Maximize the Menu
Transition to Pb-free Alloys in Capacitor Filter Assemblies *

* This paper article was originally published in the Proceedings of the SMTA 2006 PanPacific Microelectronics Conference

Introduction

This paper describes the Pb-free process development for capacitor filter assemblies of various sizes that use solder preforms in the shape of a washer. Components considered for this research are circular filter capacitors that have pins inserted into a ferrite core. SnAgCu (SAC) and SnAg alloys are considered for the replacement of the SnPb preforms currently used. The objective of this work is to study a variety of factors (including reflow atmosphere, alloy composition, platings on the capacitor array (silver-based and gold-based), and flux) on the quality and reliability of the solder joints. The reliability of Pb-free solders is evaluated by first studying the intermetallics of the control samples, those made with Pb-bearing solders, which is the baseline for the intermetallics evaluation of the Pb-free solders. Thus far, we have begun to establish a Pb-free assembly process and have made initial investigations into intermetallic determinations of species and their amounts. Process development efforts for reliability determinations, including thermal cycling are conducted with the ultimate goals being the following: reduce and control the amount of intermetallics; increase the likelihood of multiple grains of solder formed in each joint; and eliminate the voiding we have discovered due to the flux.

With the arrival of legislative restriction on the use of Pb-bearing solders, the electronics industries are rapidly progressing towards a full manufacturing transition of replacing Pb-bearing solders with those that are Pb-free. The main concern in doing so is that Pb-free solders may not behave in the same way as do Pb-bearing solders.

One of the primary challenges is to ensure that Pb-free solder joints have reliability on par with Pb-bearing (primarily, SnPb) solder joints. The microstructure of the solder joint has a great influence on the mechanical properties of the solder joint, which in turn affects the reliability of the solder joint and the capacitor. As the distribution of alloying elements of the Pb-free solders depends on the cooling rate during reflow, the characterization and optimization of a reflow process is a significant factor to be considered.

The Pb-free alloys considered in this work are near eutectic SAC (305), which is Sn 96.5 Ag 3.0 Cu 0.5, and eutectic Sn 96.5 Ag 3.5. Both Pb-free solders have high temperature strength and high interconnect applications, in contrast to the eutectic SnPb solders. Control samples with eutectic SnPb were used for comparison purposes. These Pb-free alloys are Sn-rich alloys allowing for Cu6Sn5 and Ag3Sn intermetallic compound precipitates after reflow. The size of these precipitates depends on solder composition and reflow processing conditions. Large Ag3Sn precipitates are often found to be shaped as plates, while Cu6Sn5 precipitates are rod-like.

The objective of this work is to study a variety of factors, such as reflow atmosphere, alloy composition and other factors, on the quality and reliability of the solder joints. The reliability of Pb-free solders is evaluated by first studying the intermetallics of the control samples, those made with Pb-bearing solders, which is the baseline for the intermetallics evaluation of the Pb-free solders. An interesting sub-objective of the work that ties into a production-planning goal, revolves around the need to develop identical reflow profiles for three different sizes of capacitors (small, medium, and large). To help eliminate/reduce setup times, it was of interest to establish reflow parameters wherein the zone temperature settings were fixed (regardless of size of the capacitor assembly) but that the same reflow profiles could be obtained by simply changing the belt speed to compensate for the variation in the mass of the assemblies and their fixtures (the larger the capacitors/fixtures, the slower the belt speed).



Experimental Procedure

The specimens that are used in this study are ceramic circular filter capacitor assemblies that have Cu/Ni/Au plated pins inserted through a hole in the capacitor through a ferrite core, then inserted through a hole in the second capacitor. The assemblies are of three different shell sizes (diameters) denoted as small, medium, and large. A washer shaped, no clean flux coated, solder preform is used to fill the volume between the pin and the hole in the ceramic.

