EFFECT OF Au-INTERMETALLIC COMPOUNDS ON MECHANICAL RELIABILITY OF Sn-Pb/Au-Ni-Cu JOINTS

Transcription

1 EFFECT OF Au-INTERMETALLIC COMPOUNDS ON MECHANICAL RELIABILITY OF Sn-Pb/Au-Ni-Cu JOINTS A. Zribi, R.R. Chromik, R. Presthus, J.Clum, L. Zavalij and E.J. Cotts Binghamton University; Physics Dept.; P.O. Box 6016; Binghamton, NY ABSTRACT--We have studied the effect of Au thickness and plating technology on the microstructure of Cu/Ni/Au/eutectic-Pb-Sn solder joints and the effect of joint morphology and microstructure on their mechanical reliability. Substrates with various Au platings and hence different Au thicknesses (ranging between 0.2 and 2.6 µm) were considered for this study. The concentration of Au in the studied joints was lower than 0.2 wt%. To investigate the effect of thermal annealing, we aged samples for up to 450 h at 150 C in an inert atmosphere. Optical metallography, electron metallography and Energy Dispersive Spectroscopy were used to characterize the bulk composition of aged and unaged joints. The binary Ni 3 Sn 4 compound was detected in the unaged samples at the interface solder while an additional ternary intermetallic (Au 0.5 Ni 0.5 )Sn 4 formed at the Ni 3 Sn 4 solder interface in some of the aged samples. The mechanical reliability of the joints was investigated using a tensile-testing machine with a special fixture. The aged joints tend to fail in a more brittle mode than unaged joints. Higher Au concentrations are shown to impair the mechanical strength of the joint. The fracture surface was located inside the solder balls for the ductile joints, whereas brittle joints fractured at the solder-pad metalization level. The toughness of these later joints decreased by a factor of two as the joints aged. This was correlated with the growth of the ternary (Au 0.5 Ni 0.5 )Sn 4 adjacent to the interface. I. INTRODUCTION While dissolution rates of eutectic Pb-Sn solder on Cu substrates are fairly well known, the solderability, and integrity of different solder/substrate combinations has become an issue. For instance, the increased use of Ni/Au metallizations creates the need for an increase in the scope of characterization of solder/metal systems. The trend in the microelectronics industry to use devices of submicron length scales has increased this need for understanding the interconnect materials which bond these devices to the board. In traditional attachment processes, using a Sn based solder, an intermetallic compound (M x Sn 1-x ) forms where M is some finishing or plating material, such as Pd, Ni, Cu, or Au (Ohriner, et. al., 1987; Kim, et. al., 1994; Chromik,1996; Mei, et.al., 1998). In some cases, two intermetallics may form during the soldering process and one or both may continue to grow in the solid state during thermal aging. The mechanical reliability of these intermetallic compounds clearly influences the mechanical integrity of the solder joint. In fact, brittle failure of solder balls has been identified with the growth of a number of intermetallic compounds both at the interfaces between metallic layers and in the bulk of solder balls. These compounds tend to be brittle at ambient temperatures and are considered a potential cause of crack initiation and growth in electronic assemblies. For example, the Au intermetallic compound, AuSn 4, has been noted for its particularly deleterious effect on both bulk and interface properties of solder joints Many studies, including those which introduced Au to the solder joint in Au-Pb-Sn solder, established that Au has the tendency to embrittle Au-alloyed joints as the Au concentration exceeds 2 wt % -3 wt % (Ohriner, et.al.,1987; Chromik,1996; Mei, et.al.,1998; Kramer, et.al.,1994). In the wide range of explored concentrations (2wt%-7wt%) of Au, AuSn 4 manifested as small precipitates dispersed everywhere inside a Sn-rich phase matrix. These precipitates seem to alter the creep behavior of the joint at high temperatures and to reduce its toughness at lower temperatures. An understanding of the growth of intermetallic compounds in solder joints will enhance our ability to control their influence on joint properties. The growth of intermetallic compounds in solder joints takes place by the interdiffusion of the parent phases of M (Pd, Ni, Cu, Au) and Sn. However, it is has been observed that in the case of many of these solder/metal diffusion systems that one species diffuses much more rapidly. It is also true that the first phase to form in these systems is always the most Sn rich. For instance, Ni 3 Sn 4, Cu 6 Sn 5, and AuSn 4 in the Ni-Sn, Cu-Sn and Au-Sn systems, respectively. To help understand this phase selection and preferred diffusing species, one must turn to a number of studies done concerning the solute diffusion of noble metals and near noble metals in Sn and Pb (Kramer,et.al.,1994; Kim,et.al.,1997;

