A mixed technology reliability assessment: using a mix of lead and Pb-free components, the authors found failure similar mechanisms for SAC and SnPb eutectic solder.

Availability of Pb-free components necessitated mixed technology, combining lead and Pb-free components. At Dade Behring, our intent was to provide early data that would reflect the impact on performance and reliability. This testing would help determine the need for critical changes and highlight potentially serious problems. The activity, if completed early enough, would provide time to resolve any difficult design issues. It would also permit adjustment, if necessary, of the Pb-free process, and help optimize the manufacturing process. It could also provide some specific metrics for process stabilization as volume increased. Ultimately, this activity would help reduce risk and increase confidence in our products' performance and reliability. This work compared existing lead-processed PCBs with PCBs fabricated with a Pb-free process, and used Pb-free components where available and lead components where Pb-free was not available. This latter decision itself proved problematic because it required hand soldering of some Pb-free components unable to withstand Pb-free process temperatures. As such, we probably have conducted a worst-case evaluation.

It must be stated that the entire evaluation was a joint activity, as evidenced by review of the authors' affiliations. In retrospect, this joint participation is probably a strength of the evaluation. The collective knowledge helped the planning, execution and evaluation of the investigation.

Finally, we need to make it clear that we do not advocate using lead components in designs that are claimed Pb-free. This evaluation was, as stated, "an evaluation" for investigatory purposes. The results were used to help determine how well Pb-free components and PCBs perform. This evaluation was used to guide us along a path to a Pb-free design, process and product.

SMT Process

Pb-free board assemblies were processed with a commercial SnAg3.0Cu0.5 alloy for both the solder paste and wave solder alloys. Solder paste printing and component placement followed similar conditions as those used for SnPb eutectic solder assemblies. The primary process differences between Pb-free and SnPb solder assemblies were in the reflow and wave solder profiles. Table 1 and Figure 1 show the reflow parameters and profile used for the Pb-free surface mount component attachment.

Post assembly, the boards were inspected using transmission x-ray analysis and electrical test. Details of both are presented herein.

Evaluation Plan

The evaluation plan consisted of several reliability tests. First, we subjected a sample of lead and Pb-free PCBs to a stress test, which consisted of about a weeklong procedure in an environmental chamber. PCBs were first characterized at nominal and extremes of power supply voltage, which established a baseline of performance against which to judge the rest of the performance during stress test. The PCBs' complete functionality was continuously tested and logged during the entire environmental stress test The environmental stress test ranged in temperature from -15[degrees] to 65[degrees]C and from 30% to 85% relative humidity. We did observe minor differences in performance. Figure 2 is the pattern used during stress test.

Second, a separate sample of lead and Pb-free PCBs was subjected to a 2000-hr. temperature and humidity test. The conditions were 85% RH and 60[degrees]C. We observed more lead than Pb-free failures during this test We were limited to 60[degrees]C due to the external sensor used to verify the test.

Finally, thermal cycling test was performed, again on a separate sample of lead and Pb-free PCBs. The temperature excursion was -40[degrees] to 125[degrees]C. Samples were tested at 0,200,500, 1000,1500 and 2000 cycles. This test revealed a difference between the lead and Pb-free PCBs. In brief, the lead PCBs exhibited failures before the Pb-free PCBs.

Electrical test. The PCB samples were tested to determine proper functioning of:

* Motor drive circuits.

* DC voltage levels.

* Ground.

* General purpose drivers.

* Serial/parallel interface.

* RS 232 interface.

* Analog/digital converters.

* Reference voltages.

* Internal board temperature.

* General purpose I/O functions.

* LED test

* Power supply test.

* Board serial number.

* Internal board ID.

* Firmware/EEPROM.

By and large, these tests were performed at each readout and continuously during stress testing. (Samples that failed these tests determined the % failure in the next section.)

Thermal cycling. Figure 3 shows thermal cycling results. Clearly, neither lead nor Pb-free performance was ideal or even close to theoretical best expectations. Some systematic layout issues induced overriding stress concentrations, which dominated the onset of failure. Despite the performance, however, the Pb-free process was more robust to thermal cycling.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

[FIGURE 3 OMITTED]

Failure Analysis

Following thermal cycle exposure, from -40 to 125[degrees]C, three failed assemblies were analyzed to determine the root cause. Electrical and functional testing of the boards determined failures. Test results indicated particular electrical nets that had failed, along with the most likely components within the net that had failed. Board 12 was from the Pb-free assembly group, and boards 2 and 4 were from the SnPb assembly group. Electrical tests indicated that these three assemblies might have cracks in the solder joints of the oscillators and R60 resistors post-thermal cycle exposure. Figure 4 shows optical micrographs of typical oscillator and R60 resistors post-thermal cycling.

Preliminarily failure analysis, consisting of transition x-ray analysis, was performed. Figure 5 shows some voiding in the oscillators' solder joints from boards 4 and 12. Figure 6 shows typical voiding in the R60 solder joints of boards 2 and 4. Generally, the Pb-free and SnPb boards exhibited similar degrees of voiding in the solder joints inherent to the solder paste reflow process used.

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

Next, the assemblies were analyzed using optical microscopy. Optical microscopy indicated cracking in a number of solder joints. Figure 7 shows oscillator SnPb solder joints on board 2 post-reliability testing. Figures 8 and 9 show R60 and R109 resistors for SnPb joints on board 2. In both cases, surface cracks in the bulk solder joint can be seen.

