Following the continually increasing demand for high-density interconnection and multilayer packaging for chips, solder bump size has decreased significantly over the years, this has led to some challenges in the reliability of interconnects. This thesis presents research into the resulting effects of miniaturization on the interconnection with Sn-solder, especially focusing on the full intermetallics (IMCs) micro-joints which appear in the 3D IC stacking packaging. Thereby, systematic studies have been conducted to study the microstructural evolution and reliability issues of Cu-Sn and Cu-Sn-Ni IMCs micro-joints. (1) Phenomenon of IMCs planar growth: The planar IMCs interlayer was asymmetric and composed of (Cu,Ni)6Sn5 mainly in Ni/Sn (2.5~5 μm)/Cu interconnect. Meanwhile, it was symmetric two-layer structure in Cu/Sn (2.5~5 μm)/Cu interconnect with the Cu3Sn fine grains underneath Cu6Sn5 cobblestone-shape-like grains for each IMCs layer. Besides, it is worth noticing that the appearance of Cu-rich whiskers (the mixture of Cu/Cu2O/SnOx/Cu6Sn5) could potentially lead to short-circuit in the cases of ultra-fine ( < 10 μm pitch) interconnects for the miniaturization of electronics devices. (2) Microstructural evolution process of Cu-Sn IMCs micro-joint: The simultaneous solidification of IMCs interlayer supressed the scalloped growth of Cu6Sn5 grains in Cu/Sn (2.5 μm)/Cu interconnect during the transient liquid phase (TLP) soldering process. The growth factor of Cu3Sn was in the range of 0.29~0.48 in Cu-Cu6Sn5 diffusion couple at 240~290 °C, which was impacted significantly by the type of substrates. And the subsequent homogenization process of Cu3Sn grains was found to be consistent with the description of flux-driven ripening (FDR) theory. Moreover, Kirkendall voids appeared only in the Cu3Sn layer adjacent to Cu-plated substrate, and this porous Cu3Sn micro-joint was mechanically robust during the shear test. (3) Microstructural evolution of Cu-Sn-Ni IMCs micro-joint: There was obvious inter-reaction between the interfacial reactions in Ni/Sn (1.5 μm)/Cu interconnect. The growth factor of (Cu,Ni)3Sn on Cu side was about 0.36 at 240 °C, and the reaction product on Ni side was changed from Ni3Sn4 into (Cu,Ni)6Sn5 with the increase of soldering temperature. In particular, the segregation of Ni atoms occurred along with phase transformation at 290 °C and thereby stabilized the (Cu,Ni)6Sn5 phase for the high Ni content of 20 at.%. (4) Micro-mechanical characteristics of Cu-Sn-Ni IMCs micro-joint: The Young s modulus and hardness of Cu-Sn-Ni IMCs were measured by nanoindentation test, such as 160.6±3.1 GPa/ 7.34±0.14 GPa for (Cu,Ni)6Sn5 and 183.7±4.0 GPa/ 7.38±0.46 GPa for (Cu,Ni)3Sn, respectively. Besides, in-situ nano-compression tests have been conducted on IMCs micro-cantilevers, the fracture strength turns out to be 2.46 GPa. And also, the ultimate tensile stress was calculated to be 2.3±0.7 GPa from in-situ micro-bending tests, which is not sensitive with the microstructural change of IMCs after dwelling at 290 °C.

