No abstract

No abstract

A novel thermosonic (TS) bonding process for gold wire bonded onto chips with copper interconnects was successfully developed by depositing a thin, titanium passivation layer on a copper pad. The copper pad oxidizes easily at elevated temperature during TS wire bonding. The bondability and bonding strength of the Au ball onto copper pads are significantly deteriorated if a copper-oxide film exists. To overcome this intrinsic drawback of the copper pad, a titanium thin film was deposited onto the copper pad to improve the bondability and bonding strength. The thickness of the titanium passivation layer is crucial to bondability and bonding strength. An appropriate, titanium film thickness of 3.7 nm is proposed in this work. One hundred percent bondability and high bonding strength was achieved. A thicker titanium film results in poor bondability and lower bonding strength, because the thicker titanium film cannot be removed by an appropriate range of ultrasonic power during TS bonding. The protective mechanism of the titanium passivation layer was interpreted by the results of field-emission Auger electron spectroscopy (FEAES) and electron spectroscopy for chemical analysis (ESCA). Titanium dioxide (TiO 2 ), formed during the die-saw and die-mount processes, plays an important role in preventing the copper pad from oxidizing. Reliability of the high-temperature storage (HTS) test for a gold ball bonded on the copper pad with a 3.7-nm titanium passivation layer was verified. The bonding strength did not degrade after prolonged storage at elevated temperature. This novel process could be applied to chips with copper interconnect packaging in the TS wire-bonding process.

One of the important factors affecting the accuracy of stress values obtained from the hole-drilling method is the calibration coefficient. A three-dimensional model was established to determine the calibration coefficients for integral method. The constraint conditions and loading conditions during hole-drilling can be simulated more realistically with this method. With this new model, coefficients a¯i,j and b¯i,j could be determined within one computation procedure. The relationship between calibration coefficients and plate thickness was investigated over a wide range of plate thickness. It has been found that the calibration coefficients determined in this work may vary with thickness of plates and the thickness range for thin plates was thus well defined. The calibration coefficients can thus be extended to measure the residual stresses of either thin or thick plates. Comparison of calibration coefficients with those determined by other studies was also conducted.

No abstract