2009 Reports

Effects of Corner/Edge Bonding and Underfill Properties on the Thermal Cycling Performance of Lead Free BGA Assemblies
AuthorsD. Blass and M. Meilunas
Abstract: Underfilling will almost certainly improve the performance of an area array assembly in drop, vibration, etc. However, depending on the selection of materials the thermal fatigue life may easily end up worse than without an underfill. This is even more true for lead free than for eutectic SnPb soldered assemblies. If reworkability is required the bonding of the corners or a larger part of the component edges to the PCB, without making contact with the solder joints, may offer a more attractive materials selection. A 30 mm FCBGA component with SAC305 solder balls was attached to a PCB and tested in thermal cycling with underfills and corner/edge bonding reinforcements. Two corner bond materials and 6 reworkable and nonreworkable underfills with a variety of mechanical properties were considered. All of the present underfills reduced the thermal cycling performance, while edge bonding improved it by up to 50%. One set of theFCBGAs was assembled with a SnPb paste and underfilled with a soft reworkable underfill. Surprisingly, this improved the thermal cycling performance slightly beyond that of the non-underfilled assemblies, providing for an up to three times better life than for those assembled with a SAC305 paste. 

Authors: E. Al-Momani, M. Meilunas, P. Borgesen and M. Khasawneh
Abstract: For many high end microelectronics products the long term service life may be limited by thermal mismatch induced fatigue, a factor generally assessed by accelerated thermal cycling and extrapolation of the test results to service conditions. For this purpose, thermal cycling tests are accelerated strongly by the use of larger temperature ranges, shorter dwell times at the maximum and minimum temperatures, and often higher ramp rates than encountered in service. The sensitivities of common lead free solder joints to these factors and their interactions with a range of design, materials and process parameters are very different from those of the SnPb joints we have been using for decades, and systematic research efforts are currently aimed at the establishment of new test protocols and acceleration factors. Such work has already shown lead free solder joints to be much more sensitive to the dwell times in testing, but extrapolations of this dependence to service conditions remain complicated by interactions with design 
parameters, as well as with the more obvious effects of cycling parameters.

Neural network analysis offers important advantages for the recognition of patterns, correlations and trends among empirical accelerated test results, but it usually does not lead to the kind of mechanistic understanding required for extrapolations. Nevertheless, as part of a larger effort to develop tools for the ongoing analysis and generalization of major databases, a systematic trend was identified that should come in useful in life assessment. This is particularly important because it contradicts predictions, based on the widely used Engelmaier model, of a more problematic behavior. In short, we shall suggest a practical approach to the assessment of the effect of the time at operating temperature on the life of lead free solder joints in service.

Effects of Solder Paste Volume and Reflow Profiles on the Thermal Cycling Performance of Mixed SnAgCu/SnPb Solder Joints 
Author: P. Borgesen
Abstract: Manufacturers of most high liability microelectronics products are still exempt from the requirement to eliminate Pb from their soldering process, and neither their components nor their printed circuit boards can always survive a lead free reflow profile. When faced with the need to incorporate a lead free BGA or CSP into their product they are thus faced with three alternatives – attaching it separately using a rework station, reballing the component with eutectic solder or soldering it with a eutectic SnPb solder paste. Either of these may affect the final reliability of the assembly in its own way and comparisons between them by accelerated testing are confounded by very different acceleration factors. Optimization of solder paste volumes and reflow profiles for mixed soldering is of course less affected by this, but common experiences have been that optimized mixing requires a typical lead free reflow profile. This clearly defeats the purpose of mixing in many cases. 

Two different sets of experiments were conducted to assess the consequences of soldering SnAgCu balls with a eutectic SnPb paste for the thermal fatigue resistance of the resulting solder joints. The first experiment addressed effects of solder paste volume and common reflow profile alternatives. Based on the results of this, the second experiment involved the development of an improved low temperature profile and the comparison of this to common SnPb and lead free profiles in terms of thermal cycling performance. 5 different high and moderate strain area array components were tested with SnPb, SnAgCu and mixed SnAgCu/SnPb solder joints and the results discussed on the basis of known test parameter dependencies and trends. The improved profile was shown to optimize the performance of the mixed joints, but indications are these may still compare poorly to pure eutectic ones for some components. All five components may have shorter lives with lead free than with SnPb joints in service.