A 6-zone convection reflow oven with a nitrogen atmosphere, at Binghamton University (BU), was used to generate a solder profile for the large and small assemblies (Figure 1, a-b), as the large and small assemblies account for 75% of the parts that are processed on a monthly basis. The profile investigation was performed with a thermocoupled solder fixture with an important criterion, as earlier mentioned, of having a common profile between the different sizes with minimal changes in setup. The thermocouples were attached to 5 different locations on the fixture and were passed through the oven along with a thermal profiler, to monitor the temperatures that various parts of the filters experience.
Thermocouple Location on Filter Array for Profiling:

  1. Top, Center Hole
  2. Top, Edge Hole
  3. Bottom, Edge Hole
  4. Top, Contact
  5. Bottom, Contact

Figure 1a: Large sample and test fixture
Figure 1a: Large sample and test fixture (Click to Zoom)

Figure 1b: Small sample and test fixture
Figure 1b: Small sample and test fixture (Click to Zoom)

The desired Pb-free reflow profile has a preheat dwell at 170°C (338°F) for 30 to 60 seconds and a reflow plateau at 230°C to 240°C (446°F to 464°F). The control samples, using Pb-bearing alloys, had a reflow plateau of 195°C (338°F). The control samples, having both types of coating (silver frit and gold plating on the ceramic capacitor) were reflowed, at Amphenol Aerospace Operations (AAO), using a 9-Zone convection oven with a nitrogen atmosphere.

The test specimens were cross-sectioned and mounted in epoxy moulds. Once completely cured, the cross sections were ground using a variable speed grinder and using 120, 320, 600, 800, and 1200 grit carbide papers. The fine polishing was done with polishing cloths loaded with diamond pastes (6 μm to 1 μm) and polishing oil. The final polishing was done with 0.05 μm alumina gel, followed by 0.02 μm silica gel for a very short duration just to expose the intermetallics. The aforementioned grinding and polishing operations were performed with exacting care while varying the speed and pressure, maintaining the flatness of the sample and removing any damage (particularly, scratches) acquired from any of the previous polishing steps. After the final polishing, images were taken using the bright field and cross polarized imaging techniques.

The baseline for the intermetallics and the grain structure was performed using the control samples. The reliability testing consisted of a thermal shock testing and a burn-in test. The thermal shock conditions were the same for all sizes with a temperature range of ­55°C to 125°C with a rapid transfer from one temperature to another with a 30 minute dwell period. The burn-in test was performed for 24 hours at 125°C with a dielectric withstand voltage (DWV) voltage applied. All samples in the control group had gold plating.



Optical Microscopy

The combination of bright field and cross polarized optical microscopy was used to delineate different Sn grains and the location of Sn grain boundaries in the samples. Because β-Sn is birefringent, cross polarized light microscopy reveals a different contrast between differently oriented Sn grains. For this technique to work well, it was important to have the final polishing stages free from any polishing derived defects. Such defects generally include formation of a number of small Sn grains, i.e., recrystallization of the β-Sn phase.

Results obtained after studying the above crosssections of the control samples, t subjected to the reliability tests and the newly soldered Pb-free solders are discussed below. The Pb-free samples evaluated here have not yet been through the reliability testing.



Results and Analysis

During the course of the study, the reliability of the Pb-free samples is evaluated by first studying the intermetallics of the control samples, those made with Pb-bearing solders, which is the baseline for the intermetallics evaluation of the Pb-free solders. Thus far, we have developed the reflow profiles of the Pb-free assemblies and have made initial investigations of the Pb-bearing assemblies (control samples) into intermetallic determinations of species and their amounts. With all these findings, listed here in brief, we have moved towards the reliability determinations, namely, thermal shock test and burn-in tests for the SnPb samples, again to set a baseline for Pb-free samples.

Figure 2 (a, b) shows the reflow profiles for the small and the large fixtures that were obtained by varying the belt speed and keeping the zone parameters constant. The peak temperature of the profiles was approximately 234°C (453.2°F) with a time above liquidus (TAL) of approximately 120 seconds. Using these profiles, two samples of different sizes, small and large, one each having SAC and SnAg solders were reflowed.