Les mémoires à changement de phase sont l'un des candidats les plus prometteurs pour la prochaine génération de mémoires non-volatiles. Un intense effort de recherche est requis pour optimiser les matériaux à changement de phase (PC) utilisés dans ces mémoires. En particulier, il a été démontré que le dopage améliore les propriétés de rétention des dispositifs. Par ailleurs, l'étude des effets de réduction de taille et des effets des matériaux d'interface sur les propriétés des matériaux à changement de phase est encore un sujet de recherche ouvert. Dans ce contexte, la première partie de la thèse est dédiée à l'investigation de la structure locale de GeTe amorphe dopé avec C ou N. L'effet du dopage sur la structure a été observé expérimentalement via l'apparition d'un nouveau pic dans la fonction de distribution de paires de GeTe dopé, ce qui montre la formation d'une nouvelle liaison interatomique absente dans le matériau non dopé. La présence de nouvelles configurations incluant le carbone et l'azote a été confirmée par des simulations ab initio. L'objet de la deuxième partie de la thèse est l'influence de la réduction de taille sur la cristallisation de Ge2Sb2Te5 (GST). Des agrégats nanométriques de GST ont été fabriqués par pulvérisation puis déposés et étudiés par diffraction des rayons X en utilisant le rayonnement synchrotron. Dans l'état cristallisé une très forte déformation positive des agrégats est observée et attribuée à la matrice d'Al2O3 qui entoure les agrégats. La température de cristallisation des agrégats est de 25°C plus élevée que celle d'un film de GST de 10 nm déposé dans les mêmes conditions. Ce résultat est encourageant pour les futurs développements des mémoires à changement de phase car il montre que l'effet de réduction de taille sur la température de cristallisation peut-être faible. La troisième et dernière partie de la thèse est dédiée à l'investigation des effets des matériaux d'interface sur la température de cristallisation de films minces de GeTe et GST par des mesures de réflectivité et de diffraction des rayons X. Pour les deux matériaux, la température de cristallisation de films de 100 nm est plus grande pour une interface avec du Ta que pour une interface avec du TiN ou du SiO2. Une différence aussi marquée n'était jamais montré auparavant. Les résultats suggèrent que l'interface SiO2/GeTe est énergétiquement favorable pour la nucléation et la croissance de grains avec une orientation préférentielle et que les mécanismes de nucléation et croissance sont différents pour différents matériaux d'interface. / Phase Change Memories (PCM) are one of the best candidates for the next generation of non volatile memories. A great research effort is still needed in order to optimize the properties of phase change (PC) materials which are used in PCM devices. In particular, doping has been demonstrated to improve retention in devices. Moreover, a study of the effect of scaling and interface material on PC materials properties is still an open research field. In this context, the first part of the thesis is dedicated to investigate the local structure of C or N doped amorphous GeTe. The impact of doping is observed experimentally with the appearance of a new peak in the pair distribution function of doped GeTe, indicating the formation of a bond at a new distance that is absent in the undoped amorphous material. The presence of new environments involving carbon and nitrogen is confirmed through ab initio simulations. The subject of the second part of this thesis is the impact of confinement on Ge2Sb2Te5 (GST) crystallization mechanism. Nano-sized clusters of GST have been made by sputtering, deposited and then studied through X-ray diffraction using synchrotron radiation. The crystalline clusters experience a tensile strain that can be ascribed to the effect of the embedding Al2O3 matrix. Their crystallization temperature has been found to be only 25°C higher than the one observed for a thin film of GST of 10 nm deposited under the same conditions. This result is positive for the future development PCM because it indicates that the scaling effect on the crystallization temperature in phase change material can be small. The third and last part of the thesis is dedicated to the investigation of the interface material effect on the crystallization temperature of GeTe and GST thin films through reflectivity and X-ray diffraction measurements. In both GeTe and GST film 100 nm thick interfaced with Ta the crystallization temperature is higher than in the case of TiN or SiO2 interface. Such an interface effect on relatively thick films was never reported before. The results suggest that the SiO2/GeTe interface is energetically favourable for the nucleation and growth of grains with a preferred orientation and that nucleation and growth mechanisms are different for different interface materials.

