Search results

Presents the results of a study of the effects of solder ball pad metallurgy, intermetallic compound (IMC) thickness and thermal cycling on the shear strengths of PBGA package solder balls. The study of the microstructures of solder balls revealed that only a very thin layer of intermetallic compound existed between solder balls and Ni or Ni alloy barrier layers immediately after ball placement and reflow. The protective Au layer was dissolved completely and a needle like AuSn₄ intermetallic compound was then formed and dispersed evenly in the solder balls. The overall thickness of the IMC layers was thicker than 15μm after storage at 150°C for 1,000 hours. During the shear tests failure occurred at the interface of the two IMC layers. The fracture surfaces of solder balls with electrolytic Ni and thick Au layers were smooth and brittle fracture was observed. The ball shear strength decreased dramatically with the formation of IMC layers. For the solder balls with electroless Ni and thin Au layers, only a single IMC layer was formed at the interface and its thickness was only 2.5 μm after storage at 150°C for 1,000 hours.

The purpose of this paper is to visualise the activities of three solders; Sn‐37Pb, Sn‐9Zn and Sn‐3.5Ag on Cu substrates during reflow near their melting points and to relate them with reflow reactions between solder and substrate.

Melting activities of three solders near their melting points on copper substrates are visualised in an infrared reflow furnace.

Solder balls demonstrate different ways of melting and reflowing behaviours in dissimilar times and temperature intervals. Melting of Sn‐9Zn solder balls is initiated simultaneously at the surface and joint between solder balls. This is followed by the melting at the joint between solder balls and the Cu substrate. During melting, solder balls are first merged into each other and then reflow on the substrate from top to bottom. Opposite to Sn‐9Zn, Sn‐3.5Ag solder balls start to melt at the surface and the joint between the solder and substrate, simultaneously. Balls are first reflowed from top to bottom and, in the process, liquid solder is merged. Unlike Sn‐9Zn and Sn‐3.5 Ag, melting of Sn‐37Pb solder balls is initially commenced at the surface only. This is followed by simultaneous melting at both joints. Variation in melting activities of these solders is found to be closely related to the coalescence mechanism of solder balls and the reflow reactions between the solders and the Cu substrate.

The elementary melting activities of different solders on Cu substrates is related with their reflow behaviours. This provides better understanding of solder behaviour and selection of good lead‐free solder for applications in the electronic industry.

The purpose of this paper is to describe the board level reliability test results of four IC packages with lead‐free balls/platings, soldered with lead‐free solder paste, during thermal cycling. The board level reliability test results of tin‐lead balled/plated packages soldered with lead‐free solder paste have also been included for comparison.

Four different packages, i.e. ball grid array (BGA), chip scale package (CSP), quad flat package (QFP) and thin small outline package (TSOP), were assembled on a test printed circuit board (PCB) as the test vehicle. Lead‐free and tin‐lead BGA/CSP packages were equipped with Sn‐3.0Ag‐0.5Cu and Sn‐37Pb solder balls, respectively. The lead‐frames of lead‐free QFP/TSOP leaded‐packages were plated with Sn‐58Bi and those of tin‐lead QFP/TSOP leaded‐packages, Sn‐37Pb. The lead‐free solder paste used in this study was Sn‐3.0Ag‐0.5Cu. Two kinds of surface finishes, immersion gold over electroless nickel (Au/Ni) and organic solderability preservative, were used on the PCBs. The test PCBs were thermal cycled 5,000 times within the temperature range of −40 to 125°C and electrically monitored during the thermal cycling.

It was found that the tin‐lead balled/plated BGAs, CSPs, QFPs and TSOPs soldered with lead‐free solder paste showed serious board level reliability risks as their abilities to withstand thermal cycling stresses are much weaker than those of entirely lead‐free assemblies. Neither package nor surface finish was found to have any effects on the board level reliability of test vehicles with lead‐free balled/plated BGAs, CSPs, QFPs and TSOPs. Metallographic examinations were conducted to investigate the effect of thermal cycling on the failure modes of solder joints.

The paper is of value by contributing to research in the use of lead‐free solder paste with lead‐containing packages in the industry. Currently, there is a deficiency of knowledge in this area.

