Thermosonic ball bonding is a key technology in electrical interconnections between an integrated circuit and an external circuitry in microelectronics. Although this bonding process has been extensively utilised in electronics packaging industry, certain fundamental aspects behind all the practice are still not fully understood. This thesis is intended to address the existing knowledge gap in terms of bonding mechanisms and interfacial characteristics that are involved in thermosonic gold and copper ball bonding on aluminium pads. The research specifically targets the fine pitch interconnect applications where a thin metal wire of approximately 20 µm in diameter is commonly used. In thermosonic ball bonding process, a thin gold or copper ball formed at the end of a wire is attached to an aluminum pad through a combination of ultrasonic energy, pressure and heat, in order to initiate a complex solid-state reaction. In this research, the mechanisms of thermosonic ball bonding were elaborated by carefully examining interfacial characteristics as the results of the bonding process by utilising dual-beam focused ion beam and high resolution transmission electron microscopy, including the breakdown of the native alumina layer on Al pads, and formation of initial intermetallic compounds (IMCs). The effect of bonding parameters on these interfacial behaviours and bonding strength is also investigated in order to establish an inter-relationship between them. Interfacial evolution in both Au-Al and Cu-Al bonds during isothermal annealing in the temperature rage from 175ºC to 250ºC was investigated and compared. The results obtained demonstrated that the remnant alumina remains inside IMCs and moves towards the ball during annealing. The IMCs are formed preferentially in the peripheral and the central areas of the bonds during bonding and, moreover, they grow from the initially formed IMC particles. Growth kinetics of Cu-Al IMCs obey a parabolic growth law before the Al pad is completely consumed. The activation energies calculated for the growth of CuAl2, Cu9Al4 and the combination (CuAl2 + Cu9Al4) are 60.66 kJ/mol, 75.61 kJ/mol, and 65.83 kJ/mol, respectively. In Au-Al bonds, Au-Al IMC growth is controlled by diffusion only at the start of the annealing process. A t^0.2-0.3 growth law can be applied to the Au-Al IMC growth after the Al pad is depleted. The sequence of IMC phase transformation in both Au-Al and Cu-Al bonds were investigated. Voids in Au-Al bonds grow dramatically during annealing, however, only a few voids nucleate and grow very slowly in Cu-Al bonds. The mechanisms of void formation, including volumetric shrinkage, oxidation and metal diffusion were proposed and discussed.

Tin electrodeposits produced from aqueous electrolytes are frequently used within the electronics industry due to their high solderability and corrosion protection. One limitation to using these deposits is their spontaneous formation of long conductive filament whiskers. These whiskers grow post-electrodeposition and increase the risk of unwanted electrical shorts within electronic devices. In this thesis, tin electrodeposits produced from a proprietary bright acid Tinmac electrolyte, currently used in industry, were studied. Electrodeposits were produced using a range of current densities with and without agitation and were characterised with respect to crystallographic orientation, topography and surface finish. Moreover, the intermetallic compound (IMC) growth produced at the copper substrate-tin coating interface was assessed over a period of time as its growth is considered to be a significant driving force behind whisker formation. In addition, a technique for the electrochemical anodic oxidation of tin electrodeposits on copper substrates was developed. This technique was used throughout this project for the study of IMC growth from tin electrodeposits as it was able to effectively remove the tin whilst leaving the IMCs and substrate unaffected. Ionic liquids exhibit promising electrochemical characteristics for electrodeposition but are still not widely utilised in industry. Their ability to deposit tin coatings has been studied in the present investigation. Trials concentrated on process optimisation to produce uniform electrodeposits by varying current density, SnCl2.2H2O concentration, and electrolyte composition. These deposits were then characterised and compared to tin coatings of similar thickness produced from Tinmac with respect to topography, surface finish, crystallographic orientation, IMC growth, and whisker propensity. Electrodeposits produced from the ionic liquid electrolyte exhibited a different crystallographic texture, topography, and IMC growth compared to those produced from Tinmac. Moreover, the deposit produced from the ionic liquid featured increased whisker growth compared to those produced from Tinmac, but in a wider context, far less growth than conventional tin electrodeposits in the literature. In addition, by exploiting other electrochemical characteristics of ionic liquids, such as their large potential window, future work may be able to produce novel tin or tin alloy electrodeposits which may further reduce the whisker propensity of deposits produced in this investigation.

