The knowledge of diffusion parameters provides important understanding of many physical and mechanical properties of materials. In most of the applications silicides are grown by a diffusion controlled process mainly in thin film condition. Because of this reason, most of the studies till date are available in thin film condition. Although more than one phase is present in all these systems, mainly disilicides were found at the interface. In this thesis bulk interdiffusion studies are conducted by coupling pure refractory metals (group IVB, VB and VIB elements) with single crystal Si. Several phase layers grow between binary refractory metal and Si systems. The layer thicknesses of the phases are measured from the microstructures. Composition profiles were measured in electron probe microanalyzer. Different diffusion parameters are estimated such as parabolic growth constants, integrated diffusion coefficients, activation energy for diffusion and ratio of tracer diffusivities of the components are estimated. Growth mechanisms of the phases are discussed with the help of diffusion parameters. The atomic mechanism of the diffusion is discussed considering crystal structure of the phases along with possible defects present. Solid diffusion couple experiments are conducted to analyse the growth mechanism of the phases and the diffusion mechanism of the components in the Ti-Si system. The calculation of the parabolic growth constant and of the integrated diffusion coefficients substantiates that the analysis is intrinsically prone to erroneous conclusions if it is based just on the parabolic growth constants determined for a multiphase interdiffusion zone. The location of the marker plane is detected based on the uniform grain morphology in the TiSi2 phase, which indicates that this phase grows mainly because of Si diffusion. The growth mechanism of the phases and morphological evolution in the interdiffusion zone are explained with the help of imaginary diffusion couples. The activation enthalpies for the integrated diffusion coefficient of TiSi2 and the Si tracer diffusion are calculated as 190±9 and 170±12 kJ/mol, respectively. The crystal structure, details on the nearest neighbours of the elements and the relative mobilities of the components indicate that the vacancies are mainly present on the Si sublattice. Diffusion controlled growth of the phases in the Hf-Si and Zr-Si are studied by bulk diffusion couple technique. Only two phases grow in the interdiffusion zone, although several phases are present in both the systems. The location of the Kirkendall marker plane detected based on the grain morphology indicates that the disilicides grow by the diffusion of Si. Diffusion of the metal species in these phases is negligible. This indicates that vacancies are present mainly on the Si sublattice. The activation energies for integrated diffusion coefficients in the HfSi2 and ZrSi2 are estimated as 394 ± 37 and 346 ± 34 kJ/mol, respectively. The same is calculated for the HfSi phase as 485±42 kJ/mol. The activation energies for Si tracer diffusion in the HfSi2 and ZrSi2 phases are estimated as 430 ± 36 and 348 ± 34 kJ/mol, respectively. We conducted interdiffusion studies to understand the atomic mechanism of the diffusing species and the growth mechanism of the phases. Integrated diffusion coefficients and the ratio of tracer diffusion coefficients were estimated for these analyses. The activation energies for the integrated diffusion coefficients were calculated as 550 ± 70 and 410 ± 39 kJ/mol in the TaSi2 and the Ta5Si3 phases, respectively. In the TaSi2 phase, Ta has a slightly lower but comparable diffusion rate with respect to Si, although no TaTa bonds are present in the crystal. In the Ta5Si3 phase, Si has higher diffusion rate, which is rather unusual, if we consider the atoms in the nearest-neighbor positions for both the elements. The ratio of Si to Ta tracer diffusion coefficients is found to be lower in the Si-rich phase, TaSi2, compared to the Si-lean phase, Ta5Si3, which is also unusual. This indicates the type of structural defects present. An analysis on the growth mechanism of the phases indicates that duplex morphology and the Kirkendall marker plane should only be present in the TaSi2 phase. This is not present in the Ta5Si3 phase because of the very high growth rate of the TaSi2 phase, which consumes most of the Ta5Si3 phase layer. The problems in the calculation method used previously by others in this system are also explained. Experiments are conducted in the W-Si system to understand the diffusion mechanism of the species. The activation energies for integrated diffusion are found to be 152±7 and 301±40 kJ/mol in the WSi2 and W5Si3 