Electroless copper plating is an important process to improve the electrical reliability at the bottom of the via. In this study, the effects of nickel ions added to the electroless copper plating bath on throwing power, electrical conductivity, and the recrystallization of electro copper plating films were investigated. When the electroless copper plating film to be the seed layer had high purity and low film thickness, the crystal grains of the inner layer copper and the electro copper plating showed continuity with the recrystallization at room temperature. The co-deposition of nickel in the electroless copper plating film increased the sheet resistance and hindered the continuity with the crystal grains of the inner layer copper. When continuity with the crystal grains of the inner layer copper occurred, the solder heat resistance test showed high electrical reliability.

抄録全体を表示

Power modules are used for all types of electric power control and key devices for energy saving. One of the important parts for a power module is the wire-to-semiconductor chip joint. This joint is subjected to repeated temperature cycles, and repeated thermal stress acts on the joint due to the difference in coefficient of linear thermal expansion between a wire and a die material. Wire-liftoff is the thermal fatigue caused by such repeated thermal stress. In recent years, power modules are expected to be used at high temperatures of 200℃ or higher. Finite element analysis considering both creep and plastic behavior is a powerful tool for evaluating the wire-liftoff lifetime. Stress reversal occurs in a bonding wire under a repeated thermal stress condition. It is known that the modified strain hardening rule should be utilized in the transient creep analysis instead of the conventional strain hardening rule. In this paper, we performed the fund amental study on the application of the modified strain hardening rule to wire-liftoff phenomenon for an Al wire-Si die system.

抄録全体を表示

To investigate the mechanisms by which fatigue crack networks form in die-attach joints in power semiconductors, high-speed thermal cycling test was performed using a Si/Sn-Ag-Cu/Si specimen and the formation process of fatigue crack networks in the solder layer was observed. Fatigue cracks were found to emerge around intermetallic compounds in the β-Sn dendrite boundaries or from the high-angle (high-Σ) grain boundaries of β-Sn generated by continuous dynamic recrystallization. In all of these cases, the subsequent cycles caused the individual cracks to propagate in a cross shape and to become connected to each other, resulting in the formation of fatigue crack networks. Finite element method (FEM) analysis confirmed that the solder layer was in a state of equibiaxial tensile and compressive creep in the direction parallel to the joint surface of the solder layer during high-speed thermal cycling. FEM analysis also indicated that equibiaxial tensile creep is the driving force behind fatigue fractures. FEM analysis results for the cross-shaped micro-cracks confirmed that equibiaxial tensile creep in the period of decreasing temperature caused the cross-shaped cracks to open and propagate. Further, these propagated cracks became connected to each other to form fatigue crack networks.

抄録全体を表示

We studied a method of low temperature Cu-Cu quasi-direct bonding inserting ultra-thin metal layer deposited by atomic layer deposition（ALD）. A thin Pt layer was selected as a metal intermediate layer because Pt can be easily diffused into Cu. Area selective deposition of Pt on Cu surface was achieved without any mask using ALD. The Cu-Cu bonding using an ultra-thin Pt interlayer realized enough shear strength of 9.5 MPa, which was 5.1 times larger than without using the Pt interlayer. The Pt intermediate layer promoted atomic diffusion on the bonding interface at 300℃, which can improve the shear strength of the Cu-Cu bonding. The proposed method is a useful for low-temperature mounting, which is an issue for 2.5D/ 3D mounting.

抄録全体を表示

In mixed-model production, it is important to determine a proper production sequence which minimizes deterioration of production efficiency caused by variation of required processing times among the models. Properness of a production sequence depends on part delivery planning also, because it may be impossible to deliver a required part to its destination by the starting time of the process which requires the part. For this reason, integration of sequencing and part delivery planning has been discussed. This paper provides an efficient method for solving the integrated planning problem in which the problem is formulated as a 0-1 mixed integer programming and solved using both genetic algorithm and linear programming. The proposed method was applied to some numerical examples and it was possible to find a solution which is almost optimum for a small size problem and a reasonable solution for a large size problem.

抄録全体を表示