Effects of Cycling Parameters on the Thermal Fatigue Life of Mixed SnAgCu/SnPb Solder Joints 
Author: M. Meilunas
Abstract: The mandated switch of the overwhelming part of microelectronics assembly to lead free soldering has left the manufacturers of many high reliability products, which are still exempt from the requirement to change, with a choice between imperfect alternatives. One of these involves the inclusion of lead free BGAs and CSPs in a eutectic SnPb solder paste based reflow process. In order to assess the consequences of this for reliability it is not enough to subject samples to common accelerated tests. Mixing of the alloys is almost certain to affect the acceleration factors so that comparisons of test results may not reflect relative performances under service conditions. This was illustrated with for a set of model BGA components with 20 mil (0.5mm) diameter SAC305 solder balls which were reflow soldered onto printed circuit boards using a eutectic SnPb solder paste and tested in thermal cycling together with controls soldered with a SAC305 paste. Mixed joints were found to outperform pure SAC305 in -40/125oC tests but not in 0/100oC, and general trends suggest that mixed joints may compare more poorly in service.

Interactions Between Flip Chip Underfill and Solder Alloy 
Author: D. Blass
Abstract: The effects of underfill selection on flip chip reliability were always a complex issue. Mechanical optimization of the underfill performance, achieved by the addition of appropriate fillers, is invariably a trade-off between the adhesion and the coefficient of thermal expansion (CTE) and thus also between in- and out-of-plane stresses. Another critical concern is the degradation of the underfill in processing and/or long term exposure to operating temperatures and ambient humidity. This is strongly affected by the chemical compatibility with combinations of solder mask, chip passivation, and flux residues. The latter is believed to be responsible for our observation of interactions with the solder alloy too. As for the effects of glass transition temperatures and CTE we find materials that were close to optimum for eutectic SnPb to be very far from the best options for lead free joints.

We report on two sets of systematic experiments. The first addressed the performance of combinations of underfills, no-clean fluxes and solder alloys in a JEDEC Level 3 moisture sensitivity test. The second one involved thermal shock testing of flip chip assemblies underfilled with one of five different materials after soldering with SnCu, SAC305, and SnPb.

On the Incorporation of Fine Pitch Lead Free CSPs in High Reliability SnPb Based Microelectronics Assemblies 
Authors: D. Blass and M. Meilunas
AbstractThe thermal cycling performance of lead free soldered BGAs and CSPs soldered with eutectic SnPb solder paste, so-called backward compatible assemblies, tends to drop with increasing Pb concentration in the joints. This is a particular concern for fine pitch CSPs where concentrations are invariably rather large. Flip chip assembly life is almost always improved by underfilling, although lead free soldered assemblies are more sensitive to the underfill material properties. In the case of BGAs and CSPs, however, underfilling with the wrong material may actually reduce life in thermal cycling substantially. In general, underfills with low coefficients of thermal expansion (CTE) and high glass transition temperatures (Tg) are preferred, especially for lead free solder joints. However, recent work showed a reworkable underfill with a high CTE and a very low Tg to still improve the life of SnAgCu joints with 3% Pb mixed in. 

A variety of 0.4 and 0.5mm pitch CSPs with SnPb, SAC305 or SAC405 solder bumps were assembled onto a six layer PCB with a SnPb solder paste. Assemblies were tested in 0/100oC thermal cycling both with and without a reworkable underfill with a reasonably high modulus and a Tg of 105oC. Results deviated systematically from trends previously observed for larger pitch BGAs.

Quantitative Assessment of Long Term Aging Effects on the Mechanical Properties of Lead Free Solder Joints 
Author: Vikram Venkatadri
Abstract: A limited number of research experiments in the literature address the effects of preconditioning (accelerated aging) on the performance of the solder joints in subsequent testing. But they are all quite empirical and limited to the interpretation of test results without offering a quantification of the acceleration. Therefore the focus of the present research is to:

1. Determine whether it is possible to offer a quantification of the acceleration of aging on the reliability of old lead free solder joints (i.e., how old a solder joint was simulated?).

