| Summary: | 碩士 === 國立成功大學 === 材料科學及工程學系 === 102 === SUMMARY
The present study investigated the bonding behaviors and interfacial reactions in the solder joints of TCNCP (Thermal Compression Non-conductive Paste) process. The bonding behavior was affected by the physical properties of the NCPs. The results showed that the bonding effect between solder and Cu trace could be improved by the usage of NCPs with higher gel temperature. To improve the bonding effect between solder and Cu trace, the removal of the OSP (Organic Solderability Preservative) layer on the Cu trace surface was also investigated with Ar/O2 plasma, hydrochloric acid and formic acid. Interfacial reactions between solder and Cu were investigated by SEM-EDX to analyze the reactions within the Cu pillar/solder/Cu trace structure during TC bonding. The result revealed that a uniform layer of Cu3Sn was formed at the solder/Cu trace interface. However, the growth of Cu6Sn5 might be suppressed due to the entrapment of NCP fillers at the Cu trace interface. This study further investigated the effects of pre-condition and TCT (Thermal Cycle Test) tests on the interfacial reaction. After the pre-con test, the temperature difference between Cu pillar and Cu trace would lead to the asymmetric growth of IMCs. During the TCT, Cu6Sn5 on both Cu pillar and Cu trace would react with Cu to form Cu3Sn. The thermomigration resulted in the diffusion of Cu towards Cu pillar to react with solder which gave rise to the growth of Cu6Sn5 on the Cu pillar side.
INTRODUCTION
The three dimensional integrated circuit (3D IC) has been regarded as the new trend for electronic packaging. In 3D IC, thermal compression bonding is used as the bonding technique instead of conventional flip chip bonding. The thermal compression non-conductive paste (TCNCP) has been regarded as one of the solutions for fine pitch high I/O density products packaging. With the reduction of the solder joint dimension, the reduced solder volume may cause the rapid conversion of the solder into intermetallic compounds (IMC), which would lead to the reliability problems. Moreover, the thermomigration may be significant due to the reduced size of the solder joint. The interfacial reactions within the flip chip solder bump are generally known in previous researches. However, bonding behaviors and interfacial reactions in TCNCP process have seldom been reported in the literature. Thus, in this study, the bonding behaviors and interfacial reactions in the solder joints of TCNCP process will be investigated.
MATERIALS AND METHODS
In order to remove the OSP layer, the PCB substrate was pre-treated by Ar/O2 plasma, hydrochloric acid or formic acid. Double beams - focused ion beam (DB-FIB) was used to observe the OSP morphologies and removal effects after substrate pre-treatment. Four types of NCPs were used in the TC-bonding process. The physical properties of the NCPs were measured and compared with the analysis of microstructure and composition by SEM-EDX. TC-bonding conditions were listed in Table 1. Heating modes of linear and step type were used. After TC-bonding, pre-condition test and thermal cycle test (TCT) were conducted. The schematic diagram of TCNCP solder joint is shown in the Figure 1.
RESULTS AND DISCUSSION
Effect of the NCP properties on solder bonding behaviors
Four types of NCPs were used in this study. The NCP properties and the corresponding bonding behaviors were investigated. The difference of bonding behaviors was ascribed to the physical properties of the NCPs. The results showed that the bonding effect between solder and Cu trace could be improved by the usage of NCPs with higher gel temperature. The selection of NCPs with lower gel temperature may lead to the formation of cold joints due to the rapid curing of the NCPs.
Effect of the OSP residue and the removal methods
In this study, Ar/O2 plasma, hydrochloric acid or formic acid was used to remove the OSP layer on the Cu trace surface. The analysis results by DB-FIB showed that OSP layer could be effectively removed by the usage of Ar/O2 plasma. The OSP layer could also be removed by the immersion of acid solutions. Hydrochloric acid showed better ability to remove the OSP layer than formic acid. The ability of OSP removal was related to the acid dissociation constant. The SEM analysis of the TC-bonding specimen delineated that the solder joint with OSP removal showed good wetting behavior (Figure 2). However, bonding behaviors without OSP removal resulted in cold joint (Figure 3).
Effect of the heating rate on solder bonding behaviors
The solder bonding behaviors may be influenced by the heating rate of the TC-bonding process. In this study, linear and step heating modes were used to investigate the bonding behaviors. The joint morphology of linear heating mode was shown in Figure 2. In contrast to the linear heating mode, step heating mode showed poor bonding behavior (As shown in figure 4). The difference of the bonding behaviors might be attributed to rapid crosslinking and curing for the NCP with the step heating mode.
Interfacial reactions during TC-bonding
Interfacial reactions within the Cu pillar/solder/Cu trace structure during TC bonding were investigated. The result revealed that a uniform layer of Cu3Sn was formed at the solder/Cu trace interface. However, the Cu6Sn5 thickness on Cu pillar was much thicker than that on Cu trace. It was suggested that the growth of Cu6Sn5 on Cu trace might be suppressed due to the entrapment of NCP fillers at the Cu trace interface.
Effect of pre-con and TCT tests on interfacial reactions
After pre-con and TCT tests, the total IMC thickness increased. Kirkendall voids were formed due to the diffusion flux difference of Cu and Sn atoms. With the continuous reliability tests of the TCNCP specimen, Ag3Sn grain growth was also found in the solder joint.
Asymmetric growth of IMCs
After the pre-con test, the analysis results by SEM showed that the Cu3Sn growth rate on Cu pillar is higher than that on Cu trace. During the TCT, Cu6Sn5 on both Cu pillar and Cu trace would react with Cu to form Cu3Sn, however, the thickness of Cu6Sn5 on Cu pillar kept increasing with respect to the decreased thickness of the Cu6Sn5 on Cu trace. The asymmetric growth of IMCs might be due to the temperature difference between Cu pillar and Cu trace. The thermomigration resulted in the diffusion of Cu towards Cu pillar to react with solder which gave rise to the growth of IMCs on the Cu pillar side.
CONCLUSION
This study investigated the bonding behaviors and interfacial reactions during thermal compression non-conductive paste bonding (TCNCP). The OSP on the Cu trace surface needed to be removed to make solder and Cu react with each other and formed IMCs. Solder joints with good bonding behaviors could be produced by the usage of lower gel temperature NCP and linear heating process of TC-bonding. An uniform layer of Cu3Sn was formed during the TC-bonding. The growth of Cu6Sn5 might be suppressed due to the entrapment of NCP fillers at the Cu trace interface. After pre-con and TCT tests, Cu3Sn grew slowly. However, the growth of Cu6Sn5 lead to the asymmetric growth behavior on the Cu pillar/Cu trace side. In this study, it was speculated that the asymmetric growth of IMCs was due to the thermomigration whose temperature gradient was ascribed to the specimen structure and the heat conductivity difference.
|
|---|