Neste trabalho, investigou-se a elaboração mecânica do composto intermetálico TiFe por moagem de bolas com alta energia. Uma forte aderência do material moído, particularmente nas paredes do recipiente de moagem, foi o principal problema verificado com tempos de moagem superiores a 1 hora (moinho agitador). Tentativas para resolver este problema foram realizadas inicialmente com o emprego de agentes controladores de processo (ACPs), como etanol, ácido esteárico, polietileno de baixa densidade, benzeno e ciclohexano, em diferentes quantidades (1 a 20% em massa) e tempos (1 a 40 h), mantendo-se constantes outros parâmetros de moagem como a razão bola:pó em massa (10:1) e o tamanho das bolas (=7mm). Os rendimentos mais elevados (em termos da massa de pó não aderido) foram obtidos quando se utilizaram grandes quantidades de benzeno e ciclohexano (101 e 103% em massa, respectivamente), porém com a formação de TiC ao invés de TiFe em razão da decomposição do ACP e reação do carbono com as partículas de titânio. As moagens foram realizadas posteriormente sem o emprego de qualquer ACP e também utilizando um moinho planetário. Várias estratégias foram investigadas para se tentar mitigar a aderência incluindo-se: (a) moagem de uma pequena quantidade da mistura de pós de Ti e de Fe, revestindo as paredes do recipiente e as bolas de moagem, antes da moagem da carga principal, (b) moagem pausada com aberturas intermediarias do recipiente em atmosfera ambiente, (c) moagem pausada para rotação e inversão da posição do recipiente de moagem (apenas no moinho agitador), (d) moagem isolada dos pós de Ti e de Fe, antes da moagem da mistura, e (e) moagem do pó de Fe com o Ti hidretado. Os melhores resultados, em termos de diminuição da aderência combinada com a formação majoritária do composto TiFe durante a moagem, foram obtidos quando se adotou o procedimento de inversão/rotação, juntamente com o processo de revestimento preliminar do recipiente e das bolas de moagem (26% em massa). Rendimentos maiores foram obtidos com a utilização do TiH2 no moinho planetário, porém sem a formação majoritária do TiFe durante a moagem. / In this work an investigation on the mechanical alloying of the intermetallic compound TiFe by high-energy ball milling was conducted. Strong adherence of milled material, particularly at the vial walls, was seen to be the main problem at milling times higher than 1 hour (shaker mill), hindering the compound synthesis. Attempts to prevent this problem were accomplished first by adding different process control agents (PCAs), like ethanol, stearic acid, low density polyethylene, benzene and cyclohexane at variable quantities (1 to 20 wt. %) and times (1 to 40 h), keeping constant other milling parameters like ball to powder mass ratio (10:1) and balls size (=7mm). Highest yields (related to the non adhered powder) were attained with larger amounts of benzene and cyclohexane (101 and 103 wt. %, respectively), but with TiC formation during milling instead of TiFe due to the PCA decomposition and the reaction of the carbon with and titanium particles. Milling was conducted further without adding any PCA and also using a planetary ball mill. Several strategies were tried to avoid or minimize the adherence including: (a) milling of a small quantity of the Ti and Fe powder mixture, dirtying the vial walls and the balls surfaces before milling the main charge, (b) stepwise milling with intermediate openings of the vial in air, (c) stepwise milling with the rotation and the inversion of the vial position between the steps (only in the shaker mill), (d) milling Ti and Fe powders (apart from each other) before milling the mixture of them, and (e) milling Fe powder with Ti hydride powder. Best results concerning both yield and major TiFe formation during milling were verified with the rotation/inversion procedure combined with preliminar dirtying of the vial and balls (26 wt.% in the shaker mill). Higher yields could be attained by using TiH2 powder in the planetary mill, but with no major TiFe formation during milling.

