No description available.

No description available.

No description available.

Flip chip on board assemblies have been widely used in high reliability product applications such as automotive, aerospace and defence industries. However, flip chip solder interconnections are often the weakest link in terms of product reliability. This is mainly due to the high CTE (coefficient of thermal expansion) mismatch deformation between the silicon die and organic substrate during extreme environmental operation that leads to fatigue failures. Thus, this thesis aims to develop a non-destructive and automated monitoring system using ultrasound technology to assess solder joint through-life performance during thermal excursions. The capability and feasibility of using ultrasound technology to achieve non-destructive monitoring was investigated and discussed. An automated ultrasonic inspection and monitoring system on flip chip solder joints was designed to solve the existing issues in capturing the fatigue failures during thermal cycling. Acoustic Micro Irnaging (AMI) techniques have been used along with the accelerated thermal cycling (ATC) test to inspect solder joints due to their strong capability to detect discontinuities within materials and interconnections. However, this approach has not previously been used as an effective tool in monitoring solder joints through life performance throughout environmental testing. The research regime proposed in this thesis was to track the condition of solder bonds monitored through ultrasound images from birth to failure, and to see how their reliability was affected by joint position. This work included accelerated thermal cycling tests to deliberately generate fatigue defects on custom designed test boards, and perform AMI scanning every 8 test cycles over a total period of 96 test cycles. A robust image segmentation and feature extraction system was designed and implemented on the ultrasound images to extract the image features such as histogram difference, bonding area and intensity level that represent the quality of the joint. Through data analysis and classification, the failure occurrence cycles were measured. Finally, a measure of solder joint fatigue life distribution was plotted and compared to a Finite Element prediction. The results have showed that corner joints have the lowest levels of reliability whereas joints closer to the neutral point in the centre of a chip have higher reliability. These results are in accord with the FE prediction and cross-sectional analyses. This matches flip chip failure theory where corner joints are the most critical joints during thermal excursions. AMI monitoring has displayed the defect occurrences in a more realistic manner which encounters real life conditions as compared to FE prediction. This capability allows AMI to non-destructively monitor end user products through life and to be able to reflect real 'in-field' reliability. The monitoring techniques were systematically investigated and verified in this thesis. The results have highlighted the potential of AMI in the non-destructive evaluation of solder joints throughout environmental tests.

The work in this thesis investigates modelling methods to predict the reliability of solder joints under thermo-mechanical cycling. A literature review is presented covering analytical methods, creep laws and fatigue laws, and advanced damage mechanics methods. The use of FEA (Finite Element Analysis) to model creep along with a fatigue law to predict lifetime appears to be the most widely used and validated technique at present. The FEA discretisation of elasticity problems is derived using the principle of minimum potential energy and implemented in the code FATMAN (Finite-element Analysis Tool, Multi-physics And Nonlinear). A novel implicit solution scheme called LENI is proposed to allow modelling of creep in solder. The sinh law for steady-state creep and the Armstrong-Frederick kinematic hardening law to capture primary creep have been implemented in FATMAN using the LENI scheme. The advantage over an explicit discretisation is investigated. An inverse analysis method for determining material properties is used to determine constants for the kinematic hardening law from experimental creep curves. A damage law is presented which allows the prediction of crack propagation through a solder joint. A failure criteria based on the increase in electrical resistance is used, which removes the need for an empirical fatigue law. The steady state creep law, the kinematic hardening law and the damage law are all applied to modelling of tests developed at the NPL (National Physical Laboratory) including novel crack detection tests, an isothermal fatigue test, and accelerated thermal cycling of resistors.

Within the context of the imminent implementation of the Pb-free soldering in Europe in 2006, this thesis addresses the gap in understanding that has emerged in the fundamental materials issues between well-understood and mature lead-containing solders and a plethora of new, Pb-free solders for which there are neither long term reliability data nor understanding of the materials behaviour and how these might be influenced by manufacture and in-service conditions. In addition, this thesis also addresses the question as to whether the solder joint size and geometry could become a reliability issue and therefore affect the implementation of the Pb-free solders in ultrafine micro joints. Thermodynamic calculations using MTDATA (developed by the National Physical Laboratory, NPL, UK) together with a thermodynamic database for solders under either equilibrium or Scheil conditions, have shown their usefulness in Pb-free solder design and processing, generating a wealth of information in respect of the temperature dependence of phase formation and composition. The predictions from MTDATA on a number of selected systems is generally in good agreement with the results from experimental work, and has assisted in the understanding of the microstructure and mechanical properties of the Pb-free solders and the implications of their interactions with a tin-lead solder. However, further critical assessment and the addition of new elements into the solder database, such as Ni and P, are required to make MTDA TA a more effective computational tool to assist the optimization of processing parameters and cost-effective production in using Pb-free solders. Molten solder can interact with the under bump metallizations (UBM) and/or board level metallizations on either side of the solder bump to form intermetallic compounds (IMCs) during solder reflow. In the modelling of the kinetics of the dissolution process of UBM into the liquid solder, the commonly used NernstBrunner (N-B) equation is found to have poor validity for these calculations for micro joints at 100 μm in diameter or less. Three bumping techniques, i.e. solder dipping (SD), solder paste stencil printing followed by reflow (SPR) and electroplating of solders and subsequent reflow (EPR), are used to investigate the interfacial interactions of molten Sn/Sn-rich solders, i.e. pure Sn, Sn-3.5Ag, and Sn-3.8AgO.7Cu, on electroless nickel immersion gold (ENIG) and copper pads at 240°C. The resultant bulk and interfacial microstructures from a variety of pad sizes, ranging from 1 mm down to 25 μm, suggest that in general the small bumps contain smaller β-Sn dendrites and Ag₃Sn IMC particles, nevertheless the interfacial IMC is thicker in the smalI bumps than in the large bumps. In addition, one and two-dimensional combined thermodynamic and kinetic models have been developed to assist the understanding of the kinetics of interdiffusion and the formation of interfacial intermetallic compounds during reflow. Both the experimental results and theoretical predictions suggest that the solder bump size and geometry can influence the as-soldered microstructure, and therefore this factor should be taken into consideration for the design of future reliable ultrafine Ph-free solder joints.

