In semiconductor devices interfaces play an important role. A typical interface occurring in semiconductor devices is the interface between dielectric and metal, which is necessary to guarantee the functionality of the device. Usually, the interface is between a metal or metal-based material, e.g. W, W(Ti) and TiN, and a silicate glass. The glass can be doped with different elements, e.g. boron and/or phosphorous. In this study, the interfaces between different metallizations on as-deposited and annealed borophosphpsilicate glass (BPSG) were investigated. 4-point-bending (4PB) experiments were used to perform the mechanical characterization; Auger electron spectroscopy (AES) and atomic force microscopy (AFM) were applied to identify and characterize the fracture surface of the 4PB samples. Energy dispersive X-ray spectroscopy (EDX) in the transmission electron microscope (TEM) was employed to analyse the chemical composition of intact interfaces. Another method that was further developed to characterize the mechanical properties of the interfaces was an in-situ TEM submicron-sized bending beam approach. The BPSG films revealed different mechanical behaviour after different process steps. The as-deposited BPSG showed a pronounced effect on the adhesion of W with 20 at% Ti (from here on W(Ti)) with and without a Ti interlayer. When the W(Ti) was directly applied on the as-deposited BPSG the evaluated interface energy release rate was half of the one which was achieved if a Ti interlayer was deposited in between the W(Ti) and the as-deposited BPSG. This difference, for the two systems with and without Ti interlayer, disappeared, when annealed BPSG is used, instead of the as-deposited one. There the Ti interlayer did not have any further beneficial effect. The variation of the Ti content between 15 to 25 at% in the W layer showed no pronounced effect on the adhesion of W(Ti) on annealed BPSG. All studied cases of different Ti contents in the W(Ti) layer gave almost the same interface energy release rate, which was the same as for a pure Ti layer on annealed BPSG. It was also possible to reveal that the Ti incorporated in the W layer promoted the adhesion significantly compared to a pure W layer on annealed BPSG. The EDX measurements in the TEM on an intact interface indicated a Ti enrichment at the interface between W(Ti) and annealed BPSG. The main reason for this finding is a 400 °C anneal at the end of the sample production. The AES and AFM investigations of the fracture surface of the 4PB samples revealed Ti residues on the BPSG fracture surfaces of all material systems containing Ti. These residues seem to be Ti-based particles which were ripped out of the metallization side of the interface of interest. The chosen interfaces to investigate the influence of the sample size on interfacial adhesion were Cu and W(Ti) on annealed BPSG, respectively. For the Cu/BPSG interface the 4PB experiments indicated a weaker interface than for the W(Ti)/BPSG. In addition, in-situ TEM micro-bending beam experiments showed, taking the difference in sample dimension into account, good agreement with the macroscopic 4PB tests, for the Cu/BPSG system. In case of the W(Ti)/BPSG samples the interface was too strong to be tested with the proposed in-situ TEM bending beam setup. This study shed some light on selected interfaces between metals and BPSG. Furthermore, it was revealed that it is possible to compare the results of different methods over several length scales. This provides the possibility to test semiconductor structures nearer to the device size and it will be possible to combine very localized chemical and mechanical information of the interface.