A bonding process using CuO paste, containing CuO particles and Polyethylene glycol 1000 (PEG1000) as a reducing solvent, was achieved Cu-to-Cu bonding in the air through reduction of CuO by the reducing solvent. The reduction process of CuO paste during heating process was investigated with simultaneous measurements of X-ray diffraction (XRD) and differential scanning calorimeter (DSC) and Fourier-transform infrared spectroscopy (FT-IR). Reduction process from CuO to Cu2O occurred at around 250 ℃ and Cu2O to Cu subsequently occurred at around 315℃. The difference of shear strength of joints bonded at 300℃ and at above 350℃ was caused by the generation of Cu in terms of the reduction process of CuO paste. Results of FT-IR analysis clarified the reduction process of CuO and Cu2O by PEG1000. The bonding strength of joints bonded at 400℃ for 15 min with the pressure of 15 MPa was higher than that of conventional Pb-5Sn solder.
In die attach process instead of preforms; the need of a new solder paste technology is required, which should operate better in formic acid reduction reflow. This article discusses the development and validation of such solder paste. To achieve such solder paste, a cavity between solder powders in solder paste has been developed by using a special heat resistant agent. A combination of optimized formic acid delivery process and reflow process, formic acid can penetrate deeper into the solder paste by the use of its cavities than conventional pastes. Consequently, wettability towards large surface area with low void rate can be assured as how conventional solder preform acts. In addition, very low flux residue is advantageous which retains heat resistance. Thus, wire bonding and resin molding can be done without cleaning the residue and able to obtain high reliability.
Reliability issue on bonding structures of power modules becomes more important due to new products which consist of
compound semiconductors and can operate under higher temperature condition than conventional products. Higher temperature condition leads to severer non-linear deformation of component materials. In particular, both plastic and creep deformation that are dominant factors of fatigue, drastically appear in bonding materials under the temperature close to a half of melting point. In this paper, thermal fatigue of wire bonding structure was investigated with non-linear finite element analysis considering plastic and creep deformation. Results of the analysis suggest that inelastic strain energy density can describe thermal fatigue behavior, especially the saturation of fatigue lifetime over 200℃, rather than the range of inelastic strain.
Photodefinable wafer-level underfill (PWLUF) has been developed to avoid the underfill entrapment which causes mechanical and electrical reliability issues with fine bumps. In this paper, firstly, we report the appropriate stacking profile to obtain voidfree stacks, explaining the PWLUF process flow and the melt viscosity after patterning. Secondly, the lower pressure soldering on thermal compression and collective reflow soldering without any underfill entrapment were demonstrated. It means that this PWLUF was designed to have enough low viscosity after photo-curing for patterning. The stacked sample has enough high adhesive strength even after moisture absorption and passed moisture sensitivity level 2 (JEDEC MSL2). Finally, we demonstrated the compatibility with Cu-Cu bonding application comparing with conventional no-flow type underfills (NUF).
This paper addresses how to predict the S-N curves for the fatigue fracture of neck area in aluminum wires. Because the conventional prediction methods for S-N curve were based on the assumption of the thick bulk samples, the applicability of the methods to the thin aluminum wires was examined by comparing the predicted S-N curve with fatigue test results. The results indicated that the conventional prediction methods for S-N curve were applicable to the fatigue fracture of neck area in the aluminum wires. Especially, Manson’s universal slope method predicted the S-N curve with good accuracy.