Damage Evolution and Mechanical Failure in Flip-chip Interconnects
- PDF / 1,510,670 Bytes
- 6 Pages / 417.6 x 639 pts Page_size
- 96 Downloads / 216 Views
ABSTRACT The analysis of the mechanical behaviour of flip-chip interconections is an essential aspect of the reliability of the devices using this technique. When manufacturing them and during their operation, the interconective systems are submitted to thermal and mechanically induced stresses. By cyclic micromechanical testing and from analysis by micro Raman of singularities, we studied the mechanical response and the failure of samples, consisting of a substrate, the interconections and a chip. Two sort of low melting point alloys were investigated, with different metallurgies for the pads, holding the interconections. INTRODUCTION The mechanical stability of flip-chip MCM components depends to their in service solicitations, such as thermal variations or mechanical vibrations [1,2,3]. Exerted cyclically, these types of solicitations, can produce two sorts of fatigue effects. On the one hand, thermal cycling may produce interfacial or grain boundary segregations (with or without precipitation); on the other hand, the cyclic mechanical stresses can activate microcracking or coalescence of voids, leading to catastrophic crack propagation and complete failure of the device. In a previous study, we analysed the mechanical response of flip-chip bonded samples submitted to simple tensile and shear experiments [4]. It was demostrated then that, depending on the sort of metallurgy at the interface of the bump with respect to the chip, the failure occured when an intermetallic precipitation was produced after remelting the bump during the assembling procedure. This contribution corresponds to an extension of the previous study, analysing here the mechanical response of flip-chip bonded systems submitted to mechanical fatigue conditions. Previously it was shown that depending on the different bump alloys, and on the metallurgy of the interfaces, the evolution of the alloy to failure showed different regimes. We point out here the importance of preexisting stress concentrations and/or defects near the interfaces of the solder pads, which will weaken the mechanical response and drive to unexpected brittle ruptures. EXPERIMENTAL METHODS The samples used for the mechanical testing measured 8mm by 16 mm. They consisted of 776PbSn solder joints, between a chip and a substrate, both in Silicon. These solder joints presented a truncated spherical shape, of 30gm to 50 gm high, and a mean diameter of 70gm. They had solder pads on chip and substrate sides, consisting of layers of Ti and Ni, deposited directly onto the Si0 2 covering the surface of the Si chip or substrate. In addition to these layers, a gold layer is added to protect the nickel layer from oxydation, during the fabrication of the module. The thickness of the three solder pad layers (Ti, Ni and Au) are on the order of thousands of angstroms. The experiments were performed on flip-chip samples, presenting two compositions of Pb and Sn for the solder bump: 37Wt% Pb - 63Wt% Sn (also called Pb40Sn60) and 95 Wt% Pb 5Wt% Sn (also called Pb95Sn5) and three different metallu
Data Loading...