2. Contribute to a fundamental understanding of the effects of preconditioning (isothermal aging) on the reliability of lead free solder joints, and thus to the establishment of a means of assessing the reliability of old lead free solder joints.

Pad Cratering: Characterizing Crack Propagation and the Effects of Humidity and Reflow on Reliability
Author: Gaurav Vinod Godbole
Abstract: Concern about the phenomenon of pad cratering, which is still largely misunderstood, has seen increased attention from the industry in the recent years. Pad cratering is the loss of laminate integrity under the pad leading to a “crater” under the pad which eventually leads to trace failure. Pad cratering resistance may be affected by a multitude of factors including inherent properties ascribed to pad design, trace routing, material selection, position of pads on the glass weave and distance from neutral point. It may also be affected by loading conditions and degradation mechanisms induced, among other factors, by the presence of moisture and thermal excursions.

The present work explores the impact of moisture, thermal excursions on pad performance and to an extent of loading conditions. The extension of joint level pad reliability to board level reliability has eluded researchers for a while now and the present work offers insights on how this may be due to different loading conditions and absence board level dynamics in the joint level tests. Addressing the issue of pad cratering in a comprehensive manner and correlating joint level tests with board level tests is beyond the scope of this research and would require a much larger concerted effort. Another consequence of this research has been the development of a basic understanding of the mechanics of crack propagation under the pad.

Analyzing Thermal Cycle Parameters with “Low Reliability” Packages – 2009 Update 
Author: Michael Meilunas, Emad Al-Momani
Abstract: Understanding how accelerated thermal cycle parameters such as temperature extremes and dwell time affect the reliability of a lead-free solder joint is an important focus of the AREA Consortium and, in general, the effects of cycle parameters are now know. However a method is desired which can be used to scale between various thermal cycle parameters so that a meaningful analysis of device reliability can be performed without actually testing every cycle parameter. The following is not a discussion on the scaling technique itself, but rather a full documentation of the various accelerated test results used to create and refine the technique.

Evaluating Partial Wafer-Level CSP Underfill 
Authors: Michael Meilunas, Russ Stapleton
Abstract: A wafer level solder supporting material was jet dispensed onto a silicon wafer containing an array of 5.4x5.4mm WLCSPs. The packages were saw singulated, assembled to printed circuit boards, and thermally cycled between -40 and 125°C. WLCSPs coated on the ball side with the jetted and cured solder supporting material showed significant improvements in thermal cycle reliability when compared to uncoated die with a characteristic life improvement of nearly 2X. Also, the jetting process was found to be a simplified method for applying and evaluating materials in the laboratory setting, having a good die yield of 64-84%. A version of this report has been previously published as “Reliability of 400 μm Pitch Wafer Level CSP Assemblies with Solder Supporting Material” at the International Wafer Level Packaging Conference, October 2009.

Reliability Testing of Flexible Surface Mount Capacitors 
Author: Michael Meilunas 
Abstract: The following report discusses a simple experiment designed to compare surface mount ceramic chip capacitors with “flexible” terminations to standard ceramic chip capacitors in accelerated thermal cycle testing. Flexible terminations are designed to mitigate flex cracks which are the result of shear stresses produced by board flexure. Flexible terminations are not intended to improve thermal cycle performance, however the AP Lab wanted to ensure that the thermal cycle performance of the flexible capacitors was comparable to standard capacitors. 

Lead Free Land Grid Array (LGA) Review 
Author: Michael Meilunas
Abstract: Studies conducted by Universal Instruments’ Advanced Process Lab regarding the reliability of Land Grid Array packages have produced some very unexpected results. In short, it has been observed that lead-free Land Grid Array solder joints are capable of surviving as many, and sometimes more, accelerated thermal cycles than identical lead-free Ball Grid Array solder joints. The reason for such behavior is being attributed to the fact that a single solder alloy with a high tin concentration is capable of forming very different microstructures depending upon the volume of solder applied.