This research aims to model the microstructure development of Si-containing bainitic free-machining steel, including allotriomorphic ferrite, idiomorphic ferrite, pearlite, Widmanstatten ferrite, bainite and martensite. The effect of recalescence has been included to give a better estimation of the cooling curve under natural cooling conditions. A model for estimating retained austenite size distribution in the carbide-free bainitic microstructure has been developed. Manganese sulphide particles are used in the free-machining steel to break chips during machining; its effect on the prior austenite grain size has been investigated, taking account of the sulphide shape. The theories of all the major solid state phase transformations involved in steel are reviewed in chapter 2. The theory of the simultaneous transformation model is presented in chapter 3.uu A recalescence model dealing with the heat of reaction has been developed in chapter 5 for bar-shaped products. The model is based on the integration of a heat transfer model, considering latent heat generation, into the simultaneous transformation framework. It has been found that latent heat can greatly affect the transformation, especially in the case of pearlite and Widmanstatten ferrite. Chapter 6 presents the model for estimating the size distribution of retained austenite regions. The model builds on the random division of an austenite grain by bainite sheaves, which means the sizes of the two new compartments generated by the division of an austenite grain by a bainite sheaf are allocated randomly. The next compartment to be divided is also chosen at random. Good agreement between prediction and experiment has been achieved for high carbon carbide-free bainitic microstructures. The transition temperature from upper to lower bainite is modelled in chapter 7. The model compares the time required for decarburising a supersaturated bainitic ferrite platelet and that for cementite precipitation within the ferrite platelet. Manganese, silicon and chromium are considered in the model. It is suggested that carbon and manganese favour lower bainite, whereas silicon promotes upper bainite. The effect of manganese sulphide particles on austenite grain boundary motion has been studied in chapter 8. These rod-shaped particles span many austenite grains; the result shows that the long rod-shaped particles are more effective in pinning the austenite grain boundary than spheres of the same volume, or even strings of identical spheres with the same total volume. Experimental work is presented in chapters 9 and 10. In situ synchrotron X-ray study of the bainite transformation reveals that the distribution of carbon in the residual austenite becomes heterogeneous as transformation progresses. Low carbon regions transform preferentially into martensite during cooling after isothermal bainite transformation. The partitioning of carbon was found to lag behind the bainite transformation; more time is needed as the transformation temperature is reduced. Tetragonality was not observed in either the bainitic ferrite or martensite, because the carbon content of the alloy is relatively low, and the Zener ordering temperature is below the bainite and martensite transformation temperature. No significant difference was observed in the kinetics of bainite transformation between the high sulphur and low sulphur steel.

A demanda do silício de grau solar (SiGS) utilizado na indústria fotovoltaica cresce atualmente em uma média de 35% ao ano e a projeção para a próxima década é de constante crescimento. Entretanto, a disponibilidade de matéria-prima para a produção do SiGS não tem crescido na mesma taxa, resultando em uma elevação de seu custo. Logo, a obtenção de rotas alternativas menos onerosas para a produção do SiGS torna-se crucial, principalmente sendo o Brasil o 3o maior produtor mundial de silício de grau metalúrgico. Tendo como objetivo a purificação do silício, o presente trabalho de pesquisa apresenta um estudo detalhado da solidificação direcional do silício, que é uma das etapas utilizadas em todas as rotas alternativas atualmente pesquisadas com o objetivo de purificação. Este estudo foi realizado em um forno do tipo Bridgman vertical, onde o silício líquido contido em um molde foi extraído da região quente do forno, resultando em uma solidificação direcional controlada e ascendente. Dois tipos de silício foram escolhidos como material de entrada para o processo de refino: o silício de grau metalúrgico e o silício de grau metalúrgico pré-refinado por lixiviação. Quatro velocidades de extração do molde entre 5 a 110 mm/s foram empregadas nos ensaios. A macrossegregação de impurezas nos lingotes obtidos foi medida através de análises químicas e a estrutura metalúrgica observada através de macro e micrografias. Os lingotes solidificados com menor velocidade de extração, onde se observa evidências de uma interface sólido-líquido plana, apresentam um grau de refino superior aos de maior velocidade, que mostram evidências de uma interface celular ou dendrítica. As análises químicas mostram que, após a solidificação direcional, os teores de Fe, Al, Cr