Thermal Cycling Reliability of Doped SnAgCu Solder Alloys after Long-Term Aging

Abstract

Eutectic SnPb (Tin-Lead) solder has been used in electronics since early days. This universal alloy combination was used in all soldering applications because it had good mechanical and electrical properties and hence had good reliability. But, by the end of 20th century, harmful effects of lead were identified. Since then the industry has been pushing toward lead free electronics. In 2,000, the European Union put forward two directives towards lead free electronics. The Waste of Electrical and Electronic Equipment (WEEE) directive stipulated that lead should be removed from all electrical and electronic components at the end of life. The Restriction of Hazardous Substances (RoHS) directive prohibits the use of lead in electrical and electronic components manufactured after July 1, 2006. As a result of these and similar directives, the industry went through an intense search for replacements. Research led to a series of near eutectic alloys based on tin (Sn)-silver (Ag)-copper (Cu), commonly known as SAC solder alloy. With the advent of new materials, the reliability of solder joints became a major concern, especially in harsh environment. In electronics, the reliability is typically limited by the fatigue failure of a single solder joint. Aging makes the situation worse by altering the mechanical and physical properties of the solder material. The precipitate coarsening and the growth of the brittle intermetallic compound layer over time weakens the solder joint and hence deteriorates the reliability. Several elements such as bismuth, nickel, antimony, cobalt and indium have been microalloyed with the SAC based solder alloy to mitigate the adverse effects of aging. In this study, thermal cycling reliability of aged SAC based solder alloys is examined. Twelve solder pastes from the leading manufacturers with three solder spheres, namely, SAC105, SAC305 and match (where solder paste is same as solder sphere) and three surface finishes (ENIG, ImAg and OSP) are examined. For certain solder pastes, matching spheres could not be used due to their unavailability. The test vehicle consists of three 15mmx15mmCABGA208s, three 6mmx6mm CABGA36s, three 5mmx5mm MLFQFNs and a bank of six 2512SM resistors connected in series. A printed circuit board was made of four layers of FR-4 glass epoxy substrate with non solder mask defined pads. Several boards were aged at 125 C for a period of twelve months starting immediately after the assembly. The aged boards were subjected to thermal cycling in a temperature range -40 C to +125 C with 15 minutes dwell at +125 C and 10 minutes dwell at -40 C. The profile had a ramp time of 50 minutes, which corresponds to a rate of about 3.3 C per minute. The components were continuously monitored throughout the test. The failure data collected was used for statistical analysis using Weibull and DOE-ANOVA methods. After the test, the failed components were cross-sectioned and analyzed using optical and scanning electron microscopes to have an understanding about different failure modes. From the study, it could be concluded that SAC based alloys with elements such as Bi, Sb, and Ni had better fatigue resistance and therefore better reliability than SAC305 alloy. ENIG finish with its Ni layer performed better than the other surface finishes. It could also be concluded that the reliability depends on the combination of different factors considered in the study (solder paste, sphere and finish). For example, just because of the fact that ENIG performs better than ImAg and OSP, it does not guarantee that ENIG has better reliability in all combinations of solder paste/sphere/finish as the material properties of the whole solder joint vary with the combination (interaction effect of factors).