Thermal Cycling of OFNs on Thick Test Boards 
Author: Michael Meilunas
Abstract: This report describes the printed circuit board, materials, assembly processes, test procedures and results obtained with 13 leadless surface mount package designs in -40 to 125°C thermal cycling on 2.2mm thick motherboards with Cu OSP and ENIG surface finishes.

Thermal Cycle Reliability Assessment of Surface Mount Devices on Test Board 2007 
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 December 2009. Testing continues as of this writing and updated reports will be issued as necessary

Thermal Cycling and Drop Testing of Leadless Packages Located on “QFN Process Board 2007” 
Author: Michael Meilunas
Abstract: This report describes the printed circuit board, materials, assembly processes and test procedures used to evaluate 14 leadless surface mount package designs in drop tests and thermal cycling on a 1.0mm thick motherboard. Overall, most of the packages evaluated are quite robust and perform as well as or better than similarly sized chip scale packages.

Electroless Nickel – Electroless Palladium – Immersion Gold (ENEPIG): A “Universal” Finish?
Author: Pericles A. Kondos
Abstract: A project studying the behavior of this PCB finish has started in the APL. The published literature on this topic has been surveyed and found to be inadequate with results often inconsistent and contradictory. Various samples of ENEPIG boards have been acquired and are in various stages of being studied. Issues under investigation include variations in Pd thickness (and their effect on joint properties), mechanical performance of joints, and nature/structure of pad intermetallics. Future work will expand on these and include the behavior of low-volume high-area joints (QFN, LGA etc) where the average concentration of Pd in the solder will be much higher than in a typical BGA ball.

Brittle Failures of Joints on Electrolytic Ni/Au 
Author: Pericles A. Kondos
Abstract: Experiments were continued on SAC or SnAg solder bumps (with or without Cu dust added) on electrolytic Ni/Au pads. The bumps were subjected to a variety of tests, mostly cold bump pull, under conditions that caused many or all of them to fail with brittle failures in the intermetallic region. The purpose of these tests was to determine if substrates with bad or suspect nickel produced consistently weaker intermetallics than samples with typical nickel. The effects of ball-attach and/or Cu-addition profiles were investigated. The failure surfaces and the structure of the intermetallics were studied as well. Many of the results, especially those involving microstructure studies, were in agreement with older experiments, but inconsistent behavior was seen in some cases, raising doubts about the ability of some of the tests to distinguish between nickel of different qualities. Other tests are still under evaluation.

Low-Ag Pb-Free Alloys Solderability Studies 
Author: Pericles A. Kondos
Abstract: In order to determine the reflow profile with the lowest Tmax and TAL that is required to accomplish good soldering in parts with low-Ag bumps, a "marker" property was sought that would show when the properties of the solder joint become similar to those of a joint that has been formed with a "definitely hot enough" reflow. These properties included microhardness, microstructure, and performance in periodic impact and shear fatigue tests. Samples were made with several alloy compositions on three different metallizations. Promising differences were seen in some cases, but none of the properties tried could be universally used for all combinations. Reliability tests with actual assemblies are planned. 

Microstructure of SAC105 Solder Alloy 
Author: Liang Yin
Abstract: SAC105 based alloys have been attractive for mobile applications requiring high impact/drop reliability. In this report, 30 mil SAC105, SAC305 and SAC405 balls were soldered to Cu-OSP pads and compared microstructurally. SAC105 bumps showed pronounced phase separation, in which Ag3Sn and Cu6Sn5 precipitates were seen to be included in β-Sn matrix as separated phases. For SAC305 and SAC405 bumps, the two precipitates coexisted in the ternary phase. Thermal analysis of SAC305 and SAC105 spheres by DSC (differential scanning calorimetry) showed different melting characteristic, which was consistent with the phase separation behavior in SAC105.