e Ti medidos nas regiões da base e central dos lingotes atingem níveis especificados para o SiGS por alguns autores. / The demand for solar grade silicon (SiGS) in the photovoltaic industry has grown 35% per year and a constant growth rate is predicted for the next decade. The availability of feedstock to produce SiGS, however, has not grown at the same rate, resulting in increasing SiGS cost. Consequently, the development of less expensive alternative processing routes is of paramount importance to the photovoltaic industry in general, and particularly to Brazil, which is the 3rd largest producer of metallurgical grade silicon. The objective of the present work is to study the directional solidification of silicon, which is a step adopted in virtually all alternative processing routes under development to refine silicon. The directional solidification of silicon was carried out in a vertical Bridgman furnace, where liquid silicon was poured into a quartz mold, which was extracted at constant velocity from the hot region of the furnace, resulting in ascending directional solidification. Two types of feedstock silicon were used, namely, a metallurgical grade silicon that was refined by leaching prior to directional solidification and metallurgical grade silicon as-received from the manufacturer. The quartz molds with liquid silicon were extracted at four different velocities in the range from 5 to 110 mm/s. In the solidified silicon ingots, the macrosegregation of elements were measured using accurate analytical techniques, whereas the metallurgical structures were observed in macro and micrographs. The ingots solidified at lower extraction velocities, at which evidences of a planar solidliquid interface were observed, showed greater refining effect than those solidified at larger velocities, which showed evidences of a cellular or dendritic solid-liquid interface. The contents of Fe, Al, Cr, and Ti measured at the base and center portions of the solidified ingots comply with the specifications of SiGS proposed by some authors.

Alumina é um material amplamente aplicado em diversas áreas. Uma dessas aplicações utiliza a alumina amorfa e a alumina gama, como suporte catalítico devido suas elevadas áreas de superfície específica e defeitos em suas estruturas cristalinas. Porém, problemas com a estabilidade das aluminas metaestáveis em altas temperaturas se tornaram um desafio. Assim a combinação da alumina com alguns óxidos emergiu como uma nova classe de materiais catalíticos com melhor estabilidade quando comparada com a alumina gama ou alumina amorfa convencional. Entretanto, entender como se dá à estabilização das fases através desses aditivos se tornou outro desafio. Assim, este trabalho teve como objetivo avaliar os possíveis efeitos de aditivos nas propriedades termodinâmicas da alumina. Além de sua influência sobre a transição de fases amorfo-gama, ou seja, a influência sobre o balanço energético através da evolução do tamanho de partícula. Para isto foram sintetizadas aluminas com diferentes concentrações de Mg2+ e Zr4+ através do método proposto por Pechini. As amostras foram calcinadas e caracterizadas por Calorimetria Diferencial de Varredura, Análise de Superfície Específica pelo método de Adsorção-Desorção de Gases, Difração de Raios X, Espectrometria de Infravermelho por Transformada de Fourier e Picnometria. Os resultados das análises mostraram um aumento da estabilidade para as amostras dopadas com os aditivos, porém com distintas evoluções de área de superfície específica, evidenciando que o caminho adotado pelo aditivo quando inserido em um material matriz independe de sua concentração. / Alumina is a material applied in several areas. One such application uses the gamma alumina and amorphous alumina as catalyst support. The use of alumina as the catalyst support is because of their high specific surface areas and defects in their crystal structures. However, problems with the stability of metastable aluminas at high temperatures have become a challenge. To solve this problem, we use a combination of some oxides to alumina improves the stability of the material when compared with conventional catalytic aluminas. However, a new challenge, to understand the performance of these oxides in the stabilization of alumina. This work aimed to evaluate the possible effects of additives on the thermodynamic properties of alumina, through its influence on the amorphous-phase transition range. The assessment of influence on the energy balance was made by the evolution of particle size. For this alumina powders were synthesized with different concentrations of Mg2+ and Zr4+ using the method proposed by Pechini. The samples were calcined and characterized by Differential Scanning Calorimetry, Surface Analysis by the method of Specific Adsorption-Desorption of Gases, X-Ray Diffraction, Infrared Spectroscopy Fourier Transform and Picnometry. The analysis results showed an increased stability for the samples doped with additives, but with different evolutions of specific surface area, showing the path taken by the additive when inserted into a matrix material depend on its concentration.

