| dc.description.abstract | The effect of Ni and Cu additive on electromigration(EM) in Sn solder joints and lines have been studied. Both EM test samples were stressed under the current density of 104 A/cm2. The electromigration behaviors in Sn0.7Ni solder joints were investigated at 155℃, 180℃, 200℃, respectively. We found that the Cu additives were transported to the anode side and formed Cu-Sn intermetallic compound. For Ni doped Sn solder joint, a thin layer of intermetallic compound (IMC) formed at the interface which was identified to be the Ni-Cu-Sn ternary compound phase. Under EM test, we found that for the pure Sn solder, the interfacial Cu-Sn compound layer at the cathode side will dissolve into Sn. However, for the Sn0.7Ni solder joint, remarkablely, we found that the Ni-Cu-Sn ternary compound layer is more stable than the Cu-Sn compound layer under EM test. Another intriguing finding is that a line of voids (Kirkendall voids) occurred at the interface between IMC and Cu pad. From the FIB examination, the Kirkendall voids, which at the anode side is more serious than that at the cathode side. Depending on the direction of electron flow, two cracking failure modes were observed. One is on the anode side, the cracking was occur at the IMC / Cu interface, the other is on the cathode side, the cracking was occur at the IMC / solder. Comparing the activation energy between Sn0.7Ni alloy (Q= 0.85 eV/ atom ) and Sn3Ni alloy (Q= 1.13 eV/ atom ), we know that, under the EM current stressing, the Sn3Ni has the highest activation energy than Sn0.7Ni, therefore the Cu consumption rate of Sn3Ni alloy is the slowest. The reason for Sn3Ni alloy has the slowest Cu dissolution rate than Sn0.7Ni than pure Sn (Q= 0.68 eV/ atom, reference data from Ellen Ge). The reason for the lower Cu consumption rate is mainly due to the formation of (Cu1-yNiy)6Sn5 can induced a lower Cu consumption rate[33]. The first reason is that (Cu1-yNiy)6Sn5 has a lower growth rate and is an effective diffusion barrier under the EM stressing. Otherwise, the (Cu1-yNiy)6Sn5 also is much stable under the EM stressing, as aforementioned. The second reason is that, from the growth kinetic of data for the IMC, as seen the Fig 5.2~5.5, we found that the thickness of IMC of Sn3Ni alloy is much smaller than the thickness of IMC of Sn0.7Ni alloy. Considering the mass balance of Cu, the amount of Cu consumed is equal to the amount of Cu dissolved into the solder, and incorporated with the IMC compounds. The formation IMC thickness of Sn3Ni alloy is less than the IMC thickness of Sn0.7Ni alloy, therefore, the Cu consumption rate in the Sn3Ni alloys is the slowest than in Sn0.7Ni and even in pure Sn under the EM current stressing.
Using solder lines structure, we have studied the electromigration phenomenon on Sn (Cu) alloys and Sn (Ni) alloys lines, which are Sn, Sn0.7Cu, Sn3Cu, Sn0.7Ni, Sn3Ni. Two EM test temperature were at the room temperature and 60℃. The samples were stressed under the current density of 104 A/cm2. Mass accumulation near the anode and void nucleation near the cathode were observed during the current stressing. The electromigration rates of the above alloys were determined by knowing the extrusion where at the anode side. Our results show that the magnitude were Sn0.7Cu＞Sn＞Sn3Cu＞Sn0.7Ni＞Sn3Ni. We known that Sn0.7Ni alloy and Sn3Ni alloy shows much higher electromigration resistance than Sn (Cu) alloys. Furthermore, the grain boundary diffusion is known to be the main kinetic path of atomic transport in EM. From the Fig 5.12, we know that the Sn (Ni) alloy has a larger grain size than Sn (Cu) alloy, compare from the Table III, we know that the grain size of Sn (Cu) alloys just 3~10μm, however, the grain size of Sn (Ni) alloy is almost 20μm, it’s means that the Sn (Ni) alloy has the less grain boundary density and the electromigration’s diffusion path is decrease. Therefore, the Sn (Ni) alloys can retard the electromigration than the Sn(Cu) alloys. However, a more detail study is needed to verify this suggestion. Otherwise, we can estimate the accumulation region, the atomic flux driven by the electromigration can be calculated. Then the effective charge number and DZ* value of the different solder alloy lines were obtained. The DZ* values agree with the literature data. | en_US |