2 Ferguson,et.al.,1997). Researchers found that solutes in the platinum and noble metal groups of the periodic table diffused very rapidly in Sn and Pb. This rapid diffusion was attributed to an interstitial diffusion mechanism. These results have been used previously to explain the first phase selection of the most Sn rich phase (Nakahara, et.al.,1981). An atom, such as Cu, may diffuse easily into Sn by this rapid, interstitial mechanism, where Sn diffuses in Cu by a slower, substitutional mechanism. An aspect of these systems which has been studied recently (Chromik, 1996) is the possible existence of a continued interstitial diffusion mechanism within the alloys. Averaged interdiffusion coefficients in a number of alloys of interest have been estimated, and reaction constants have been measured, in a number of solder alloys such as Ni 3 Sn 4, Cu 6 Sn 5, and PdSn 4. These determinations were compared to the solute diffusion coefficients for the relevant systems. Comparisons have resulted in the recognition of the trend in these solder metal systems for a decreasing averaged interdiffusion coefficient as a function of increasing metal content of the solder alloy. Consideration of the Sn content of a solder alloy will indicate the rate of formation of and likelihood for that solder ally to form in a solder/metal diffusion couple, with alloys of greater Sn content forming first and fastest. It has been more difficult to understand the growth of solder alloys in more complex, multi component systems such as those found in ball grid array Cu/Ni/Au/Pb/Sn solder joints. Problems in loss of joint toughness upon thermal annealing at intermediate temperatures (150 o C) have been previously observed. The presence of Au in these joints is primarily meant to address the problem of Ni oxidation before reflow. However, Au has a rather high solubility in molten eutectic solder and dissolves very fast during reflow (Kim,et.al., 1994; Chromik, 1996; Mei,et.al.,1998). The ultimate joint morphology and microstructure depends upon the initial thickness of Au, the reflow profile and the annealing history of the sample. However, previous works (Ohriner,et.al.,1987) reported the growth of AuSn 4 at the Ni/solder interface. This compound was observed to form inside the solder spheres and apparently migrate after annealing towards the binary intermetallic Ni 3Sn 4 located at the Ni/solder interface. In this work we seek to characterize this process in order to better understand the driving force and mechanism. This work concerned ball grid array solder joints with a low gold concentration of less than 0.2 wt%. In fact, we observe similar decreases in joint toughness in these ball grid array Cu/Ni/Au/PbSn solder joints upon thermal aging, but we correlate this with the growth of a ternary (Ni 0.5 Au 0.5 )Sn 4 at the Ni 3 Sn 4 /PbSn interface. The identification of this ternary phase allows for an increased basis for the understanding of this process on the basis of trends identified for binary soldermetal systems: that the most Sn rich phase in the system tends to grow faster and form first at the solder metal interface. Substrates from two different vendors, with three Au/Ni plating categories (electrolytic, immersion and selective) were used to prepare the samples. We investigated the effect of thermal annealing on the mechanical properties of the joints, and correlated thermal aging and changes in mechanical properties with the formation of solder alloys. II. EXPERIMENTAL We used substrates from two different vendors with the same pad metallurgy, Ni/Au-coated copper, to prepare the samples. Three plating techniques were used to plate Ni and Au on top of the copper, identified here as electrolytic, immersion and selective. These three types of sample were distinguished by the thickness of the Au layers, 0.75, 0.25 and 0.02 µm in the electrolytic, immersion and selective samples, respectively (limited study of a second series of electrolytic samples with 2.6 µm Au layers is included as well).the thickness of the Au layer was measured for all the samples using a Scanning Auger Microprobe. Substrates were populated with BGA (ball grid array) eutectic solder spheres arranged in one row around the perimeter of the board layout. The solder balls were placed with a placement machine then annealed (thermal reflow) in a convection oven. The reflow was carried out according to vendor recommendations, which are summarized in the temperature profile shown in Fig. 1. After cooling, the populated boards were cut into smaller samples, sealed in argon filled glass Temperature ( o C) o C reflow temperature 183 o C melting point of eutectic solder 45 s Time (s) Figure 1: Temperature profile.