Foram realizados ensaios de dilatometria em ligas não comerciais que apresentam microestrutura bainítica após o resfriamento contínuo. As variações de composição química foram realizadas sobre o aço destinado para construção mecânica AISI 5120 com adições de molibdênio, boro e nióbio. Os ensaios foram conduzidos no dilatômetro com atmosfera e temperatura controlada. No dilatômetro, foram aplicados resfriamentos contínuos em diferentes velocidades a partir da temperatura em que normalmente peças forjadas são reaquecidas. Tradicionalmente, a classe dos aços apresentados neste trabalho é exposta ao tratamento térmico de têmpera e revenimento e apresentam microestrutura martensítica. Com o objetivo de eliminar o tratamento térmico realizado pós conformação, foi proposto como substituição os aços bainíticos. Aços bainíticos não exigem tratamento térmico pós conformação e, apenas com a aplicação de um resfriamento controlado, é possível obter uma microestrutura que apresenta propriedades (tensão de escoamento e tenacidade) iguais ou melhoradas em relação ao material temperado e revenido. As microestruturas obtidas nas diferentes ligas resfriadas de maneira contínua foram caracterizadas de modo a estabelecer relações entre a velocidade de resfriamento e produtos formados, morfologia e fração de microconstituíntes. A caracterização microestrutural foi realizada de maneira intensiva de modo a relacionar desde propriedades magnéticas com padrões de difração de raios X das amostras para medição da fração de austenita retida. O objetivo deste trabalho foi investigar qual é a influência do molibdênio, boro e nióbio no resfriamento contínuo de aços bainíticos, bem como estabelecer o intervalo de velocidades de resfriamento em que é possível obter de maneira homogênea a estrutura bainítica. Após o resfriamento, os corpos de prova foram caracterizados por metalografia (microscopia óptica e eletrônica de varredura), dureza, saturação magnética, difração de raios x e EBSD. De fato foi verificado o efeito do molibdênio, boro e nióbio na cinética de decomposição da austenita no resfriamento contínuo e estabelecido relações entre a microestrutura obtida, velocidade de resfriamento e composição química. Foi observado também o efeito do molibdênio, boro e nióbio em evitar a transformação ferrítica para baixas velocidades de resfriamento de modo a obter uma estrutura bainítica sob um maior intervalo de resfriamento. / Dilatometry tests were carried out in a non commercial alloy that showed bainitic microstructure after continuous cooling from the austenitization temperature. The chemical composition variations were performed on a base chemical composition of a commercial steel (AISI 5120), additions were of molybdenum, boron and niobium. The tests were conducted on the dilatometer with atmosphere and temperature control. In the dilatometer, continuous cooling was carried out at different rates from the temperature in which the reheating of forged parts is usually performed. Traditionally, the steels used for this application are quenched and tempered and present a predominantly tempered martensite microstructure; bainitic steels were proposed as a substitution in order to eliminate further heat treatments after forging. The bainitic steels do not require post-conformation heat treatment: only with the application of a controlled continous cooling is possible to obtain a homogenous bainitic microstructure which has equal or improved properties (yield strength and toughness) comparing to quenched and tempered material. The microstructures obtained from the different alloys continuously cooled were characterized in order to establish relations between the cooling rate and formed products, morphology and volume fraction of phases. The microstructural characterization was carried out intensively and correlated with magnetic properties and X-ray diffraction patterns of the samples. The objectives of this work were to investigate the influence of molybdenum, boron and niobium on the continuous cooling of bainitic steels, as well as to establish the range of cooling rates needed in order to obtain an homogeneous bainitic structure. After cooling, the specimens were characterized by metallography (optical and scanning electron microscopy), hardness, magnetic saturation, x-ray diffraction and EBSD. The effect of molybdenum, boron and niobium on the kinetics of austenite decomposition in the continuous cooling was verified and relationships established between the microstructure, cooling rate and chemical composition. It was also observed the effect of molybdenum, boron and niobium in avoiding ferritic transformation at low cooling rates in order to obtain a bainitic structure under a longer cooling interval.