Figure 2a: Initial reflow profile for the small sample test fixture
Figure 2a: Initial reflow profile for the small sample test fixture (Click to Zoom)

Figure 2b: Initial reflow profile for the large sample test fixture
Figure 2b: Initial reflow profile for the large sample test fixture (Click to Zoom)

The initial findings (see Figure 3, for example) show that the tin-lead microstructure consists of Sn-rich and Pb-rich lamella grains. When observed under the polarized light microscopes, the Pb-rich grains appear as bright regions while the Sn-rich grains appear significantly darker in color. The grain growth and the grain coarsening indicate an excessive time at temperature above the phase temperature of the specific intermetallic alloy being formed. SEM analysis was conducted to identify the content of different alloys in the intermetallics. Due to the presence of different metallurgies (platings) on the planar capacitor (silver-based and gold-based) we assumed that the dissolved Ag and Au form intermetallics with Sn and hence we estimated how much Sn is consumed by intermetallics during reflow, i.e., the intermetallic loading of the solder. The presence of gold (Au) in the plating (for the Au-plating case) and on the pins (in both cases) has led to the consumption of a large amount of Sn in the form of AuSn4 intermetallics [Shah et al., 2005]. In case of small samples the gold plating with both Ag and Au going into solution, approximately 70% of the free Sn can be consumed. In case of large samples the budget of Sn used to form intermetallics in the sample having gold plating is approximately 65 percent.

!!!FIG
Figure 3: SEM image of large sample with SnPb solder having gold plating. The bright phase in the solder is the Pb phase and the gray background is the SN phase

The optical micrographs and the semi-quantitative EDS analysis provided the composition of each element in the weight percentage present in that phase. The composition formed near the solder-pin interface and the ceramic-solder interface depends on the type of frit or plating on the capacitor. Figure 3 is one example of an SEM image of a large sample having SnPb solder. The main phase found in this solder joint is dendrites of the eutectic Sn, Ag3Sn, and AuSn4. Crystals of Ag3Sn are generally found at the pin-solder interface and also at the ceramic-solder interface, whereas AuSn4 manifests as rods having indefinite cross sectional shape and size.

Reliability of the solder joint formed using these profiles was an important issue and the void formation in the solder joints is one of the many critical factors governing the reliability of the solder joint. A usual suspicion for voids is outgassing of the flux that gets entrapped in the solder joints. However, the flux used with the solder preforms is non-foaming, which reduces its likelihood as the major contributor to void formation. Cullen (2005) has suggested that another possibility/ contributor for the creation of microvoids may be due to there being an entrapment of organics during the board surface finishing process. We may be experiencing a similar situation with the silver-frit on ceramic process. Visual comparisons of the bright field and the cross polarized images of SnPb samples and Pb-free samples give an indication that the number of voids is considerably reduced in the Pb-free samples.

Having these findings as the baseline, we began our study of reliability data and intermetallic analysis of Pb-free solders. Figure 4 (a-b) shows bright field and cross polarized micrographs of the large sample having SnPb solders with gold plating that was tested for reliability analysis.

Figure 4a: Bright field image of large sample having SnPb solder with gold plating. The bright phase in the solder is the Pb phase and the gray background is the Sn phase.
Figure 4a: Bright field image of large sample having SnPb solder with gold plating. The bright phase in the solder is the Pb phase and the gray background is the Sn phase. (Click to Zoom)

Figure 4b: Cross polarized image of a large sample having SnPb solder with gold plating. The bright phase in the solder is the Pb phase and the gray background is the Sn phase.
Figure 4b: Cross polarized image of a large sample having SnPb solder with gold plating. The bright phase in the solder is the Pb phase and the gray background is the Sn phase. (Click to Zoom)

For the reliability analysis, the large samples were subjected to thermal shock for 2 cycles at the temperature range of ­55°C to 125°C. These samples depicted failure at a rate greater than 50% at the end of the thermal cycle; once those passed the thermal cycle, they depicted failure at the same rate subsequent to the 24 hour burn-in tests. The failures that occur were seen as large spider web cracking across the face of the array. To compensate with the failures, the large sample was then reflowed using the medium profile and these samples did not fail during the thermal shock tests or the burn-in test. Figure 5 (a - b) shows the bright and cross polarized images of samples reflowed using the medium profile.

