Fatigue Properties and Reliability of Solder Joints in BGA Assembly
Type of DegreePhD Dissertation
Industrial and Systems Engineering
MetadataShow full item record
Lead-free near-eutectic Sn-Ag-Cu (SAC) solder system has raised increasing attention in the electronics industry since the hazardous effects of SnPb solders. One of the critical considerations for the reliability of an electronic product is the fatigue failure of interconnected solder joints. Numerous studies have investigated fatigue properties of SAC solder materials using large bulk samples, and later the more realistic individual solder joints. However, solder joints in BGA components would suffer from more complex situations and thus make the fatigue life of the component different from the individual solder joints. This dissertation includes two types of test vehicles: the individual solder joint and the customized solder joints in BGA assembly. The latter test vehicle has 3×3 solder joints interconnected between two substrates to represent a realistic chip. The first study of the research focused on the reliability of SAC-based individual solder joints. Low-temperature solder (LTS) alloys have recently received considerable attention because of their inexpensive price and the reduced defects in complex assemblies. The shear and fatigue properties of individual solder joints were tested using an Instron micromechanical testing system. Two LTS (Sn-58Bi-0.5Sb-0.15Ni and Sn-42Bi) with low melting temperatures were examined and compared with Sn-3.5Ag and Sn-3.0Ag-0.8Cu-3.0Bi. The surface finish was electroless nickel-immersion gold (ENIG) during the test. Sn-3.5Ag solder with organic solderability preservative (OSP) surface finish was tested as well, for comparison. Shear testing was conducted at three strain rates, and the shear strength of each solder alloy was measured. A constant strain rate was used for the cyclic fatigue experiments. The fatigue life of each alloy was determined for various stress amplitudes. The failure mechanism in shear and fatigue tests were characterized using scanning electron microscopy/energy-dispersive spectroscopy (SEM/EDS). The results revealed that Sn-3.0Ag-0.8Cu-3.0Bi had superior shear and fatigue properties compared to other alloys but was more susceptible to brittle failure. The shear strain rate affected the failure modes of Sn-3.0Ag-0.8Cu-3.0Bi, Sn-58Bi-0.5Sb-0.15Ni, and Sn-42Bi; however, Sn-3.5Ag was found to be insensitive. Several failure modes were detected for Sn-3.5Ag in both shear strength and fatigue tests. For Sn-3.5Ag solder alloy, the OSP surface provided better interfacial toughness than the ENIG surface finish. In the second part of the research, customized sandwich test vehicles with 3×3 solder joints connected between two substrates were manufactured. Instron Micromechanical Tester is used to test the SAC305 solder joints using both the stress-controlled and strain-controlled methods at room temperature. The testing was conducted at a constant strain rate of 0.05s-1. Four stresses and four strain levels of the solder alloy Sn-3.0Ag-0.5Cu (SAC305) were examined using organic solderability preservative (OSP) and electroless nickel-immersion silver (ENIG) surface finishes. The work per cycle and plastic strain range were computed based on a systematic recording of the stress-strain (hysteresis) loops of each sample. A novel approach based on inelastic work is developed to calculate the fatigue life of a BGA assembled test vehicle. The results of the stress-controlled and strain-controlled tests indicated that the OSP surface finish outperformed the ENIG surface finish. Regardless of the testing process and surface finish, the Coffin-Manson and Morrow energy models were acceptable for SAC305. The third study investigated the fatigue performance of some other micro-alloying solder alloys besides Sn-3.0Ag-0.5Cu (SAC305). These lead-free solder alloys are Sn-3.5Ag-0.7Cu-3Bi-1.5Sb0.125Ni (SAC-I), Sn-3.41Ag-0.52Cu-3.3Bi (SAC-Q), and Sn0.92Cu-2.46Bi (SAC-R). The fatigue performance of these solder alloys was compared considering the effects of surface finishes (OSP and ENIG) and testing approaches (stress-controlled and strain-controlled). The SEM and EDS were utilized to determine the microstructure and failure mechanism of each solder alloy. The results showed that OSP surface finish outperformed the ENIG surface finish, regardless of testing methods and solder alloys. The interfacial IMC layer of SAC305 with the OSP surface finish was scallop-like Cu6Sn5, whereas smoother layers were observed in the SAC-R, SAC-Q, and SAC-I solder joints. SAC-Q and SAC-I associate with ENIG surface finish performed the brittle failure. They are more susceptible to changes in strain and stress, particularly strain. The composition of the IMC layer was dependent on the concentration of Cu in the solder alloys. The final study proposed a mechanical fatigue test method under low temperature (248K), room temperature (298K), and elevated temperature (348K). The same solder joints were assembled in the BGA configuration. The investigated solder alloys were SAC305 (Sn-3.0Ag-0.5Cu) and SnPb (Sn-37Pb). Two types of surface finishes (OSP and ENIG) were utilized for all the testing solder alloys to study the effect of surface finish. Strain-controlled tests were performed using the Instron Micromechanical Tester. It was found that the characteristic fatigue life decreased with the increase of strain level or testing temperature because the solder joint experienced more damage every cycle. The higher testing temperature also led to the larger plastic strain range, the more inelastic work, and decreased peak stress for solder joints in BGA assembly. The temperature of 348K tends to amplify this effect. The OSP surface finish outperformed the ENIG surface finish regardless of strain level or testing temperature due to the failure mechanisms. An empirical model was suitable to describe the effects of strain level on the fatigue behavior of SAC305 and SnPb solder joints at the temperature of 248K, 298K, and 348K. The modified empirical model was proposed to correlate fatigue life, strain level, and testing temperature. The failure mode in each case was identified.