La thèse porte sur deux aciers emboutissables à chaud, soudés en configuration homogène ou hétérogène en termes de composition chimique et d'épaisseur. Les solutions en flans raboutés laser présentent en effet de remarquables performances à l'impact et connaissent un fort développement dans le contexte actuel de l'industrie automobile (réduction des émissions de CO2 et amélioration de la sécurité passive des véhicules). L'opération de soudage laser peut générer une hétérogénéité de la jonction soudée, due au mélange imparfait des deux matériaux et des conditions thermiques de solidification et d'emboutissage à chaud. Les propriétés mécaniques du joint soudé sont donc déterminées par la qualité du mélange, la composition chimique locale et le cycle thermique.La première partie de l'étude est consacrée à la quantification de l'hétérogénéité du joint soudé et la compréhension de la microstructure. Une nouvelle méthodologie a été développée pour étudier le comportement métallurgique et mécanique de ces joints hétérogènes, en élaborant des coulées synthétiques représentant les différentes compositions chimiques attendues dans le joint réel. Le comportement du joint soudé est reconstruit à partir des données obtenues par la caractérisation de ces coulées synthétiques. La deuxième partie de l'étude porte sur la compréhension de la formation de la microstructure de la zone fondue avant et après traitement thermomécanique simulant l'emboutissage à chaud. Des modèles quantitatifs ont été établis pour calculer les températures de changement de phase, les proportions relatives des constituants microstructuraux ainsi que la dureté de ces alliages en fonction de la composition chimique et du cycle thermique. La dernière partie de l'étude porte sur le comportement mécanique après traitement thermique. Des lois de comportement des matériaux synthétiques ont été déterminées en fonction de leur composition chimique et du cycle thermique. La sensibilité à la rupture fragile par clivage a été évaluée par des essais de traction sur éprouvettes sévèrement entaillées. Les lois de comportement obtenues ont été appliquées pour déterminer la tolérance du joint soudé à un défaut géométrique, sous sollicitation de traction uniaxiale. / This thesis is focused on the study of welded steel grades deformed by hot stamping. Typically, a laser welded blank is formed by butt joining two steel sheets, which can be identical or differ in chemical composition and/or sheet thickness. Such laser welded blank solution offers excellent crash performance, significant CO2 emission reduction and passive car safety thus it is of great interest for the automotive industry. The laser welding process may induce chemical and metallurgical heterogeneities arising from an incomplete mixture of the two base metals and from the thermal cycle conditions during welding and the subsequent heat treatment. The mechanical behaviour of the weld is therefore determined by the mixing quality, the chemical composition and the thermal cycle.The first part of the study aims at the quantification of the weld heterogeneity and the characterisation of the resulting microstructure. In this work, a new method is proposed to study the metallurgical and mechanical behaviour of those heterogeneous welds, by elaborating synthetic casted alloys representative of the different chemical compositions that can be locally found in the weld. The behaviour of the weld is then reconstructed from the results obtained from the characterisation of those synthetic casted alloys. The second part of the study concerns the understanding of the microstructure formation in the molten zone before and after a thermo-mechanical treatment which replicates the hot stamping conditions. Quantitative models have been established to calculate the phase transformation temperatures, the proportion of the principal constituents in the microstructure, and hardness of the weld as a function of chemical composition and thermal cycle parameters. The last part of the study is dedicated to the evaluation of the mechanical behaviour of the synthetic alloys after heat treatment. The constitutive material behaviour laws for each alloy have been identified. The sensitivity to brittle cleavage fracture has been assessed by tensile tests on severely notched bars. The constitutive material models are then applied to determine the sensitivity of the weld to geometrical defects under uniaxial loading.

