Reliability of Mixed-alloy Solder Joints
December 31, 1969 |Estimated reading time: 9 minutes
Completely mixing tin/lead solder with the SAC ball is critical to form uniform and homogenous microstructures. This article details a study on the reliability of mixed-alloy solder joints at different process conditions, and compares results with lead-free control samples. By Jennifer Nguyen; David Geiger; Daniel Rooney, Ph.D.; and Dongkai Shangguan, Ph.D., Flextronics International
he complete mixing of tin/lead solder with a lead-free SnAgCu (SAC) ball is critical to the formation of uniform and homogeneous microstructures. A previous study1 showed that the complete mixing of tin/lead solder with SAC depends on the volume of the tin/lead solder relative to that of the SAC ball and the soldering temperature. It can happen at reflow temperatures below 217°C if the correct amount of tin/lead solder paste is used. Is a homogeneous solder joint more reliable than heterogeneous joints? Is the mixed-alloy solder joint between tin/lead paste and a SAC ball less reliable than a homogenous alloy solder joint? To answer these questions, the reliability of mixed-alloy solder joints at different process conditions was studied, and results were compared with the lead-free control samples. Thermal-cycle, drop, shear, and bend tests were performed for the mixed-alloy solder joints and lead-free solder joints.
Assembly Process Conditions
The mixed-alloy samples of different lead-free area-array packages were prepared with different percentages of tin in the solder joints, and reflowed under various reflow profile conditions (Table 1). The solder ball composition used was Sn3.0Ag0.5Cu (SAC 305). In the mixed-alloy samples, lead-free (SAC) components were assembled with tin/lead solder paste. Control samples were prepared with lead-free (SAC) area-array packages assembled with lead-free solder paste. The percentage of lead and tin in the mixed-alloy solder joints (Table 1) is calculated using the following equations:
It is noted that the actual solder volume is different from the tin/lead solder paste volume applied. The stencil thickness and aperture control the amount of tin/lead paste.
FIGURE 1. X-ray radiograph of BGA package on 95% tin and reflowed at 205°C (best-case scenario of void).
X-ray InspectionX-ray inspection results showed that samples of high tin content and low reflow temperature had the least amount of voids and smaller voids in the solder joints (Figure 1); low tin content and high reflow temperatures resulted in more and larger voids (Figure 2). A previous study also showed that voids decreased when tin percentage increased in the BGA solder joints, and that voids increased when higher reflow peak temperature were used. This means that conditions for a greater degree of mixing also correlate with more voiding. However, purely lead-free solder joints of the BGA at higher reflow temperatures displayed somewhat less voiding than mixed-alloy parts with lower reflow temperatures.
Figure 2.X-ray radiograph of BGA package on 87% tin and reflowed at 230ºC (worst-case scenario of voids).
Solder Joint MicrostructureCross sectioning was performed to analyze the mixed-alloy solder joint characteristics. For BGAs, good mixing of lead into the SAC ball was seen in most cases, and lead was distributed fairly uniformly from the top to the bottom of the solder joints, with the only exception of 95% tin reflowed at 205°C (Figure 3). In this case, the lead-rich phase looks like veins in the cross-section image, and little lead was seen near the component side of the solder joint. The microstructure of its solder joint looked much rougher than typical tin/lead and lead-free solder joints. There were noticeable differences in the microstructure of mixed-alloy joints assembled at different process conditions. Samples with higher tin content and reflow temperature (Figure 4) appeared to exhibit smaller and more finely dispersed lead-rich phases than samples with a lower tin content and reflow temperature (Figure 5). The microstructure of mixed-alloy solder joints assembled at a low-tin content and high reflow temperature is shown in Figure 6, and the solder joint microstructure of a sample assembled at a high-tin content and low reflow temperature is shown in Figure 7. In general, the solder joints of low reflow temperatures result in a larger grain size and lead-rich phase (Figures 4 and 6). Solder joints of high reflow temperatures look more homogeneous and uniform (Figures 5 and 7). The cross-section of chip-scale package (CSP) solder joints revealed that the lead-rich phase was finely dispersed. The microstructure in wafer-level CSP solder joints looked homogeneous and more like eutectic tin/lead alloys than a mixed alloy. This is expected because CSP mixed-alloy joints had only 72% tin and 28% lead.
Thermal Cycle Test
The thermal-cycle test was performed in an air-to-air thermal-cycle chamber from -40 to 85°C with a 15-min. dwell time at each temperature exterme. Ramp rates are about 24°-29°C per minute. Thermal-cycle test results showed that the CSP failed much earlier than the BGA package. The mixed-alloy solder joint CSP components started to fail at about 600 cycles, while no BGA failures were found in up to 2,500 thermal-shock cycles.
Figure 3. Cross section image of BGA package with 95% tin in the mixed alloy, reflowed at 205ºC.
Weibull analysis of the CSP failure data is shown in Figure 8. The results showed that lead-free solder joints performed better than the mixed-alloy solder joints in the thermal-cycling test, which aligns well with other studies.2,3 Mixed-alloy samples reflowed at a higher peak temperature performed better than the mixed-alloy samples reflowed at a lower peak temperature. This is because lead in the solder joint of the mixed-alloy samples with higher reflow temperatures was more finely dispersed than in samples with a lower reflow temperature.
The cross sections of the failed solder joints in the mixed-alloy CSP components exhibited cracks in a phase-coarsened band within the bulk solder at the component side. These failed samples exhibited a microstructure that is typical of thermo-mechanical fatigue failures in tin/lead solder alloys. Lead-free CSP samples also exhibited cracked solder joints from thermo-mechanical fatigue.