The demands for high density, fine pitch interconnections in electronics systems has seen solder-based approaches for such interconnections miniaturized to the scale of tens of micro meters. At such a small scale, such 'micro joints' may contain only one or a few grains and the resultant mechanical behaviour may not be that for a polycrystalline aggregate, but rather for a single crystal. Since the ~-Sn matrix of SnAgCu solder has a contracted body-centred tetragonal (BCT) structure, such a solder grain is expected to demonstrate a considerably anisotropic behaviour. In such cases the reliability of a Phfree solder is strongly dependent on the local microstructural features, such as the size and orientation of the grains. This thesis presents the investigation of the evolution of microstructure within a joint or at the interface and, the influence of such microstructural features on the meso-scale mechanical behaviour of the Ph-free solder. It includes Evolution of the interface between a molten solder and the Cu substrate To form a joint, the solder alloy is heated and molten, wetting a solid under-bump metallization. After solidification, layers of brittle intermetallic compounds (IMCs) are formed at the interface. In this project, facilities were set up to obtain interfacial reactants at an arbitrary moment of the liquid/solid reaction. Formation and evolution ~ during reflow of SnCu IMCs at the interface between the molten SnAgCu alloy and the Cu UBM was captured and presented for the first time. Formation of phases and IMCs with the body of a liquid SnAgCu solder during solidification The formation behaviour of basic components for a SnAgCu grain (including Sn dendrites, AIDSn and Cu6Sns IMCs) during solidification was investigated. Relationships between the growth behaviour of these components and their internal lattice orientation were studied. The characteristic growth and coupling of AIDSn IMCs and the Sn matrix to form eutectics has been elaborated and presented in this study for - 1- the first time. Based on the results, the forming process of a eutectic SnAgCu grain under the non-equilibrioum solidification condition was illustrated; and major factors that determine the lattice-orientation, size and substructure of the grain were discussed. Meso- and Micro- scale mechanical behaviour of a SnAgCu solder joint To study the size effect on the microstructure, and subsequently, the meso-scale mechanical behaviour, solder joints were manufactured with varying geometries. Shearing tests were performed on these meso-scale joints. The results first demonstrated that the anisotropic characteristics of a SnAgCu grain play an important role in the mechanical behaviour of both a meso-scale solder joint and the adjacent interfacial IMCs. To further investigate the micro-scale deformation and damage mechanisms, micro-mechanical tests were preformed within a SnAgCu grain. Constitutive equations for a SnAgCu grain Based on the experimental results, a crystal model was established to describe the local microstructure-dependent mechanical behaviour. The constitutive equation was implemented by means of the finite element approach, and applied in solder joints of a Flip Chip (FC) package by a multi-scale method. To describe the crystal behaviour at the higher temperature, the model was improved to account for deformations due to vacancy diffusion and thermal expansion. This model was integrated by an implicit approach, and implemented in a full three dimension (3D) finite element (FE) model.

Cette étude présente le brasage isotherme sous vide des alliages d’aluminium appliqué à la fabrication d’échangeurs thermiques. Ainsi, on étudie les évolutions microstructurales des nuances 3003 (Al-Mn) et 4004 (Al-Si-Mg) au cours des différentes étapes du cycle de brasage. Une double approche est mise en œuvre. Dans un premier temps, des échantillons modèles sont traités thermiquement en laboratoire. On suit l’évolution des phases en présence dans les deux alliages et les phénomènes de diffusion à l’état solide grâce à des analyses EDS. Nous montrons que les outils de simulation thermodynamique Thermo-Calc et DICTRA sont fiables à des températures supérieures à 400°C. On propose ensuite une description des mécanismes gouvernant la fusion du métal d’apport. Nous montrons qu’elle aboutit à la ségrégation d’un liquide enrichi en Si à la surface du métal d’apport. Dans un deuxième temps, des essais sont réalisés en industrie afin de prendre en compte les paramètres du brasage réel. Nous mettons en évidence des phénomènes de dissolution excessive et de pénétration de liquide aux joints de grains. Nous identifions les mécanismes qui gouvernent l’apparition de ces problèmes métallurgiques au cours du brasage. Ainsi, une faible taille de grains du métal de base et une diffusion préférentielle aux joints de grain sont mises en cause / This study deals with the vacuum TLP (Transient Liquid Phase) brazing of aluminum alloys applied to the manufacture of heat exchangers. Thus, the microstructure evolutions of 3003 (Al-Mn) and 4004 (Al-Si-Mg) alloys during the whole assembly process are studied. Firsty, model samples are heat treated in laboratory. The phase transformations and the solid state diffusion between the filler alloy and the base alloy are studied. The results are compared to thermodynamic predictions obtained with both Thermo-Calc and DICTRA softwares. We conclude that these tools are reliable at temperatures above 400°C. The fusion path of the filler alloy is described. It is shown that a Si enriched liquid is formed at the clad surface. On a second time, tests are carried out in industrial conditions, in order to take actual brazing parameters into account. Excessive dissolution and liquid penetration at grain boundaries are observed. The fine grained structure of the base alloy associated to a preferential diffusion at grain boundaries appear to be the main causes