De nos jours on utilise de plus en plus les aciers pré-revêtus dédiés aux applications pour emboutissage à chaud afin de protéger la surface de l’acier contre la décarburation et l’oxydation durant le traitement thermique. Le revêtement est déposé à chaud en continu par immersion de la bande d’acier dans un bain d’Al-Si. Pendant l’austénitisation le revêtement Al-Si se transforme par réactions d’inter-diffusion et de solidification. Ces réactions conditionnent la microstructure finale et en particulier la surface du revêtement, responsable des propriétés d’emploi telles que le soudage par point et l’adhérence peinture.A l’état de livraison le revêtement Al-Si contient les phases suivantes : des grains d’Al, l’eutectique ternaire Al-Fe-Si, une couche intermétallique ternaire Al-Fe-Si et une couche intermétallique binaire Al-Fe. Après austénitisation les phases se transforment en deux types d’intermétalliques : intermétalliques ternaires riches en Si et intermétalliques binaires pauvre en Si.Durant cette étude une identification complète des différentes phases a été établie. Les phases initialement riches en Si se transforment en phases fusibles par réactions eutectique ou péritectique. La présence de phase liquide accélère la diffusion locale de fer permettant aux phases riches en Si de garder leur contenu en Si pendant tout le traitement thermique. C’est la diffusion du fer qui stabilise les gradients en Si dans le revêtement, malgré les lois de diffusion classiques. En utilisant le diagramme de phases ternaire il a été démontré comment l’enrichissement en fer s’établit dans le revêtement tout en gardant l’équilibre aux interfaces entre les phases riches et pauvres en Si / Nowadays more and more pre-coated steels are applied in hot-stamping process to prevent the steel surface against iron oxidation and decarburization during the heat-treatment. The coating is deposited by hot-dipping the coil in an Al-Si bath. During the austenitization, the Al-Si coating transforms completely by inter-diffusion and solidification reactions, which define the final microstructure and particularly the top layer responsible for the in-use properties like spot welding and painting adhesion.At the delivery state, the Al-Si coating is a multiphase coating containing the following phases: Al-grains, Al-Fe-Si ternary eutectic phase, an Al-Fe-Si intermetallic layer and a binary Al-Fe interfacial layer. After the heat-treatment, all the phases transform into two types of intermetallic compounds: Si-rich ternary and Si-poor binary compounds.During this study, a complete identification of the different intermetallic phases has been conducted. Initial ternary Al-Fe-Si phases transform into fusible phases by eutectic or peritectic reaction. The presence of liquid state enables rapid local iron diffusion which allows to the Si-rich phases to keep their high Si content during the whole treatment. So the iron diffusion stabilizes the Si gradients in the coating despite the classical diffusion laws. Using the ternary phase diagram it was shown how iron enrichment in the coating proceeds by keeping thermodynamical equilibrium along the interfaces between Si-rich ternary and Si-poor binary phases. A phenomenological model of phase transformations explaining the formation of differente coating microstructures during the austenitization step is proposed

The present work has been produced as part of an on-going collaboration between the University of Manchester and Amec, with the primary aim of furthering mechanistic understanding of corrosion processes in zirconium alloys out-of-reactor. Zirconium alloys are used as cladding material for nuclear fuel pellets, and correct understanding of the corrosion process in autoclave is essential to predicting material behaviour in-reactor. This EngD thesis is composed of five proposed papers that investigate observations and hypotheses under the theme of mechanical degradation in oxides formed on zirconium alloys in autoclave. First investigations concern observed stress relaxation in zirconium oxide. Finite element analysis is used to capture mechanical aspects of the corrosion process and apply this to stress behaviour determined previously using synchrotron x-ray diffraction. The results indicate that a mechanism other than creep or hydrogen induced lattice strain must be present to account for the observed stress relaxation. One such potential mechanism is crack formation; statistical analysis of scanning electron microscopy images has been used to identify a link between the development of roughness at the metal-oxide interface, crack formation in the oxide and transition points or acceleration in the corrosion kinetics. Parameters such as the median radius of curvature and profile slope (Rdq) have been applied, as these parameters do not require the definition of a periodic wavelength or amplitude. These and other parameters are related to information in