To meet the state‐of‐the‐art requirements of BGA assemblies necessitates direct coupling of field conditions, simulation tools for life‐time study and advanced experiments for the assessment of physical degradation. For conventionally soldered SMD components, transformations between test and field conditions are still not completely known. For new types of array components, the answers critically depend upon ‘Component age’ and change in fatigue mechanisms. The increasing complexity of microelectronic assemblies and the hidden joints of BGAs lead to an increase in reliability problems in this field. Therefore, to describe failure‐free times for different applications, fatigue relevant parameters of the ball solder joints need to be studied. With regard to the thermal coefficient of expansion, BGAs are mainly asymmetrical, consequently residual strains and stresses are generated in the solder joint array. The level of strains and stresses depends upon the global and local mismatch, the applied operating conditions and the temperature distribution in the ball solder joint array (chip location, ambient and operating temperature). For thermo‐mechanical cycling procedures, hold and ramp times at upper and lower temperatures (e.g., −20°C/+100°C) are used to initiate strains in materials and interfaces. BGAs and PCBs show comparable thermal levels with regards to the test procedures mentioned before, and the resulting stress conditions in the ball solder joints are a function of package size, DNP, etc. The test results with regard to the generation of cracks are not directly comparable to the fatigue behaviour under operating conditions. Therefore, different types of degradation tests were developed: thermo‐mechanical, mechanical, electrical and/or corrosive procedures. Depending upon the chip location in the BGA package (symmetrically: PBGA, TBGA, CBGA; asymmetrically :MCM‐BGA) frequencies, lateral and vertical temperature distribution under simulated power dissipations, and the internally generated heat will be used to induce stresses in the ball solder joints. For different values of power dissipation and ambient conditions, thermal measurements were performed, screening the top to the bottom side of the BGA and the array field. The resulting information is a precondition in order to define power cycle parameters. For different test procedures, locations of defects, crack initiation and growth in ball solder joints were studied by metallographic analysis. The practical measurements serve as analytical input to compare thermal and power cycle tests and they are a necessary step to perform a lifetime prediction.

The goal of this work is to evaluate the feasibility of soldering tin‐silver‐copper balled BGAs using tin‐lead‐based solder and to investigate the influence of different production parameters on the microstructure of the solder joint.

The soldering of the BGAs was done with various temperature profiles and two conveyor speeds under a nitrogen atmosphere in a full convection oven. One specimen from each temperature/time combination was cross‐sectioned. The cross sections were analysed with optical microscopy, scanning electron microscopy with energy dispersive X‐ray spectroscopy (SEM/EDS) at 30 kV and focused ion beam microscopy (FIB).

The cross sections show a metallurgical bond between the solder and the tin‐silver‐copper balls of the BGA, even at a peak reflow temperature of 210°C. However, the balls alloy only partially with the solder, as the liquidus of tin‐silver‐copper balls is 217°C. As soon as the peak temperature exceeds the liquidus of the ball, the solder is totally dissolved in the material of the ball. A reflow profile with a peak temperature of about 230°C on the BGA gives a homogenous reaction of the solder with the ball with a minimum formation of voids.

The dependence of varying reflow parameters on reliability requires detailed study. Especially the effect of a partially melted ball on the degradation of the solder joint needs to be investigated.

From the findings, it can be said that soldering lead‐free balls with tin‐lead solder is possible. This is useful during the transitional period that the industry is in at the moment. More and more component manufacturers are changing their components to lead‐free, often without notice to the customer. If a production line is still running a tin‐lead process it is essential to know how to process these components with tin‐lead solder.

The purpose of this paper is to study the variation of the mechanical strength and failure modes of solder balls with reducing diameters under conditions of multiple reflows.

The solder balls with diameters from 250 to 760 µm were mounted on the copper-clad laminate by 1-5 reflows. The strength of the solder balls was tested by the single ball shear test and pull test, respectively. The failure modes of tested samples were identified by combing morphologies of fracture surfaces and force-displacement curves. The stresses were revealed and the failure explanations were assisted by the finite element analysis for the shear test of single solder ball.

The average strength of a smaller solder ball (e.g. 250 µm in diameter) is higher than that of a larger one (e.g. 760 µm in diameter). The strength of smaller solder balls is more highly variable with multiple reflows than larger diameters balls, where the strength increased mostly with the number of reflows. According to load-displacement curves or fracture surface morphologies, the failure modes of solder ball in the shear and pull tests can be categorized into three kinds.