In past years, legislative pressures (particularly in Japan and Europe) had forced the electronics industry to eliminate Pb from their end products and manufacturing processes. With respect to factors such as ease of converting existing tin-lead plating systems, ease of manufacture and compatibility with existing assembly methods, pure tin plating is seen by many in the industry as a potentially simple and cost effective alternative to SnPb-based systems. The problem of spontaneous tin whisker formation, a characteristic of pure tin, still needs to be addressed, as it can lead to device failure by shorting two terminals on electronic devices. This possibility gives rise to major reliability concerns. The study relates to an electronic component with pure tin deposit layer on the part for electric connection, wherein pure tin deposit layer is a fine grained tin deposit layer composed of grains with smaller size in the direction perpendicular to the deposit surface than in the direction parallel to the deposit surface. It is called irregular tin grain structure. It applies a process for plating an electronic component, so as to form a pure tin deposit layer on the part for electric connection, comprising the steps of: adjusting the composition of tin plating solution in which starter additive and brighter additive are included; moving the electronic component through the tin plating solution, so as to form a fine grained tin deposit layer on the part for electric connection. We performed a DoE by depositing different tin grain structures with variant thickness. After whisker test in high temperature/high humidity and room condition, we confirmed corrosion mechanism, intermetallic morphology, and different behaviour of tin atoms. To summarize the studies, as compared with the prior arts, irregular grain structure can validly inhibit the whisker growth.

none

Friction stir processing (FSP) is applied to produce particulate-reinforced aluminum matrix composites with ultrafine grained structure from elemental powder mixtures of Al-Cu, Al-Ti and Al-Si. The microstructures of the composites were characterized by the use of XRD, SEM and TEM. Microhardness, tensile and compressive tests were performed to evaluate the mechanical properties of these composites. The mechanisms of microstructure evolution in FSP and the strengthening mechanisms in these composites are discussed. In the Al-Si system, the Si particles were broken and uniformly distributed in the stir zone by the application of multiples-pass FSP. The average size of Si particles and Al grains were refined to below ~2

New environmental regulations enforce the electronic industry to replace Pb-Sn solder due to Pb could contaminate our environment. Pure Sn has good material properties such as solderability, conductivity and anti-corrosion. Pure Sn is a good candidate to replace Pb-Sn solder. One of the disadvantages of pure Sn is the whisker growth phenomenon. Whisker problem has become a major concern in electronic industry due to the trend toward component miniaturization and pitch reduction. It is well understood that the root cause for tin whisker growth is the compressive stress within the tin layer. In the literature, the main stress sources are, (1) the intermetallic layer induced interface stress, (2) the difference of thermal expansion coefficient between Sn layer and substrate and (3) the mechanical residual stress from trim-form operation after tin plating. In our study, we used the electrochemical electrolysis method and Cross-section Polisher (CP) to examine the tin whisker growth mechanism. In the result, we can clearly show the Cu6Sn5 phase grow up in the tin grain boundary regions and demonstrate that the Cu6Sn5 phase formation is the main cause of the tin whisker growth. We also discuss the relationship of tin whisker growth and tin-plating process parameters that include the temperature effect; Ni underlay effect and tin-plating bath effect. For the temperature effect, the Cu6Sn5 is the major phase at 150¢XC aging. The mechanism behind its growth mechanism was grain boundary diffusion at the earlier stage and then the bulk diffusion in the later stage. The application of 150¢XC post-heat treatment could drive the bulk diffusion and form a layer type Cu6Sn5 phase to eliminate the whisker growth. For the Ni underlay effect, the Ni underlay can block the Cu atom diffusion to the tin layer and changed the tin layer stress state from compressive to tensile. Therefore, the tin whisker can be eliminated. For the tin-plating bath effect, in the sulfuric acid base and uses Triton X-100 as the surface active agent, may transform the whisker type to particular tin grain type. Thus, this tin-plating solution can restrain the tin whisker growth.