Drop Test
A board-level free-fall drop test was performed. Samples were held horizontally 5' from the floor and dropped. The component sat on the bottom of the board, and a fixture was used to hold the edges of the board during the drop test. This face-down drop is the most severe for mobile phones, and is close to the free-fall product-drop condition.4 The acceleration of the drop test was measured; the average peak amplitude is 2398G, and the time span is about 0.37 msec.
Figure 4.SEM micrograph of cross-sectioned sample with high tin content (95%) reflowed at 230ºC.
Drop-test results showed that the CSPs survived better under mechanical stress than the BGA packages. CSP components survived as many as 200 drops without any cracked solder joints, whereas the BGA samples assembled with lead-free paste survived an average of 34 cycles, and mixed-alloy solder joint BGAs survived from an average 50-80 drop cycles, depending on the process conditions. This is expected because CSPs are much smaller than BGAs. For the drop test, mixed-alloy solder joints performed better than lead-free solder joints. However, there is no statistical difference in the reliability of mixed-alloy solder joints assembled at different process conditions (different tin content and peak temperature).
Figure 5. SEM micrograph of cross-sectioned sample with low tin content reflowed at low temperature.
Cross sectioning revealed cracks at the component side and the PCB side; however, the extent of cracks was much more severe on the component side. The cracks found on the PCB side did not extend through the entire solder joint. The solder failures appear to be interfacial at the nickel metallization on the component solder pads. One reason that the solder joints fail at the component side is that the component solder pads are smaller that the PCB solder pads for this component. Another reason is that the component solder pads are plated with nickel rather than the copper material on the PCB solder pads, and solder joints to nickel are weaker mechanically than these to copper.
Figure 6. SEM micrograph of cross-sectioned sample with low tin content reflowed at high temperature.
Shear-test results for the CSP showed that lead-free solder joints were stronger than mixed-alloy solder joints. There was no significant difference in the shear strength for samples reflowed at low and high peak temperatures. Four-point bending tests were performed using the JEDEC method 22B113. The bending test consists of two stationary and two moving anvils. The span on the moving beams is 75 mm, the span on the stationary beams is 110 mm. Total displacement is 3 mm; the repetition rate is 2.25 Hz (135 cycles/min.).
Figure 7. SEM micrograph of cross-sectioned sample with high tin content reflowed at low temperature.
Statistically, there are no differences in the bending test reliability of mixed-alloy solder joints assembled at different process conditions. However, it is noticed that the slope of the Weibull curve from the high tin content (95% tin) and low temperature (205°C) group is steeper than the other groups. This group has heterogeneous microstructures with incomplete mixing of lead in the solder joints. This might create non-uniform stress distribution within the solder joint and makes it more vulnerable to mechanical shock. The cross section shows that the cracks were located in the bulk solder near the component side. The cracks initiated at the location where the soldermask on the component contacts the solder, which suggests that the solder mask on the component is acting as a stress riser.
Figure 8. Thermal-cycle reliability - CSP8 at -40º to 85ºC cycling, up to 2,500 cycles.
ConclusionThis study shows that sensitivity of mixed-alloy solder joints relability to process conditions depends on the type of environmental loading. Assembly process conditions affect the thermo-mechanical reliability of the mixed-alloy solder joints significantly. Thermal-cycle testing (-40° to 85°C) has shown that lead-free solder joints performed better than the mixed-alloy solder joints; mixed-alloy samples reflowed at higher peak temperature performed better than the mixed-alloy samples reflowed at a lower peak temperature.5 However, under a mechanical-shock environment (such as drop, shear, and bending tests), there is no significant difference in the reliability of mixed-alloy samples assembled at different process conditions. This difference in the sensitivity to the process condition is due to different failure modes. Under thermo-mechanical loading conditions, solder joint failure is within the solder and is more sensitive to the microstructural characteristics of the solder. In turn, process condition affects this significantly. For mechanical shock, solder joint failure typically is interfacial in nature and not as sensitive to the microstructural characteristics of the solder itself; therefore, it is less sensitive to the process condition.
ACKNOWLEDGMENTS
The authors thank DongJi Xie for support in the reliability test plan and Ninh Vu for support during the test-board assembly.
REFERENCES
- J. Nguyen, D. Geiger, D. Rooney, and D. Shangguan, “Backward Compatibility Study of Lead-Free Area Array Packages with Tin-Lead Soldering Processes,” APEX 2006.
- C. Handwerker, J. Bath, E. Benedetto, etc., “NEMI Comparison between PbSn and SnAgCu Reliability and Microstructures,” SMTAI 2003.
- JPan, J. Bath, “Lead-free Soldering Backward Compatibility,” IPC/JEDEC 12th Int’l Conference on Lead-free Electronics Components and Assemblies, Santa Clara, Calif., 2006.
- D. Xie, D. Geiger, D. Shangguan, D. Rooney, and L. Gullo, “Characterization of Fine Pitch CSP Solder Joints Under Board-Level Free Fall Drop (BFFD),” IPACK 2005.
- T. Gregorich, P. Holmes, “Low-temperature, High-reliability Assembly of Lead-free CSPs”, IPC Lead-free Conference, Germany, 2003.
For a complete list of references and figures, please contact the authors.
Jennifer Nguyen, Flextronics International, may be contacted via e-mail: jennifer.nguyen@flextronics.com. David Geiger, Flextronics International, may be contacted via e-mail: david.geiger@flextronics.com. Daniel Rooney, Ph.D., Flextronics International, may be contacted via e-mail: dan.rooney@flextronics.com. Dongkai Shangguan, Ph.D., vice president - Assembly Technology and Platform Realization, Flextronics International, may be contacted at (408) 428-1336; e-mail: dongkaishangguan@flextronics.com.