Microstructure of Ag-Sn Bonding for MEMS Packaging
ABSTRACT Different metallization systems and bonding designs of Ag-Sn bonding were investigated to achieve good bonding. The bonding strength was evaluated by shear force. The microstructure of bonding interface was inspected by scanning electronic microscopy and ED AX. Shear force test was performed for as-bonded dice. The test results indicate differences among different metallization systems. The bonding pair with Ti/Au as the UBM has a quite low shear strength because of the bad adhesion on the silicon substrate. The bonding pair of Ti/Ni/Sn/Au and Ti/Ni/Au/Ag obviously has higher shear strength than that of Ti/Ni/Sn/Au and Ti/Ni/Au/Ag/Au. The former is 55.17 MPa on average while the later is 36.05 MPa. The shear strength of the pair of Ti/Ni/Sn/Au and Ti/Ni/Au/Ag is similar to that of Ti/Ni/Sn/Au and Ti/Ag which has the shear strength of 55.32 MPa on average. The Ni and Au in the Ag-Sn bonding system have significant effect on the microstructure of the bonding interface. The diffusion of Au into Sn is quicker than both Ag and Ni. The diffusion between Au and Sn would induce the obstacle of the inter-diffusion between Sn and Ag. Ni will also diffuse quickly into Sn and form Ni3Sn4. The existence of Ni in Sn will also influence the diffusion of Ag into Sn and make the bad wettability during bonding. After several metallization systems have been investigated, finally a uniform bonding layer has been achieved by excluding Ni and Au in the bonding system. The bonding interface is Ag3Sn layer dispersed with some pure Ag.
- [show abstract] [hide abstract]
ABSTRACT: Interfacial reactions in bimetallic Ag-Sn thin film couples have been investigated by measurement of electrical resistance and contact resistance as a function of time and temperature in order to understand kinetic behaviour in the above system where the intermetallic phase γ-Ag3Sn is formed. Since the reaction is found to start at room temperature, the conventional vacuum coating unit has been modified for preparing such films and conducting subsequent measurements without breaking the vacuum. The results from the above different methods of resistance measurement indicate that interfacial reactions are characterized by a mean diffusion coefficient of 10-13 cm2 s-1 at room temperature. X-ray diffraction indicates growth of the γ-Ag3Sn phase immediately after deposition. Scanning electron microscopy confirms the diffusion of tin into silver by grain boundary diffusion rather than by bulk diffusion. The results from transmission electron microscopy confirm the presence of a γ-Ag3Sn phase.Thin Solid Films 01/1987; 155(2):243-253. · 1.60 Impact Factor
- [show abstract] [hide abstract]
ABSTRACT: A series of Sn-Ag solders were prepared by arc melting and their phase evolution was investigated as a function of cooling rates. It was found that bulk Ag3Sn intermetallic compounds (IMCs) separated out only in the slowly cooled Sn-4.0Ag solder. This would be explained by the strong kinetic undercooling, arising from the rapid cooling conditions, which leads to the actual eutectic point shifts in the direction of higher Ag concentration. Thus, the eutectic and hypereutectic alloys experience a metastable hypoeutectic solidification route instead. All formed fractions of bulk Ag3Sn IMCs in solders, measured by quantitative microstructural analysis and thermal analysis, are larger than those predicted by the equilibrium phase diagram. The reasoning for this could be attributed to fine Ag3Sn phases, which cling to the primary Ag3Sn crystal during the eutectic reaction for their matching crystalline orientation relationship. Furthermore, the fraction of bulk Ag3Sn IMCs increases gradually with increasing the cooling rate in the slowly cooled Sn-4.0Ag alloy, which fits with the prediction of eutectic solidification theory: the increase of cooling rate would decrease the surface energy of fine Ag3Sn particles and primary Ag3Sn crystal, and make fine Ag3Sn particles cling to primary Ag3Sn crystal easily to form bulk Ag3Sn IMCs.Journal of Electronic Materials 01/2005; 34(12):1591-1597. · 1.64 Impact Factor