The effects of intermetallic compound (IMC) formation on reliability of Pb-free Sn-based solders for flip chip and three-dimensional (3D) interconnects were studied. The dissertation is organized into four parts. In the first part, the effect of Sn grain orientation on electromigration (EM) reliability of Pb-free Sn-based flip chip solder joints was studied. The Sn grain microstructure in flip chip solder joints was characterized using the electron backscatter diffraction (EBSD) technique, and wa found to be closely related to the EM failure mechanims. The approach to grain structure optimization for improved EM reliability was also explored. In addition to the experimental work, a kinetic analysis was formulated to investigate the early EM degradation mechanism in Sn-based solder joints with Ni under-bump metallization (UMB). The aforementioned kinetic analysis, the intrinsic diffusion coefficients were not readily available in the literature. In the second part of the work, a Monte Carlo method known as simulated annealing was applied to estimate the unknown diffusion coefficients using a multi-parameter optimization method by fitting to experimental measurements. The intrinsic diffusion coefficients of Ni and Sn in Ni₃Sn₄ between 150 and 200°C, and those of Cu and Sn in Cu₃Sn and Cu₆Sn₅ between 120 and 200°C were estimatd. The activation energies for these diffusion coefficients were also determined. Together, this provides the diffusivity parameters to predict the intermetallic growth as a function of temperature. The third objective focused on the EM reliability of Sn-based microbump joints in 3D interconnects with through-silicon vias (TSVs). No EM-induced bump failure was observed, showing a robust EM reliability in microbumps. High temperature thermal annealing test was also performed on microbumps with three different metallizations in an effort to explore structural and process optimization. Finally, interfacial reaction induced stress in IMC microbumps was investigated. A numerial analysis was formulated to study the concurrent diffusion, phase transformation, and deformation in the process of IMC formation. Stress generation due to unbalanced diffusion rates and volumetric change upon phase transformation was considered. The coupled analysis was applied to investigate Ni₃Sn₄ growth in the Ni-Sn microbumping system. A simulation approach based on finite difference method with moving boundaries was employed to numerically solve stress evolution in Ni₃Sn₄. The equilibrium stress was also investigated using a modified model with a finite thickness of solder. Simulation predictions were found to be in good qualitative agreement with experimental observations. / text

Neste trabalho, investigou-se a elaboração mecânica do composto intermetálico TiFe por moagem de bolas com alta energia. Uma forte aderência do material moído, particularmente nas paredes do recipiente de moagem, foi o principal problema verificado com tempos de moagem superiores a 1 hora (moinho agitador). Tentativas para resolver este problema foram realizadas inicialmente com o emprego de agentes controladores de processo (ACPs), como etanol, ácido esteárico, polietileno de baixa densidade, benzeno e ciclohexano, em diferentes quantidades (1 a 20% em massa) e tempos (1 a 40 h), mantendo-se constantes outros parâmetros de moagem como a razão bola:pó em massa (10:1) e o tamanho das bolas (=7mm). Os rendimentos mais elevados (em termos da massa de pó não aderido) foram obtidos quando se utilizaram grandes quantidades de benzeno e ciclohexano (101 e 103% em massa, respectivamente), porém com a formação de TiC ao invés de TiFe em razão da decomposição do ACP e reação do carbono com as partículas de titânio. As moagens foram realizadas posteriormente sem o emprego de qualquer ACP e também utilizando um moinho planetário. Várias estratégias foram investigadas para se tentar mitigar a aderência incluindo-se: (a) moagem de uma pequena quantidade da mistura de pós de Ti e de Fe, revestindo as paredes do recipiente e as bolas de moagem, antes da moagem da carga principal, (b) moagem pausada com aberturas intermediarias do recipiente em atmosfera ambiente, (c) moagem pausada para rotação e inversão da posição do recipiente de moagem (apenas no moinho agitador), (d) moagem isolada dos pós de Ti e de Fe, antes da moagem da mistura, e (e) moagem do pó de Fe com o Ti hidretado. Os melhores resultados, em termos de diminuição da aderência combinada com a formação majoritária do composto TiFe durante a moagem, foram obtidos quando se adotou o procedimento de inversão/rotação, juntamente com o processo de revestimento preliminar do recipiente e das bolas de moagem (26% em massa). Rendimentos maiores foram obtidos com a utilização do TiH2 no moinho planetário, porém sem a formação majoritária do TiFe durante a moagem. / In this work an investigation on the mechanical alloying of the intermetallic compound TiFe by high-energy ball milling was conducted. Strong adherence of milled material, particularly at the vial walls, was seen to be the main problem at milling times higher than 1 hour (shaker mill), hindering the compound synthesis. Attempts to prevent this problem were accomplished first by adding different process control agents (PCAs), like ethanol, stearic acid, low density polyethylene, benzene and cyclohexane at variable quantities (1 to 20 wt. %) and times (1 to 40 h), keeping constant other milling parameters like ball to powder mass ratio (10:1) and balls size (=7mm). Highest yields (related to the non adhered powder) were attained with larger amounts of benzene and cyclohexane (101 and 103 wt. %, respectively), but with TiC formation during milling instead of TiFe due to the PCA decomposition and the reaction of the carbon with and titanium particles. Milling was conducted further without adding any PCA and also using a planetary ball mill. Several strategies were tried to avoid or minimize the adherence including: (a) milling of a small quantity of the Ti and Fe powder mixture, dirtying the vial walls and the balls surfaces before milling the main charge, (b) stepwise milling with intermediate openings of the vial in air, (c) stepwise milling with the rotation and the inversion of the vial position between the steps (only in the shaker mill), (d) milling Ti and Fe powders (apart from each other) before milling the mixture of them, and (e) milling Fe powder with Ti hydride powder. Best results concerning both yield and major TiFe formation during milling were verified with the rotation/inversion procedure combined with preliminar dirtying of the vial and balls (26 wt.% in the shaker mill). Higher yields could be attained by using TiH2 powder in the planetary mill, but with no major TiFe formation during milling.