Microstructure Evolution in Accelerated Thermal Cycling of Pb-Free Solder Joints 
Author: Liang Yin
Abstract: Accelerated thermal cycling (ATC) test is often used to evaluate assembly-level solder joint reliability by simulating service temperature variation in an accelerated fashion. Sometimes the test is used for qualification, while other times the test results are utilized to predict service life by using Coffin-Mason or Engelmaier models. After typical ATC testing Pb-free solder alloys are often seen to undergo recrystallization of β-Sn grains. In this report, evolution of recrystallization was investigated in 0-100°C thermal cycling tests. Two dwell times and five pitch sizes were used. It was found that global recrystallization occurred at cycle numbers approximately 25-50% of the average fatigue life, independent of dwell time and pitch size. Fatigue crack propagation appeared mostly to be transgranular, indicating that crack growth was facilitated by dynamic recrystallization.

Fatigue Damage Evolution in Pb-Free Solder Alloys 
Author: Liang Yin
Abstract: The report presents a literature survey of damage mechanisms in Pb-free solder alloys during various fatigue tests. The focus is on cyclic softening, an irreversible change of stress-strain behavior induced by cyclic straining/stressing. In mechanical fatigue tests, grain boundary cracking apparently attribute to cyclic softening. In thermo-mechanical tests, dynamic recrystallization and strain-enhanced coarsening could introduce additional softening of solder joints. Experimental results from Consortium studies are also reviewed and compared. 

Recrystallization of Pb-Free Solder Alloys in Mechanical Fatigue Testing 
Author: Liang Yin
Abstract: In accelerated thermal cycling (ATC) test recrystallization is often observed to facilitate the growth of intergranular fatigue cracks. In this report, the focus is on isothermal mechanical fatigue tests. Recrystallization behavior of Pb-free solder joints was studied in room temperature bending and vibration tests after electrical failures had occurred. In vibration test minimal level of recrystallization was observed with mostly transgranular cracks. In bending test some level of recrystallization was observed with occasional intergranular cracks. The relatively low level of recrystallization in mechanical fatigue testing indicates that crack propagation might well be different from that in ATC test. 

Recrystallization of Pb-Free Flip Chip Solder Joints 
Author: Liang Yin
Abstract: The report presents an investigation of the recrystallization behavior of lead-free flip chip solder joints during liquid-toliquid thermal shock (LLTS) testing. The focuses are the grain structure evolution of solder joints and its connection to fatigue crack propagation. LLTS of 2500 cycles led to noticeable recrystallization throughout the joints. Fatigue cracks were seen to be intergranular, transgranular and along the interface between large Ag3Sn or Cu6Sn5 pallets and Sn matrix. The large IMC pallets were resulted from large undercooling associated with the small solder volume. 

Recrystallization of Lead-free Solder Joints – Confounding the Interpretation of Accelerated Thermal Cycling Results?
Authors: Ahmad Mayyas, Liang Yin and Peter Borgesen
Abstract: A primary concern of any reliability engineer is whether failure in accelerated testing may be caused by a different damage mechanism than the one that determines life in service. The recent suggestion that thermal expansion mismatch induced fatigue and failure of lead free solder joints under realistic service conditions may progress without the recrystallization seen to play a major role in accelerated thermal cycling thus requires systematic investigation. 

A total different concern is the ever recurring question of to which extent thermal cycling can be replaced by much faster isothermal cycling tests. The observation that the damage mechanisms tend to be completely different, one leading to transgranular cracking the other to intergranular cracking of lead free BGA solder joints, would seem to invalidate such ideas completely. Importantly, most practical service scenarios involve the combination of more than one loading conditions. This may be as simple as thermal excursions to different temperatures or vibrations with two different frequencies or amplitudes, but it may also involve a combination of thermal excursions and vibration, for example. Reliability prediction almost invariably relies on the explicit or implicit assumption of some principle of damage accumulation, most commonly Miner’s rule. These assumptions do, however, break down much more often than commonly recognized. One important reason for this, albeit clearly not the only one, would be the different effects of recrystallization. 

All of these concerns prompted an ongoing investigation to identify the parameters determining whether recrystallization will affect lead free solder fatigue life and what would be the practical consequences. 