3 Figure 2: Load-displacement curve for an unaged sample. tubes, and annealed for different times at a temperature of 150 C. The samples were subsequently cross-sectioned and polished for inspection by means of optical and electron microscopy. For mechanical testing, the samples were mounted to a special fixture designed for the test and each solder sphere was individually tested using an INSTRON tensile-testing machine. The testing machine generates force-displacement curves used to determine the failure mode (brittle or ductile), the breakage energy and the strength of the studied joint. III. RESULTS Separate samples of all categories were aged at 150 C for 0.5 h, 1 h, 4 h, 9 h, 40 h, 150 h and 450 h and prepared for mechanical testing, or optical and electron metallography inspections, together with unaged speciemens. In all these samples Cu, Ni, Ni 3 Sn 4 and Pb- Sn solder phases were identified by optical and electron microscopy together with EDS (Energy Dispersive Spectroscopy) spectra. As expected, Ni 3 Sn 4 nucleated along the Ni-solder interface. The thickness of this intermetallic compound was observed to increase with the annealing time. After 150 h of aging at 150 C, a new ternary Ni-Au-Sn intermetallic was viewed adjacent to the interface solder/ni in the electrolytic samples. This phase continued growing at the same location as a result of continued aging of the electrolytic specimens. The thickness increase of this phase has been detected using 1000X-optical micrographs and EDX (Energy Dispersive X-ray analysis) linescans performed across the interface solder/ni. After 450 h at 150 C the same phase was detected in the immersion samples. Microchemical analysis was undertaken to determine the average composition of this solder Figure 3: Load-displacement curve for a specimen aged 150 h at 150 C. alloy that forms in our samples. Using the backscattered electrons imaging mode of our microprobe, we were able to locate the studied phase. Semi-quantitative analysis (ZAF) was than carried out inside this phase to reveal the approximate stoichiometry (Au 0.5 Ni 0.5 Sn 4 ). Similar trends were observed occasionally in some of the immersion samples whereas no ternary intermetallic compounds were detected in the selective samples. Mechanical testing revealed two distinct modes of failure of the BGA assemblies. These failure modes were identified using the forcedisplacement curves plotted for aged and unaged samples. The brittleness or ductility of a joint were quantified by the strength level and the breakage energy. Fig. 2 and Fig. 3 respectively show typical force-displacement curves of a ductile and a brittle type joint. Fig. 2 corresponds to an as-reflowed sample, whereas Figure 3 belongs to an electrolytic sample annealed for 150 h at 150 C. As illustrated in these figures, the tensile strength and the breakage energy (surface area under the force-displacement curve) of the joint for the unaged samples exceeded that of the aged samples. This alteration of the mechanical properties describes a drop of the joint ductility, which is identified with the formation of the ternary intermetallic (Au 0.5 Ni 0.5 Sn 4 ) adjacent to the interface solder-ni. IV. DISCUSSION Some understanding of the growth of (Au 0.5 Ni 0.5 )Sn 4 can be gained by considering the general tendency for the most Sn rich phases in a solder-metal system to grow first and fastest. First consider the physical situation during and right after reflow. During reflow, the interaction between molten solder and the solid metalizations of the substrates resulted in interfacial dissolution reactions. Considering that the dissolution rates of Au is approximetely three orders of magnitude higher than