<p>The lack of a thorough understanding of the solidification behaviors of the proeutectic Al₃Sc and the Al-Al₃Sc eutectic in a hypereutectic Al-Sc alloy stimulates the present dissertation. The major findings for the single-phase growth of the proeutectic Al₃Sc is summarized as follows: At a low cooling rate (~1 ºC·s^-1), the proeutectic Al₃Sc phase’s formation was governed by the lateral growth, exposing six flat {100} facets. At an intermediate cooling rate (~400 ºC·s^-1), the proeutectic Al₃Sc grew in a dendritic manner, with well-defined backbones extending in eight <111> directions and paraboloidal dendrite tips, although the dendrite tips and side-branches turned into faceted steps at a late growth stage,when the lateral growth prevailed. At a high cooling rate (~1000 ºC·s^-1), the proeutectic Al₃Sc primarily crystallized into an entirely seaweed-structured particle, which was composed of interior compact seaweeds and exterior fractal seaweeds. In order to verify the proposed dendritic and seaweed growth mechanisms for the proeutectic Al₃Sc, various morphological stability criteria were used, and fair agreement between the observed and the estimated characteristic length scales was reached.</p><p>On the Al-Al₃Sc eutectic side, it was found that a rod-typed Al₃Sc eutectic phase prevalently existed in an as-cast hypereutectic Al-Sc alloy that solidified via both slow cooling in air (~1 ºC·s⁻¹) and rapid cooling in a wedge-shaped copper mold (up to ~3000 ºC·s⁻¹). Al-Al₃Sc eutectic dendrites were identified within a narrow region near the edge of the wedge. The eutectic dendrites had an equiaxed dendritic contour and a rod eutectic structure inside. Quantitative assessments revealed that an interface undercooling of 48.2 ºC was required to form the eutectic dendrites, or in other words, to enter the coupled zone of the Al-Al₃Sc phase diagram. Furthermore, a phenomenon of scientific interest was discussed: When crystallizing under a near-equilibrium condition, the eutectic Al₃Sc phase formed a non-faceted morphology, in contradiction to its faceted nature. Based on the competitive growth criterion, it was deduced that the non-faceting of the eutectic Al₃Sc phase essentially reduced the interface undercooling for the resultant regular eutectic, in comparison to an otherwise irregular eutectic that would contain a faceted eutectic Al₃Sc phase.</p>