No description available.

Electromigration (EM) and thermomigration (TM) reliability of Pb-free solder joints are emerging as critical concerns in advanced packages. In this study, EM and TM phenomena in Sn-2.5Ag solder joints with thick Cu or thin Ni under-bump metallurgy (UBM) were investigated. A series of EM tests were performed to obtain activation energy (Q) and current density exponent (n), and to understand failure mechanisms. Joule heating was also taken into account. Q and n values were determined as follows: for Cu UBM solders, Q = 1.0 eV and n = 1.5; for Ni UBM solders, Q = 0.9 and n = 2.2. Important factors limiting EM reliability of Pb-free solder joints were found to be UBM dissolution with extensive intermetallic compound (IMC) growth and current crowding. IMC growth without current stressing was found to follow the parabolic growth law whereas linear growth law was observed for Cu₆Sn₅ and Ni₃Sn₄ under high current stressing. For Cu UBM solders, the apparent activation energy for IMC growth was consistent with the activation energy for EM, which supports that EM failure was closely related to IMC growth. In contrast, for Ni UBM solders the apparent activation energy was higher than the EM activation energy. It was suggested that the EM failure in the Ni UBM solders could be associated with more than one mass transport mechanism. The current crowding effect was analyzed with different thicknesses of Ni UBM. It was found that the maximum current density in solder could represent the current density term in Black's equation better than the average current density. FEM studies demonstrated that current crowding was mainly controlled by UBM thickness, metal trace design, and passivation opening diameter. A large temperature gradient of the order of 10³ °C/cm was generated across the sample to induce noticeable TM and to compare its effect against that of EM. TM-induced voiding was observed in Ni UBM solders while UBM dissolution with IMC formation occurred in Cu UBM solders. However, the relative effect of TM was found to be several times smaller than that of EM even at this large temperature gradient. / text

Cette thèse est dédiée à l’étude, à la réalisation et à la caractérisation de nouvelles périphéries de TRIAC. L’objet de cette recherche est de réduire l’espace occupé par la périphérie en tentant de conserver le même niveau de performances au blocage. Deux voies d’amélioration ont été poursuivies : l’une concerne la réalisation de caissons d’isolation par thermomigration d’aluminium, l’autre implique l’intégration du silicium poreux dans le caisson d’isolation. La thermomigration d’aluminium est une technique attractive permettant de remplacer les techniques de diffusion conventionnelles. Son industrialisation subit cependant quelques verrous technologiques, notamment le retrait des résidus aluminés et la formation de billes. Deux procédés de gravure ont été développés en vue d’enlever sélectivement l’ensemble de résidus. L’origine des billes a été analysée à l’aide d’observations expérimentales et de modélisations numériques. En utilisant un motif incluant des trous carrés aux intersections, des résultats encourageants ont été démontrés malgré une uniformité thermique encore optimisable. La deuxième voie d’innovation consiste à profiter des propriétés diélectriques du silicium poreux. Un procédé de masquage par fluoropolymère a été développé pour la localisation du silicium poreux. Les conditions d’anodisation adéquates ont été déterminées. La caractérisation de prototypes a montré des tenues au blocage largement améliorées par rapport à l’étude précédente. Bien que les tenues en tension nécessaires n’aient pas été atteintes, des courants de fuite inférieures à 10 μA ont été constatés jusqu’à plusieurs centaines de volts. / This thesis is dedicated to the study, the realization and the testing of “Planar” type TRIAC with novel peripheries. The aim of this research is to shrink the device periphery while maintaining the same level of blocking performances. Two paths of innovation have been pursued: one concerning Al-Si thermomigration for the production of through-wafer isolation walls, and the other involving porous silicon and its integration in the isolation walls. Al-Si thermomigration is an attractive mean allowing to replace conventional diffusion technologies. However, several remaining issues, such as the removal of the unintentional residues and the ball formation phenomenon, block its commercial application. Two different etching procedures have been developed in order to selectively remove all residues. The origin of the ball phenomenon has been analyzed using experimental observations and numerical modeling. By using a new pattern including square holes at intersections, encouraging results have been demonstrated in spite of an optimizable thermal uniformity. The second way of innovation is to take advantage of the dielectric properties of the porous silicon. A fluoropolymer masking process has been developed for local porous silicon formation. The appropriate anodization conditions have been determined. The characterization results showed improved blocking performances compared to the previous study. Although the necessary voltage requirements are not met, leakage currents of less than 10 μA have been observed up to several hundred volts.