Reliability Investigation of Sn/Cu/Ni Solder Joints 
Authors: Brian Roggeman, Ursuala Marquez de Tino and Denis Barbini
AbstractThe electronics manufacturing industry has continued its search for a single alloy that can be utilized throughout the manufacturing line. To date, there is minimum success in the implementation of an alloy across the manufacturing process as compared to the tin lead based processes. The ability to employ one alloy will allow manufacturers to reduce complexity and cost while increasing yield. 

The Sn/Cu/Ni (SCN) alloy has been widely used in wave solder applications. This report continues the investigation of SCN in SMT assembly, extending the scope to use in handheld/portable devices. The drop/shock reliability was tested under standard JEDEC conditions, and compared against SAC305 and SAC105. Results indicate that SCN joints provided the equivalent reliability to SAC305, but not as good as SAC105 in a shock environment. 

Comparison of Joint Level Test Methods for PCB Pad Cratering Investigations 
Author: Brian Roggeman
Abstract: The introduction of less compliant lead free solders together with weaker and more brittle laminate materials has led to major concerns with respect to the resistance to pad cratering of printed circuit boards. For purposes of laminate selection as well as for the quantification of acceptable handling and use conditions there is a general interest in the testing for cratering at the joint level, rather than by testing entire assemblies. Joint level testing is cheaper, faster, and offers more quantitative results. Joint level testing also allows the elimination of confounding factors such as PCB and component stiffness. 

In developing test procedures and performing the test itself, it is important to distinguish between wear out under repeated loads, failure due to a single overstress or even failure due to a static load. Focusing on strength testing we have compared alternative methods and identified the most relevant approaches. Occasions may arise where shear testing, vertical pull or angled pull may be appropriate, however these may not correlate with each other and rankings of materials may be impacted. We recommend selecting the most relevant test procedure and taking into account other factors such as loading rate and temperature effects if these are significant to the end product. 
Author: Brian Roggeman
Abstract: Industries with an exemption from RoHS are facing supply chain issues with SnPb devices. One possible solution is to re-ball PbFree devices, or even procure un-balled LGA devices and ball them with SnPb. In doing so, the consistency of the balling process and integrity of the device must be considered. This report examines the integrity of 0.5mm pitch BGA devices that have been balled by 3rd party suppliers. Two test cells have had their Pb-Free solder balls removed and SnPb solder balls attached, while the rest of the test cells were balled on LGA style packages. Several metrics were used in the investigation, including solder ball dimensions, shear strength, component warpage and delamination.

Reliability of SnPb and PbFree BGAs on Various Surface Finish PCBs in Random Vibration 
Author: Brian Roggeman
Abstract: Random vibration is used to investigate the mechanical reliability of various conditions of electronic assemblies. Seven different surface mount devices were assembled to test cards that used four different surface finishes. SnPb, PbFree and mixed (backward compatible) technologies were used. The results indicate that only the largest BGA components provided verifiable failures during the test, likely due to the low stressing amplitudes induced by random vibration. Parylene coating was applied to each test cell and showed a potential improvement in lifetime, except for the case of PbFree on Electrolytic Ni/Au. In that case, the Ni is used as an etch resist and creates an overhang on the copper pad, preventing wetting of the solder ball to the sides of the pad. 

Stencil Design for 0.4mm CSP: Prelude to Drop Testing
Author: Jeff Schake and Brian Roggeman
Abstract: Future handheld electronic products will be slimmer than today and deliver more functions, enabled by innovative electronics packaging design using smaller components with greater I/Os assembled in higher density. As solder interconnects between component and circuit board shorten, they also become weaker. This causes us greater concern on the survivability of such delicate electronic interconnects under normal handling impacts and serves motivation for formal study. This investigation will evaluate the influence of stencil printed solder volume on CVBGA97 electronic component lifetime in mechanical stress testing. A stencil aperture design to print a lower limit of solder paste volume has been thoroughly characterized as the first step towards determining the range of print volumes exhibiting the greatest influence on drop, bend, and die shear test performance. In this printing focused piece of work, print volume measurements were found varied across different circuit board pad designs with no change in aperture size. Highest paste volume transfer consistently occurred with solder mask defined pads. Stencil aperture and circuit board pad design variables are discussed in detail.