Author: Liang Yin
Abstract: The recrystallization of β-Sn profoundly affects deformation and failure of Sn-Ag-Cu solder joints in thermo-mechanical fatigue (TMF) testing. The numerous grain boundaries of recrystallized β-Sn enable grain boundary sliding, which is absent in assolidified solder joints. Fatigue cracks initiate at, and propagate along, recrystallized grain boundaries, eventually leading to intergranular fracture. The recrystallization behavior of Sn-Ag-Cu solder joints was examined in three different TMF conditions for five different ball grid array (BGA) component designs. Based on the experimental observations, a thermal fatigue model is proposed: (1) strain enhanced coarsening of secondary precipitates of Ag3Sn and Cu6Sn5 starts at joint corners, eventually allowing for recrystallization of the Sn grain there as well; (2) coarsening and recrystallization continue to develop into the interior of the joints, while fatigue cracks growth lags behind; (3) intergranular fatigue cracks finally progress through the recrystallized region. Independent of the TMF condition the recrystallization appeared to be essentially complete after somewhat less than 50% of the characteristic life while it took another 50-75% of the life time for a fatigue crack to propagate through the recrystallized region.
Authors: Liang Yin and Peter Borgesen
Abstract: Soldering to Cu interconnect pads with Sn-containing alloys usually leads to the formation of a layered Cu3Sn/Cu6Sn5 structure on the pad/solder interface. Frequently microscopic voids within Cu3Sn have been observed to develop during extended thermal aging. This phenomenon, commonly referred to as “Kirkendall voiding”, has been the subject of a number of studies and speculations but so far the root cause has remained unidentified. In the present work, 103 different Cu samples, consisting of 101 commercially electroplated Cu and 2 high-purity wrought Cu samples, were surveyed for voiding propensity. A high temperature anneal of the Cu samples before soldering was seen to significantly reduce the void level in subsequent thermal aging. For several void-prone Cu foils the anneal led to significant pore formation inside the Cu. In the mean time, Cu grain growth in the void-prone foils showed impeded grain boundary mobility. Such behaviors suggested that the root cause for voiding is organic impurities incorporated in the Cu during electroplating, rather than the Kirkendall effect.
Authors: Venkatesh Raghavan, Brian Roggeman, Michael Meilunas and Peter Borgesen
Abstract: The present work addresses a significant risk generally overlooked in the design and accelerated testing of high reliability electronics. Manufacturers of servers and other expensive high reliability electronics equipment are becoming increasingly concerned with the risk of solder pad cratering. Their focus is, however, on cratering in testing, handling or transport, while the risk of premature wear-out due to thermal excursions (cycling) in service is completely ignored. This is a result of reliability testing that almost invariably addresses individual loading conditions separately. Under such conditions it is for example extremely rare for electrical failures in thermal cycling or high-cycle vibration testing to be associated with pad cratering.
Authors: Brian Roggeman and David Rae
Abstract: Pad cratering does not always present itself as a catastrophic failure. Any amount of partial cracking under connecting pads can significantly reduce the reliability of the device. This report discusses the effects of partial pad craters on drop/shock fatigue life.
Authors: Michael Meilunas and Barry Berger
Abstract: An accelerated thermal cycle experiment comparing similarly constructed BGA devices with 10, 12, or 16 mil diameter SAC305 solder balls was performed. Five package designs, containing 64 I/Os at 1.0, 1.2, 1.4, 1.8, or 2.2mm pitch, were subjected to three thermal cycle conditions in order to promote 2nd level solder fatigue. The test results were compared using Weibull analyses. The results show that time to failure is highly influenced by the package pitch and thermal cycle condition in a manner predicted by simple mechanics. However, there were instances in which the effect of solder ball size did not fit the traditional solder joint reliability model in which increasing standoff height increases the number of cycles necessary to produce failure. A theory is proposed that substantial differences in SAC305 solder joint microstructures may explain, at least partially, the discrepancies and evidence to support this theory is presented.
Author: Michael Meilunas
Abstract: 0.3mm pitch components have arrived. These packages offer substantial functionality in very small form factors but they must be integrated into new products carefully. This report describes many of the potential issues with 0.3mm pitch integration and an initial experiment performed to address several assembly and reliability issues.
Authors: Michael Meilunas and Brian Roggeman
Abstract: A wafer level solder supporting material was applied over a silicon wafer containing an array of 5.4x5.4mm WLCSPs using an expose and etch process. The packages were saw singulated, assembled to printed circuit boards, and thermally cycled between -40 and 125°C or drop tested. WLCSPs containing the pre-applied underfill showed significant improvements in thermal cycle reliability when compared to uncoated WLCSPs. Results for the drop tests were more complex. In these experiments, preapplied underfill samples could perform significantly better in harsh drop, but worse in milder drop conditions than their non-coated counterparts.
Author: Michael Meilunas
Abstract: In 2007 the Unovis Solutions’ Advanced Process Lab asked AREA Consortium members to suggest lead-free surface mount devices for the upcoming 2008-2009 thermal cycle reliability assessment program. Fifteen package designs were provided and copper OSP and ENIG motherboards were designed to accommodate the samples. The devices were assembled and testing began in March of 2008. The following report documents the basic assembly processes, test procedures and results through January 2011 (when testing was terminated). In general, most of the devices performed as well as could be expected for their package type. Of particular note is that several infant mortalities were detected which would most likely escape common screening tests and that some assemblies may have failed “mechanically” but the assembly remained “electrically” good.
Author: Pericles A. Kondos
Abstract: Three different types of an ENEPIG substrate, identical in design but from different manufacturers, were populated with SAC, SnAg, and SnPb solder spheres and used in ball shear strength and microstructure studies involving samples aged for various lengths of time at an elevated temperature. One purpose of these experiments was to verify the claim by various ENEPIG chemistry suppliers that this finish, used with lead-free solder, does not have the reliability issues it has been known to have with SnPb. While this claim appeared indeed to be true for SAC and in most cases for SnAg as well, the behavior of the latter indicated that under some conditions it might develop the weak structure normally seen in aged SnPb joints. Observations in a sample from an APL customer project raised this possibility even for Cu-containing solder. One of the substrate types had some features usually associated with black pad, to which ENEPIG is supposed to be immune.
Author: Pericles A. Kondos
Abstrac: In order to determine the reflow profile with the lowest Tmax that is required to accomplish good soldering in parts with low-Ag bumps, samples were made with several alloy compositions on three different metallizations. A variety of tests were performed in order to determine which profile had given solder bumps with properties similar to those of bumps formed with a “definitely hot enough” reflow. These properties included microstructure and performance in ball shear and ball shear fatigue tests. In addition, one of these alloys was used to make parts that were assembled on PCB’s with a range of reflow profiles and then subjected to accelerated thermal cycling. While none of these tests, viewed separately, gave a conclusive answer to the original question, taken together and combined with last year’s results they all pointed to the same “good enough” profile.
Author: Liang Yin
Abstract: The report presents a case study of creep corrosion of a PCB assembly from field return. Two identical immersion Ag boards showed distinct creep corrosion behaviors. The board showing extensive creep corrosion had more Cu on the via surface, suggesting a thin Ag coverage might contribute to the migration of Cu2S on the solder mask. On the same board, surface contamination was seen as well. A brief literature review of recent studies is also presented in this report. Apparently there are three necessary conditions for creep corrosion to occur. However, the mechanism of creeping product migration on solder mask or FR-4 surface remains a poorly understood phenomenon.