L'assemblage tridimensionnel des circuits microélectroniques et leur utilisation dansdes conditions environnementales extrêmement sévères nécessitent ledéveloppement d’alternatives plus robustes pour les contacts électriques. Unetechnique prometteuse est la transformation des contacts de brasure conventionnelleen composés intermétalliques (IMC). Ce processus est appelé « Transient LiquidPhase Soldering » (TLPS).Dans ce contexte, des tests accélérés permettant la formation d’IMC parélectromigration et thermomigration ont été effectués sur des structures « Packageon Package ». L'objectif principal est le développement d'un modèle généralpermettant de décrire la formation des IMC dans les joints de brasure. Combiné avecune analyse par éléments finis ce modèle pourra être utilisé pour prédire la formationdes IMC dans les joints de brasure pour des structures et des profils de missiondifférents. Le modèle de formation des IMC pourra être utilisé pour optimiser unprocessus TLPS. / A further miniaturization of microelectronic components by three dimensionalpackaging, as well as the use of microelectronic devices under harsh environmentconditions, requires the development of more robust alternatives to the existing Snbased solder joints. One promising technique is the diffusion and migration driventransformation of conventional solder bumps into intermetallic compound (IMC)connections. The related process is called transient liquid phase soldering (TLPS).Against this background an investigation of the IMC formation under consideration ofelectromigration and thermomigration was performed. For the stress tests Packageon Package structures are used. The final result is a general model for the IMCformation in solder joints. Combined with a Finite Element Analysis (FEA) this modelis used to predict the IMC formation in solder joints for a broad range of boundaryconditions. In future the model of the IMC formation can be used to optimize a TLPSprocess.

Nuclear and gas turbine power plants, computer chips, and other devices and industries are running hotter than ever for longer than ever. With no apparent end to the trend, the potential arises for a phenomenon known as thermotransport to cause undesirable changes in these high temperature materials. The diffuse-interface method known as the phase-field model is a useful tool in the simulation and prediction of thermotransport driven microstructure evolution in materials. The objective of this work is to develop a phase-field model using practical and empirical properties of thermodynamics and kinetics for simulating the interdiffusion behavior and microstructural evolution of single and multiphase binary alloy system under composition and/or temperature gradients. Simulations are carried out using thermodynamics and kinetics of real systems, such as the U-Zr solid metallic fuel, with emphasis on the temperature dependencies of the kinetics governing diffusional interactions in single-phase systems and microstructural evolution in the presence of multiple driving forces in multi-phase systems. A phase field model is developed describing thermotransport in the γ phase of the U-Zr alloy, a candidate for advanced metallic nuclear fuels. The model is derived using thermodynamics extracted from the CALPHAD database and temperature dependent kinetic parameters associated with thermotransport from the literature. Emphasis is placed upon the importance of the heat of transport, Q*, and atomic mobility, β. Temperature dependencies of each term are estimated from empirical data obtained directly from the literature, coupled with the textbook phenomenological formulae of each parameter. A solution is obtained via a finite volume approach with the aid of the FiPy® partial differential equation solver. Results of the simulations are described based on individual flux contributions from the gradients of both composition and temperature, and are found to be remarkably similar to experimental results from the literature. iv In an additional effort the thermotransport behavior of a binary two-phase alloy is modeled, for the first time, via the phase-field method for a two-phase (γ + β) U-Zr system. The model is similarly built upon CALPHAD thermodynamics describing the γ and β phases of the U-Zr system and thermotransport parameters for the γ phase from literature. A parametric investigation of how the heats of transport for U and Zr in the β phase affect the redistribution is performed, and the interplay between system kinetics and thermodynamics are examined. Importantly, a strict control over the microstructure that is placed into the temperature gradient ( ) is used to eliminate the randomness associated with microstructural evolution from an initially unstable state, allowing an examination of exactly how the β phase thermotransport parameters affect the redistribution behavior of the system. Results are compared to a control scenario in which the system evolves only in the presence of thermodynamic driving forces, and the kinetic parameters that are associated with thermotransport are negligible. In contrast to the single-phase simulations, in the presence of a large thermodynamic drive for phase transformation and stability, the constituent redistribution caused by the thermotransport effect is comparatively smaller.

Ce mémoire présente nos travaux sur la thermomigration de l'aluminium dans le silicium comme alternative à la diffusion bore pour la réalisation des murs d'isolation dans les composants de puissance bidirectionnels. Dans un premier temps, nous avons ainsi mis en évidence les limites de la diffusion du bore à l'état solide, limites dues essentiellement à son bilan thermique prohibitif. Parmi les solutions alternatives envisagées, la thermomigration de l'aluminium présente un certain nombre d'avantages parmi lesquels un bilan thermique très faible et un dopage élevé et constant. Le procédé consiste en la migration d'un alliage liquide Al/Si sous l'effet d'un gradient de température vertical avec cristallisation, dans le sillage de la goutte, d'une solution solide de silicium dopé aluminium (à 1019 at/cm3 environ). L'exigence de gradient thermique vertical impose l'utilisation d'un four de recuit rapide spécialement conçu à cet effet. L'étude expérimentale du phénomène nous a permis de mettre en évidence les problématiques "industrielles" liées au procédé et à l'équipement. Ainsi, l'utilisation nécessaire d'oxygène pendant le recuit de thermomigration perturbe très fortement le déroulement du procédé et nous avons dû apporter des réponses nouvelles à ce problème, notamment en considérant les paramètres géométriques du motif d'aluminium. De même, l'analyse approfondie des résultats obtenus sur le four de laboratoire nous a permis de donner les spécifications d'un nouveau four en vue du transfert industriel de la thermomigration. Enfin, grâce à la maîtrise relative du procédé dans son ensemble, nous avons conçu et réalisé une structure nouvelle de puissance, le thyristor sur épitaxie, dont la fonction de tenue en tension inverse a été démontrée. Même si un certain nombre de problèmes restent en suspens, les résultats obtenus au cours de cette étude sont très prometteurs en vue d'une industrialisation future du procédé.