4 that of Ni, and that the Ni layers were approximately two orders of magnitude thicker than the Au layers, we understand that the top, Au layer completely dissolves during reflow, while only a fraction of the Ni layer dissolves. During reflow a reaction at the Ni/solder interface forms a layer of thickness on order of 1 µm of the Sn rich Ni-Sn phase, Ni 3 Sn 4. Upon cooling to room temperature, a limited amount of the Au in solution may percipitate as AuSn 4 in the solder ball. Thus the condition of the solder joint is determined before thermal aging. After reflow we find a layer of Ni 3 Sn 4 between solder and Ni (additional Ni-P phases are also identified between the Ni 3 Sn 4 and Ni in the electroless samples). The Au is essentially dispersed in the solder ball. It is clear from phase diagrams that the system can lower its free energy by forming agglomerations, or a single agglomeration of AuSn 4, with a thickening of the Ni 3 Sn 4 layer. It may be less clear why AuSn 4 would tend to agglomerate as a layer at the Ni 3 Sn 4 /solder interface, as previously reported (Mei, et.al.,1998). When evidence that this phase is (Au 0.5 Ni 0.5 )Sn 4 is considered, once can better understand why this (Au 0.5 Ni o.5 )Sn 4 phase grew at this interface. Relative to Sn rich solder alloys such as AuSn 4 or PdSn 4, Ni 3 Sn 4 is a relatively slow growing phase. If a more Sn rich phase was present in the Ni- Sn phase diagram, there is little doubt that it would grow preferentially. Consideration of the Au-Ni-Sn ternary phase diagram leads us to essentially identify such a phase: (Au 0.5 Ni 0.5 )Sn 4. Fig. 4 shows the Snrich corner of a room-temperature section of the Au- Ni-Sn ternary phase diagram (Anh ck, et.al.,1998). Based on previous studies of solder-metal systems, it Figure 4: The Sn-rich corner of a roomtemperature Au-Ni-Sn phase diagram (Anh ck,et.al., 1998). is expected that Ni will diffuse much more rapidly in this ternary phase than in Ni 3 Sn 4. Considering this along with the rapid diffusion rate of Au in Sn (PbSn solder), we have the kinetic preference for this growth of (Au 0.5 Ni 0.5 )Sn 4 at the Ni 3 Sn 4 /solder interface. V. CONCLUSIONS We correlated the growth of ( Au 0.5 Ni o.5 )Sn 4 phase at the solder/ni 3 Sn 4 interface in ball grid array Cu/Ni/Au/PbSn solder joints upon thermal aging at a temperature of 150 o C with decreases in the mechanical toughness of the solder joints. Comparisons of the growth of this ternary alloy to the growth by solid state diffusion of binary metal-sn alloys in other solder joints provided some understanding of this phenomenon. VI. ACKNOWLEDGMENTS This research was funded by the Integrated Electronics Engineering Research Center (IEEC) located in the Watson School at Binghamton University. The IEEC receives funding from the New York State Science and Technology Foundation, the National Science Foundation and a consortium of industrial members. We gratefully acknowledge discussions with Peter Borgesen, and the use of mechanical testing facilities at Universal Instruments. The Au thickness measurements were provided by Universal Instruments. We thank Ms. Susan Pitely for her efforts in helping us edit and prepare this paper for submission. VII. REFERENCES Anh ck, S., Oppermann, H., Kallmayer, C., Aschenbrenner, R., Thomas, L., Reichl, H., "Investigations of Au/Sn alloys on different endmetallizations for high temperature applications" IEEE/CPMT Electronics Manufacturing Technology Symposium, 1998, p Chromik, R. R., "The Thermodynamics and Kinetics of Solid State Reactions In The Pd-Sn System", Masters Thesis SUNY at Binghamton, Ferguson, M. E., Fieselman, C.D., Elkins, M.A., "Manufacturing Concerns When Soldering with Au Plated Component Leads or Circuit Board Pads", IEEE Transactions on Components, Packaging, and Manufacturing Technology-Part C, Vol. 20, No. 3, July 1997, p Kim, H. K., Liou, H.K., Tu, K.N, "Morphology of instability of the wetting tips of eutectic SnBi, eutectic Sn-Pb, and pure Sn on Cu", Materials Research Society, Vol. 10, No.3, July 1994, p

5 Kim, P.G. and Tu, K.N., "Fast Soldering Reactions On Au Foils", Materials Research Society, Vol.445, 1997, p Kramer, P.A., Glazer, J., Morris, J.W., Jr., "The Effect of Low Au Concentrations on the Creep of Eutectic Sn-Lead Joints", Metallurgical and Materials Transactions A, Vol.25A, June 1994, p Mei, Z., Kaufmann, M., Eslambolchi, A., Johnson, P., "Brittle Interfacial Fracture of PBGA Packages Soldered on Electroless Ni/Immersion Au", 1998 Proceedings of the 48th Electronic Components and Technology Conference 1998, p Nakahara, S., McCoy, R.J., Buene, L., Vandenberg, J.M., "Room Temperature Interdiffusion Studies Of Au/Sn Thin Film Couples", Thin Solid Films, No. 84, (1981), p Ohriner, E. K., "Intermetallic Formation in Soldered Copper-Based Alloys at 150 C to 250 C", Welding Journal: Research suppliment, July 1987, p.191s/202s.