literature to indicate that for samples of Zircaloy-4 and ZIRLOTM, which go through transition, the interface roughness changes in a way that would increase localised stress concentrations. The third material is an experimental low tin alloy, which under the same oxidation conditions, and during the same time period, does not appear to go through transition and does not develop an interface roughness in the same way. A critical assessment of finite element analysis applied to oxidising non-planar interfaces shows the significant limitations in the existing mechanism for representing oxidation expansion and stress formation. Autoclave oxidation experiments of artificially roughened samples of Zircaloy-4 were carried out to further understand the impact of out-of-plane stress generation. The results indicate a divergence based on surface roughness after ~86 days oxidation. SEM examination of images in cross section highlighted accelerated oxidation above surface roughness peaks, and an increased crack area in rougher samples. Finally, finite element analysis of the tetragonal to monoclinic phase transformation showed that biaxial compressive stress relaxation, or the tri-axial tensile stress associated with an advancing crack tip, could reduce the transformation strain energy and destabilise the tetragonal phase. The volumetric expansion and shear strain associated with the phase transformation produces stress in the surrounding oxide sufficient to generate nano-scale cracks perpendicular to the metal-oxide interface. This would allow fast ingress routes for oxygen containing species, and therefore acceleration in the corrosion kinetics.

The current work contains experimental and theoretical work about the formation of martensite from the austenitic state of the steel Hardox 450. Simulation of rolling and subsequent quenching of martensitic steel plates requires a model that can account for previous deformation, current stresses and the temperature history, therefore dilatometry experiments were performed, with and without deformation. Two austenitization schedules were used and in the standard dilatometry the cooling rates varied between 5-100 °C/s, in order to find the minimum cooling rate that gives a fully martensitic microstructure. Cooling rates larger than 40°C/s gave a fully martensitic microstructure. The cooling rate of 100 °C/s was used in the deformation dilatometry tests where the uniaxial deformation varied from 5-50 %. The theoretical work involved modelling of the martensite formation and the thermal/transformation strains they cause in the steel. Characterizations were done using light optical microscopy, hardness tests and electron backscatter diffraction technique. The parent austenite grains of the martensitic structure were reconstructed using the orientation relationship between the parent austenite and the martensite. Kurdjumov-Sachs orientation relationships have previously been proven to work well for low-carbon steels and was therefore selected. The standard implementation of the Koistinen-Marburger equation for martensite formation and a more convenient approach were compared. The latter approach does not require the storage of initial austenite fraction at start of martensite formation. The comparison shows that the latter model works equally well for the martensite formation. The results showed that the use of martensite start and finish temperatures calibrated versus experiments for Hardox 450 works better when computing thermal expansion than use of general relations based on the chemistry of the steel. The results from deformation dilatometry showed that deformation by compressive uniaxial stresses impedes the martensite transformation. The simplified incremental model works well for deformation with 5 % and 10 %. However, the waviness in the experimental curve for deformation 50 % does not fit the model due vi to large barrelling effect and the large relative expansion for the material that the sample holders are made of. Crystallographic reconstruction of parent austenite grains were performed on a hot-rolled as-received reference sample and dilatometry samples cooled with 60 °C/s and 100 °C/s. The misorientation results showed that the samples match with the Kurdjumov-Sachs orientation relationship in both hot rolled product and dilatometry samples. When misorientation between adjacent pixels are between 15° and 48°, then the boundary between them was considered as a parent austenite grain. The austenitic grain boundaries of the sample cooled at 100 °C/s is in general identical with the hot rolled sample when considering high angle boundaries (15°-48°). The results from the hardness tests showed that the rolled product exhibits higher hardness as compared to samples cooled by 100 °C/s and 60 °C/s. This can be attributed to the formation of transition-iron-carbides in the hot rolled product due to longer exposure of coiling temperature.