Figure 5b: Cross polarized image of a large sample having gold plating reflowed using a medium profile. This sample did not fail the thermal shock.
Figure 5b: Cross polarized image of a large sample having gold plating reflowed using a medium profile. This sample did not fail the thermal shock. (Click to Zoom)

Here we hypothesize that the types of intermetallics formed in the SnPb alloys are similar to the initial findings (shown in Figure 3). Figure 5(a) shows a lesser number of AuSn4 intermetallics when compared with Figure 4(a), thus the solder joint is likely to be more brittle and susceptible to failures.

From the above micrographs, it is visible that the number of intermetallics in case of samples reflowed using the medium profile is significantly less than in the case of the samples reflowed using the large profile.

Having all this information as the baseline towards Pb-free solders, the Pb-free samples of both sizes, small and large, and solders, SAC and SnAg, were reflowed. Figures 6(a-d) show the micrographs of Pb-free samples reflowed using the large profile and Figures 7(a-d) show the micrographs of Pb-free samples reflowed using the small profile.

Figure 6a: Bright field image of a SAC solder reflowed using the large profile
Figure 6a: Bright field image of a SAC solder reflowed using the large profile (Click to Zoom)

Figure 6b: Cross polarized image of a SAC solder reflowed using the large profile
Figure 6b: Cross polarized image of a SAC solder reflowed using the large profile (Click to Zoom)

Figure 6c: Bright field image of a SnAg solder reflowed using the large profile
Figure 6c: Bright field image of a SnAg solder reflowed using the large profile (Click to Zoom)

Figure 6d: Cross polarized image of a SnAg solder reflowed using the large profile
Figure 6d: Cross polarized image of a SnAg solder reflowed using the large profile (Click to Zoom)

In the pins, there is a low Cu concentration covered with a Ni/Au finish. Chen et al. (2002) have shown that with the reflow of SnCu solders over a Ni-bearing finish, a slight variation in Cu concentration would produce a completely different reaction product. The same phenomenon (of the long, seemingly detached-from-the-pin intermetallic band) has occurred in the case of SAC alloys reflowed either using the large profile (Figure 6(a)) or using the small profile (Figure 7(a)). This band will likely lead to field failures and we need to eliminate this phenomenon. While we have yet to determine the species of this intermetallic, our suspicion is that it is Cu/Ni6Sn5. A similar phenomenon was found by Chen et al. (2002). To control this behavior, they suggest that the amount of Cu content should either be increased or completely eliminated. However in our case, the Cu concentration of the solder cannot be changed. The copper in the pins should not contribute to intermetallics in the joint due to the nickel plating on them. Further, as we desire to use the more readily available SAC alloys (0.5% Cu) we will not be able to increase or remove the Cu amount. As a result, an avenue that we will take will be to modify the reflow profiles so that we reduce the TAL to inhibit this phenomenon. The findings presented here are ongoing investigations being conducted to study the effect of thermal stresses on these different sizes of samples on the ceramic capacitor filter assemblies to understand the change in the microstructures in moving from SnPb to Pb-free solders.

Figure 7a: Bright field image of a SAC solder reflowed using the small profile
Figure 7a: Bright field image of a SAC solder reflowed using the small profile (Click to Zoom)

Figure 7b: Cross polarized image of a SAC solder reflowed using the small profile
Figure 7b: Cross polarized image of a SAC solder reflowed using the small profile (Click to Zoom)

Figure 7c: Bright field image of a SnAg solder reflowed using the small profile
Figure 7c: Bright field image of a SnAg solder reflowed using the small profile (Click to Zoom)

Figure 7d: Cross polarized image of a SnAg solder reflowed using the small profile
Figure 7d: Cross polarized image of a SnAg solder reflowed using the small profile (Click to Zoom)