The strength of solder balls will not deteriorate when the diameter of solder ball is decreased with a reflow, but a smaller solder ball has a higher failure risk after multiple reflows. The failure modes for shear and pull tests can be identified quickly by the combination of force-displacement curves and the morphologies of fracture surfaces.

A computational parametric study on the solder joint reliability of a plastic ball grid array (PBGA) with solder bumped flip chip (FC) is presented. The basic configuration of the PBGA is 27mm package‐size and 1.27mm ball‐pitch. There were three kinds of ball population: four‐row perimeter grid array with/without thermal balls, and full grid array. A total number of 24 cases, involving various chip sizes, chip thicknesses and substrate thicknesses, were studied. The diagonal cross‐section of the PBGA‐printed circuit board (PCB) assembly was modeled by plane‐strain elements and was subjected to uniform thermal loading. Through mismatch of coefficient of thermal expansion (CTE), and lack of structural compliance, the solder joints were stressed to produce inelastic deformation. The accumulated effective plastic strain was evaluated as an index for the reliability of solder joints. The present study revealed the effects of aforementioned design parameters on the solder joint reliability of FC‐PBGA assemblies. Some peculiar phenomena were identified.

Computational stress analysis was performed in this study to investigate the solder joint reliability of plastic ball grid array (PBGA) packages with various configurations. The packages under investigation were 27 mm body‐size, 1.27 mm ball‐pitch, perimeter PBGAs with and without thermal balls at the centre. The diagonal cross‐section of the PBGA‐printed circuit board (PCB) assembly was modelled by plane‐strain elements. The model was subjected to a uniform thermal loading and the solder joints were stressed due to the mismatch of coefficient of thermal expansion (CTE). A total number of 24 cases, involving different solder ball populations, chip sizes, and substrate thicknesses, were studied. The accumulated effective plastic strain was evaluated as an index for the reliability of solder joints. The results of this study revealed the effects of the aforementioned parameters on the solder joint reliability of perimeter PBGA assemblies. The findings are very useful for the design of plastic ball grid array packages.

Plastic ball grid array packages were aged for up to 2000 hours. Various solder ball pad metallurgies were studied and solder ball shear tests were conducted at a range of ageing times. The solder ball shear strength was found to decrease after an initial hardening stage. The deterioration of solder ball shear strength was found to be mainly caused by the formation of intermetallic compound layers, together with microstructural coarsening and diffusion related porosity at the interface. For the ball pad metallurgy, two distinct intermetallic compound layer structures were observed to have formed after ageing. Once two continuous intermetallic compound layers formed fracture tended to occur at their interface. For the ball pad metallurgies which do not form two continuous intermetallic compound layers, the shear strength still decreased, due to the coarsening of the microstructure, intermetallic particle formation and diffusion related porosity at the surface of the Ni₃Sn₄.

The purpose of this paper is to assess the effect of different temperature cycling profiles on the reliability of lead‐free 388 plastic ball grid array (PBGA) packages and to deeply understand crack initiation and propagation.

Temperature cycling of Sn‐3.8Ag‐0.7Cu PBGA packages was carried out at two temperature profiles, the first ranging between −55°C and 100°C (TC1) and the second between 0°C and 100°C (TC2). Crack initiation and propagation was analyzed periodically and totally 7,000 cycles were run for TC1 and 14,500 for TC2. Finite element modeling (FEM), for the analysis of strain and stress, was used to corroborate the experimental results.

The paper finds that TC1 had a characteristic life of 5,415 cycles and TC2 of 14,094 cycles, resulting in an acceleration factor of 2.6 between both profiles. Cracks were first visible for TC1, after 2,500 cycles, and only after 4,000 cycles for TC2. The crack propagation rate was faster for TC1 compared to TC2, and faster at the package side compared to the substrate side. The difference in crack propagation rate between the package side and substrate side was much larger for TC1 compared to TC2. Cracks developed first at the package side, and were also larger compared to the substrate side. The Cu tracks on the substrate side affected the crack propagation sites and behaved as SMD. All cracks propagated through the solder and crack propagation was mainly intergranular. Crack propagation was very random and did not follow the distance to neutral point (DNP) theory. FEM corroborated the experimental results, showing both the same critical location of highest creep strain and the independence of DNP.

Such extensive work on the reliability assessment of Pb‐free 388 PBGA packages has never been performed. This work also corroborates the results from other studies showing the difference in behavior between Pb‐free and Pb‐containing alloys.