Conventional IC packages form a rigid shell around silicon IC dies. Their purpose is to provide environmental protection, electrical interconnect and heat dissipation. Despite the fact that majority of current silicon IC?s are realized in a very thin top layer of the silicon substrate (<10µm), the typical thickness of packaged IC dies generally exceeds 150 µm. Continuous system miniaturization and performance improvement leads to new mass volume applications where packaging technology has to be reviewed. Here only the essential part of silicon IC?s i.e. the 10-20 µm thick top layer could be retained after thinning of the wafer. In the wafer thickness range of 10–30 ?m, silicon substrates become mechanically flexible and consequently offer a large field of new products and innovative applications. A promising development of flexible and stretchable substrates is proposed. 3D deformable electronics could be realized by the vertical thinning and lateral partitioning of the silicon substrate on sub-millimeter scale. The partitions or so-called segments can be combined to larger electronic systems by connecting many of these through electrical bridges. By varying the dimensions and/or the geometry of the segments and the gap size in between the segments as well as the geometry of the electrical bridges, the level of deformations can be controlled. In practical realization such patterned silicon structures have to be embedded/sandwiched into a polymer film to provide environmental protection and to prevent mechanical damage because of overstretching. In order to evaluate the influence of segment size and gap size on the occurrence of failure under bending and stretching, so-called 1st generation flexible and stretchable test samples were designed and prepared. The test samples being considered have hexagonal or square segments (varying in size from 150 to 2000 ?m) being embedded in polyimide. Special tensile and bending test tools were designed and fabricated to in situ observe the occurrence of cracks during loading. An optical microscope with the possibility of recording and analysing the digital images is used for establishing the crack density and width. Experimental and simulation results for the onset of cracking are quite disappointed. It is shown that the first cracks appear in the oxide layers in the gaps in between the silicon segments. The crack density appears to increase rapidly at early stage of loading and subsequently increases slightly. However, the width of the cracks appears to increase steadily during loading. Only at higher (mean) deformations the cracks propagate (or are generated) within the silicon itself. The onset of cracking depends significantly on the silicon segmentation size. The segment size and gap size also affects the crack density and the crack width at larger (mean) deformation levels. There is no crack detected for bending around glass cylinders (even not for the cylinder with the smallest diameter, ?= 2 mm) for samples with a square segment with 450 ?m side length and 120 ?m gap size and for samples with a hexagonal segment with 300 ?m side length and 40 ?m gap size. The remaining bending results show that for other samples with square segments failure always occurs for bending around the glass cylinder with the smallest diameter (?= 2 mm). From the tensile testing as well as from the simulation results we learned that occurrence of cracks in the oxide layers severely limits the stretchability of the substrates. Because of the early damage initiation found for the 1st generation samples, a modified design was proposed and worked out. So-called 2nd generation samples were designed and fabricated with fully segmented polycrystalline silicon segments with flexible aluminium interconnections which are supported by flexible poly-silicon support structures. Again polyimide was used as the embedding material. The samples being considered have varying segment sizes (from 150 to 450 ?m) and varying gap sizes (from 20 to 200 ?m). Various (more or less) sinusoidal interconnections were chosen with various numbers of half waves and various wave amplitudes. When the samples were bent around the (smallest) cylinder with 2 mm diameter, no damage of the segments was detected. Resistance measurements did not show a resistance increase larger than 5%. Compared to the 1st generation samples, for tensile testing of the 2nd generation samples the (mean) strain at onset of failure (which now is segment cracking) is significantly improved. In order to gain more insight into the occurrence of interconnection failures various FE simulations were performed for wave-shaped interconnections of samples with square segments (under stretching only). The local model used is made up from a single gap (of polyimide) with metallic interconnection and poly-silicon support structure in between two embedded segments. Comparison of the experimentally obtained strain values for the resistance change of 5% and the “sample mean strain” at (assumed) onset of failure, did not give a good match. Apparently the assumed onset of failure, defined by reaching the ultimate strength in the aluminium (that only occurs at some local) is not a good measure for the degradation of the electric conductivity. The “work of plastic deformation” might be better correlated to the change in resistance. The sinusoidal wave interconnection shows the best electrical performance compared to the straight interconnection and the semi-circular interconnection. The influence of wave amplitude, number of half waves and line width of the sinusoidal interconnection is explored. However, the sample mean strain at onset of interconnection failure appears to be limited to a few percents only. From both the experiments and the interconnection FE simulations it is concluded that again insufficient flexibility is obtained for all considered interconnection shapes. It is believed that this is caused by the embedding of the segments and interconnections within the polyimide. For this reason in Chapter 4 embedding in a much softer material is worked out. Also the case of a completely free interconnection (not embedded) is considered (in Chapter 5 ). In this manner a new concept of future flexible and stretchable substrates is introduced. In the concept of “Future flexible and stretchable substrates I” (Chapter 4), the segments and interconnections are fully embedded into ELASTOSIL RT 601 (a kind of silicone rubber, from now to the whole thesis, the silicone rubber is ELASTOSIL RT 601). Adequate material models for silicone rubber are essential for getting insight into the mechanical behavior of the new design through FE modeling. In particular, FE modeling is used to forecast possible failure. The mechanical properties of silicone rubber were characterized by various methods including tensile testing, cyclic tensile testing and DMA. The ultimate tensile elongation of the silicone rubber foil can reach about 176% at room temperature. Visco-elastic behaviour of the silicone rubber at room temperature is not relevant. The 3rd order Mooney model was selected for the constitutive description of the silicone rubber for the FE simulations. Based on the FE simulation results for the “future flexible and stretchable substrate I” it is expected that when increasing the mean sample strain, first the Si support structure will fail and subsequently the silicone rubber will fail during tensile loading. Failure of the Si-segments is likely not to occur at all. Compared to the 2nd generation substrates, the concept the “future flexible and stretchable substrate I” only gives an improvement of (about) a factor 2 for the main strain level at failure occurrence. The limiting factor for the improvement is the disappointing behavior of the Si-support structure. Apparently, the embedment of the Si-support structure by rubber very much reduces the “spring” behavior of the sinusoidal support structure. Therefore, a major improvement is suggested for the “Future Flexible and Stretchable Substrate II”, by not completely embedding the interconnection by silicone rubber, but only sandwiching this structure between two silicone rubber foils. For the concept of “Future Flexible and Stretchable Substrate II” the aluminum wave interconnections are replaced by copper wave interconnections because the better mechanical and electronic performance of copper. Free-standing interconnection copper lines (without support structures), sandwiched in between silicone rubber sheets, connect the Si-segments. Three types of the interconnections, meander shaped, horseshoe shaped and meshed shaped, were designed with various parameter sets. Simulations for these basic parts were performed to evaluate the influence of the geometric parameters on the flexibility and stretchability. The free-standing interconnection shapes and their geometric parameters have significant influence on the stretchability. From the three types of interconnections being considered, the meander shaped design appears to be most favourable. Compared to the results for the “Future Flexible and Stretchable Substrate I” it can be concluded that an enormous improvement of the stretchability of the interconnect structure is found. It is realized that because of the enormous flexibility of the meander shaped interconnection design, the maximum mean strain of the substrate is limited by the maximum mean strain that other parts can withstand. Here it should be noted that the maximum mean strain of the silicone rubber sheets is limited to about 176%, or less. With this elongation limit the (found) most favourable meander interconnection (W=5 ?m, R=100 ?m and ?=30 degree) will behave fully elastic and thus will not be damaged, even not under cyclic elongation.