phases, respectively. In both the phases, Si has a much higher diffusion rate compared to W. The result found in the WSi2 phase is not surprising, if we consider the nearest neighbors in the crystal. However, it is rather unusual to find that Si has higher diffusion rate in the W5Si3 phase, indicating the presence of high concentration of Si antisites in this phase. In the group IVB, VB and VIB M-Si systems are considered to show an interesting pattern in diffusion of components with the change in atomic number in a particular group. MSi2 and M5Si3 are considered for this discussion. Except in the Ta-Si system, activation energy for integrated diffusion of MSi2 is always lower than M5Si3. Interestingly, in both the phases, the relative mobilities measured by the ratio of tracer D* diffusion coefficients, S i decreases with the increase in atomic number in both the DM* groups. Both the phases have similar crystal structures in a particular group in which these parameters are calculated. In both the phases Si has higher diffusion rate compared to M. Absence of any M-M bonds in MSi2 and increase in the diffusivities of M with the increase in atomic number substantiates the increasing concentration of M anti-sites and higher interactions of M with vacancies. Only one or two Si-Si bonds are present in M5Si3, however, the higher diffusion rate of Si indicates the presence of vacancies mainly D* on its sublattice. On the other hand, increase in S i with increasing atomic number in DM* Both the groups substantiates increasing interactions of M and vacancies.

The electro–mechanical connection between under bump metallization (UBM) and solder in flip–chip bonding is achieved by the formation of brittle intermetallic compounds (IMCs) during the soldering process. These IMCs continue to grow in the solid–state during storage at room temperature and service at an elevated temperature leading to degradation of the contacts. In this thesis, the diffusion–controlled growth mechanism of the phases and the formation of the Kirkendall voids at the interface of UBM (Cu, Ni, Au, Pd, Pt) and Sn (bulk/electroplated) are studied extensively. Based on the microstructural analysis in SEM and TEM, the presence of bifurcation of the Kirkendall marker plane, a very special phenomenon discovered recently, is found in the Cu–Sn system. The estimated diffusion coefficients at these marker planes indicate one of the reasons for the growth of the Kirkendall voids, which is one of the major reliability concerns in a microelectronic component. Systematic experiments using different purity of Cu are conducted to understand the effect of impurities on the growth of the Kirkendall voids. It is conclusively shown that increase in impurity enhances the growth of voids. The growth rates of the interdiffusion zone are found to be comparable in the Cu–Sn and the Ni–Sn systems. EPMA and TEM analyses indicate the growth of a metastable phase in the Ni–Sn system in the low temperature range. Following, the role of Ni addition in Cu on the growth of IMCs in the Cu–Sn system is studied based on the quantitative diffusion analysis. The analysis of thermodynamic driving forces, microstructure and crystal structure of Cu6Sn5 shed light on the atomic mechanism of diffusion. It does not change the crystal structure of phases; however, the microstructural evolution, the diffusion rates of components and the growth of the Kirkendall voids are strongly influenced in the presence of Ni. Considering microstructure of the product phases in various Cu/Sn and Cu(Ni)/Sn diffusion couples, it has been observed that (i) phases have smaller grains and nucleate repeatedly, when they grow from Cu or Cu(Ni) alloy, and (ii) the same phases have elongated grains, when they grow from another phase. A difference in growth rate of the phases is found in bulk and electroplated diffusion couples in the Au–Sn system. The is explained in AuSn4 based on the estimated tracer diffusion coefficients, homologous temperature of the experiments, grain size distribution and crystal structure of the phase. The growth rates of the phases in the Au–Sn system are compared with the Pd–Sn and the Pt–Sn systems. Similar to the Au–Sn system, the growth rate of the interdiffusion zone is found to be parabolic in the Pd–Sn system; however, it is linear in the Pt–Sn system. Following, the effect of addition of Au, Pd and Pt in Cu is studied on growth rate of the phases. An analysis on the formation of the Kirkendall voids indicates that the addition of Pd or Pt is deleterious to the structure compared to the addition of Au. This study indicates that formation of voids is equally influenced by the presence of inorganic as well as organic impurities.