Cross-sectional analysis was used to further evaluate the solder joint cracking. Scanning electron microscope (SEM) imaging and energy dispersive x-ray (EDX) elemental analysis was performed on cross-sections of several oscillators and resistors from boards 2, 4 and 12. Solder cracking was evident in several solder joints of the oscillator and resistors for all three board assemblies post-thermal cycling. All cracking occurred in the bulk solder material. The cracking appears to form a "slip band" effect, or standard solder fatigue cracking, as a result of the -40[degrees] to 125[degrees]C thermal exposure. There was no evidence of cracking caused by processing defects.

[FIGURE 6 OMITTED]

[FIGURE 7 OMITTED]

[FIGURE 8 OMITTED]

[FIGURE 9 OMITTED]

Figure 10 shows typical SEM micrographs and EDX analysis of oscillator Pb-free solder joints post-thermal cycling, exhibiting through cracks in the solder joints. Cracks appear to have initiated along the outer part edge and solder joint interface, propagating through the bulk solder toward the part's center.

Figure 11 shows typical SEM micrographs and EDX analysis of R60 Pb-free solder joints post-thermal cycling. A higher concentration of crack growth appears on one side of R60. Moreover, the Pb-free solder does not appear to provide well-defined heel fillets beneath the components.

Figure 12 shows oscillator SnPb solder joints post-thermal cycling, exhibiting through cracks in the solder joints. Similar to the Pb-free solder case, cracks appear to have initiated along the outer part edge and solder joint interface and propagated through the bulk solder toward the part's center.

Figure 13 shows the R60 SnPb solder joints post-thermal cycling. Similar to the Pb-free case, a higher concentration of crack growth appears on one side of R60. Moreover, the SnPb solder joints appear to be slightly starved of solder volume and exhibit no heel fillets, further indicating a lack of solder volume and potentially excessive placement force.

The micrographs also show the solder grain structure. As expected, the solder grain structure has coarsened because of thermal cycle exposure. This coarsening also accelerates solder fatigue cracking.

Conclusions

Failure mechanisms of the Pb-free solder and SnPb eutectic solder assemblies were similar. Solder cracking was evident on comparable solder joints of oscillators and resistors post-thermal cycling. All cracking occurred in the bulk solder material. The cracking appears to form a "slip band" effect, or standard solder fatigue cracking, due to the -40[degrees] to 125[degrees]C thermal exposure. There was no evidence of cracking caused by processing defects.

The lead PCBs failed sooner than Pb-free PCBs for a given form factor, such as dimension on PCB and size of component. The failure mode and mechanism appear to be similar, if not identical. Therefore, we observed no new failure mechanism. Component availability issues were discovered early in the design cycle. These early indicators permit redesign where necessary. Finally, it appears changing from an SnPb to a Pb-free process can improve PCB thermal cycle reliability.

[FIGURE 10 OMITTED]

[FIGURE 11 OMITTED]

[FIGURE 12 OMITTED]

[FIGURE 13 OMITTED]

References

1. B. Trumble, "Get the Lead Out!" IEEE Spectrum, May 1998.

2. J. Kloeser, et al., "Fine Pitch Stencil Printing of Sn/Pb and Lead Free Solders for Flip Chip Technology," 47th Electronic Components and Technology Conference Proceedings, May 1997.

3. J. Linton, "The European Union's New Environmental Directives," CIRCUITS ASSEMBLY, September 2000.

4. C. de Wit, "Brominated Flame Retardants," Swedish EPA Report 5065, 2000.

5. R.G. Robertson and J. Smetana, "Fundamental Concerns in Pb-free Implementation," CIRCUITS ASSEMBLY, September 2000.

6. J. Hwang, "Lead-free Solder: the Sn/Ag/Cu System," SMT, July 2000.

7. P. Houston, B. Lewis, D.F. Baldwin, and P. Kazmierowicz, "Taking the Pain Out of Pb-free Reflow," IPC Apex Conference Proceedings, March 2003.

8. G. Baynham, D.F. Baldwin, K. Boustedt, A. Johansson, C. Wennerholm, D. Patterson and P. Elenius, "Flip Chip with Lead-Free Solders on Halogen Free Microvia Substrates," Electronic Components and Technology Conference Proceedings, pp. 1101-1105, May 2000.

9. G. Baynham, D.F. Baldwin, K. Boustedt, C. Wennerholm, D. Patterson and P. Elenius, "Flip Chip with Lead-free Solders on Halogen-Free Microvia Substrates," Electronic Components and Technology Conference Proceedings, pp. 1135-1139, May 2001.

10. John Sohn, "Are Lead-Free Solder Joints Reliable?" CIRCUITS ASSEMBLY, June 2002.

Ed.: This article was first presented at the SMTA Medical Electronics Symposium in May 2006 and is reprinted with permission.

Dr. Lucian Kasprzak is staff engineer at Dade Behring Inc. (dadebehring.com); [e-mail]. Milind Sawant is test engineer and Gerry Adams is staff engineer with Dade Behring; Brian Lewis is advanced process engineer; Paul N. Houston is advanced process development engineer, and Dr. Daniel Baldwin is president and CEO of Engent Inc. (engentaat.com); Mike Nahorniak is test engineer at Libra Industries Inc. (libraind.com).

Dr. Lucian Kasprzak, Milind Sawant, Gerry Adams, Brian Lewis, Paul N. Houston, Dr. Daniel Baldwin and Mike Nahorniak

Table 1. Pb-Free Solder Reflow Profile Parameters

Ramp Rate
([degrees]C/sec)  Time >217[degrees]C (sec.)  Peak Temp ([degrees]C)

0.85              68                          240