Conclusions and Ongoing Work

As the lead-free process development for these capacitor filter assemblies has progressed, the results of reliability testing (thermal shock test and burn-in test) for the baseline (Pb-bearing assemblies) are presented. Hypothesizing that the intermetallics formed in the SnPb solders are similar to the initial findings (Figure 3) there is a presence of more AuSn4 intermetallics in samples reflowed using the large profile as compared with samples using the medium profile. By reducing the time at reflow temperature, medium vs. large profile, solder reflow appears to reduce the intermetallics, improves the number of solder grains, and reduces the size and magnitude of gas bubbles formed from the trapped organics. The excessive number of AuSn4 intermetallics would likely make the solder brittle and, hence, lead to failure of these samples during the reliability analysis. As a result, our ongoing efforts are focusing on reducing the TAL for all capacitor sizes. The intermetallic study of the Pb-free samples displayed some promising features ­ mainly the effect of low Cu concentration on the Ni finish pins.

As a side-note, a production-based objective is also met wherein, with the different-sized assemblies, the same reflow profile can be achieved by only changing belt speed (and not temperature settings). Thus far, we have begun to establish reliability data for lead-free assembly and have made initial investigations into intermetallic determinations of species. Thermal cycling for our lead-free samples is underway.



Acknowledgements

The authors acknowledge the financial support for this research byAmphenolAerospace Operations. Thanks also go to the Integrated Electronics Engineering Center (IEEC) at Binghamton University for use of their equipment. The authors are grateful to Dr. Eric Cotts, Materials Science and Physics Department, Binghamton University for use of Optical Microscopy equipment.



References

  1. V. A. Shah, A. S. Kane, A. M. Shakir, D. L. Santos, M. DiPietro, L. P. Lehman, L. Krantz: Pb-Free Process Development and Microstrutural Analysis of Capacitor Filter Assemblies Using Solder Preforms, Proceedings of SMTA International, Chicago, IL, September 2005.

  2. A.U. Telang, T. R. Beiler, J.P. Lucas, K.N. Subramanian, L.P. Lehman, Y. Xing and E.J. Cotts: Grain Boundary Character and grain Growth in Bulk Tin and Bulk Lead-free Solder Alloys, Journal of Electronic Materials, December 2004, Volume 33, No.12, pp 1412-1423.

  3. J. Rathod, D. Santos, J.Belmonte, and A.J. McLenaghan: Comparisons of Microstructures in SAC305 and SAC405 Solder Joints and the Influence of the Surface Finish, Advanced Technology Symposium, Boston, MA, June 2002.

  4. S. Park, R. Dhakal, L. Lehman, E. Cotts: Grain Formation and Intergrain Stresses in Sn-Ag-Cu Solder Balls, Proceedings of ASME InterPACK, San Francisco, CA, July 2005.

  5. W. T. Chen, R.S. Tsai, Y. L. Lin, and C. R. Kao: Effect of Copper Concentration on the Solid-State Aging Reactions Between Tin-Copper Lead-Free Solder and Nickel, Journal of SMT, October 2002, Volume 15, Issue 4, pp 40-44.

  6. B.L. Partee: Microstructure Behavior of Lead-Free Solder Joints, Advanced Packaging Technology Conference Proceedings; June 2001.

  7. L.P. Lehman, S.N. Athavale, T.Z. Fullem, A.C. Giamis, R.K. Kinyanjui, M. Lowenstein, K. Mather, R. Patel, D. Rae, J. Wang, Y. Xing, L. Zavalij, P. Borgesen and E. Cotts: Growth of Sn and Intermetallic Compounds in Sn-Ag-Cu Solder, Journal of Electronic Materials, December 2004, Volume 33, No.12, pp 1429-1439.

  8. D.W. Henderson et.al: The microstructure of Sn in Near Eutectic Sn-Ag-Cu Alloy Solder Joints and its Role in Thermomechanical Fatigue, Journal of Electronic Materials, 2004, Volume 19, No.6, pp 1608-1612.

  9. D.P. Cullen: Avoiding Microvoids, Assembly (magazine), October 2005, 54-61.