y. Yagi and H. Ohtani: Proc. 8th Symp. Microjoin. Assembly Techno. Electro. (Mate2002) Yokohama, 2002, p. 169 [in Japanese]. 4) M.-J. Yim and K.-W. Paik: IEEE Trans. Adv. Packag. 22 (1999) ) L. Ye, Z. Lai, J. Liu and A. Tholen: IEEE Trans. Electr., Packag., Manufact. 22 (1999) ) S. Kotthaus, R. Haug, H. Schafer and O. D. Hennemann: IEEE Trans. Comp., Packag., Manufact. Technol. A 21 (1998) ) D. Lu, Q. K. Tong and C. P. Wong: IEEE Trans. Electr., Packag., Manufact. 22 (1999) ) D. Lu, Q. K. Tong and C. P. Wong: IEEE Trans. Electr., Packag., Manufact. 22 (1999) ) R. Dudek and H. Berek, T. Fritsch, and B. Michel: IEEE Trans. Comp., Packag., Technol. 23 (2000) ) D. Lu and C. P. Wong: IEEE Trans. Comp., Packag., Technol. 23 (2000) ) D. Lu and C. P. Wong: J. Applied Polymer Science 74 (1999)