Cooling Rates in Lead-free and Tin/lead Reflow

Estimated reading time: 6 minutes

The impact of a fully controlled and robust reflow process on solder-joint quality has been the topic of many studies, and is well understood for processes using eutectic tin/lead as the interconnect alloy. However, many questions remain unanswered regarding lead-free alloys.

By Denis Barbini and Ursula Marquez

Current forced convection reflow ovens are designed to control the heating process with great stability and uniformity. The impact of controlled cooling on joint quality and yield typically is not evaluated. Traditionally, cooling of electronic assembly emphasized the board’s exit temperature and board handling post-reflow. Studies have shown that cooling rates influence the formation of joint microstructure and consequently, joint quality. The benefits associated with increased cooling rates include decreased exit temperatures, minimized time at elevated temperature of the board and minimized exposure of heat-sensitive components and flux/paste to higher temperatures. Conversely, slower cooling minimizes the internal stresses due to differences in the various materials’ Coefficient of Thermal Expansion (CTE) or thermal capacity. The impact of the following factors should be taken into consideration in the reflow soldering of lead-free alloys:

• Time above liquidus (TAL) on interfacial intermetallic growth and joint shear strength;
• Cooling rate on the joint strength and rate of intermetallic formation;
• Linear and ramp-soak-spike profiles;
• Various surface finishes on joint shear strength;
• Aging on joint shear strength.

An experiment was conducted to characterize the cooling process as it relates to lead-free reflow on circuit assemblies where lead-free alloys are used as the interconnect material. As a baseline, tin/lead samples were also assembled using modified profiles. The main objective of this experiment was to study the effects of cooling rates on joint formation, shear strength of solder joints and microstructure formation. Characterization of the cooling rate observed on typical boards using a modern forced convection oven was critical in defining the window of practical manufacturing profile attributes. Two profiles were developed to characterize the range of cooling rates. Results are listed in Table 1. A standard board was used, and the hottest and coldest components were monitored.

FR-4 test boards with three different surface finishes: Copper Organic Solderability Preservative (Cu-OSP), Immersion Tin (Imm/Sn) and Electroless Nickel Immersion Gold (ENIG) were used. Solder spheres of 95.5Sn/3.8Ag/0.7Cu and 63Sn/37Pb were attached to the boards by applying a no-clean tacky flux designed for lead-free processing. Typical manufacturing profiles were simulated in a Differential Scanning Calorimetry (DSC). Parameters investigated included the impact of TAL, 60 or 90 seconds, profile type, linear (L) or ramp-soak-spike (RSS) and the cooling rate starting at a peak temperature of 245°C for lead-free, and 210°C for tin/lead samples. The fastest cooling rate was approximated at -2.5°C/sec and the slowest cooling rate studied was -0.5°C/sec. Shear testing, cross section, SEM analysis and aging studies consequently were performed on the samples to characterize the impacts.

Once coupons were assembled, the shear strength of the soldered joints was measured using an Instron Instrument. The load cell had a maximum capacity of 100 N and the crosshead speed was 0.5 mm/min. The shear test procedure follows the JEDEC Standard JESD22-B117. A total of 16 solder spheres per assembly types were sheared individually, and the maximum force to break was collected dynamically. Representative coupons for each type of assembly also were cross-sectioned to analyze the microstructure evolution and the intermetallic formation using both a Metallograph and Scanning Electron Microscope (SEM).

For the aging study, coupons with Cu/OSP surface finish and solder spheres of SAC and Sn/Pb were used. Samples were assembled using the simulated RSS profile for each material with 90 seconds TAL and slow and fast cooling rates. The samples then were aged at a constant temperature of 125°C for 500 hours. Twenty solder spheres were assembled for each reflow profile.

Methodology

The sample coupons were cut in size and shape to fit the furnace of a DSC working in a nitrogen environment. The substrate material consists of glass-reinforced, tetra-functional FR-4 epoxy with a glass-transition temperature of 175°C and a thickness of 0.81 mm. The pads are non-solder-mask defined with a diameter of 0.56 mm.

Eight reflow profiles for each lead-free and tin/lead assembly were simulated. Reflow profiles were validated by attaching Omega Type-K thermocouple with UV-curable adhesive on the top surface of the coupon. Flux was printed onto the pads with a custom-made mini-stencil and hand squeegee. Solder spheres were hand-placed on the pads using fine tweezers. Solder spheres of 63Sn/37Pb and 95.5Sn/3.8Ag/0.7Cu, with a diameter of 0.76 mm, were chosen for the study. Four solder spheres were assembled on each sample coupon.

Findings

The window for a robust lead-free reflow process already is narrow, as defined by the joint JEDEC IPC-STD-020B standard. By identifying and isolating reflow parameters, impact on joint formation can be investigated and a successful lead-free manufacturing process can be implemented. This experiment has identified several keys to take into account when developing a comprehensive process for lead-free assembly:

• For lead-free samples, slower cooling rates result in the increased formation of the Ag3Sn, as well as Cu6Sn5 (for Imm/Sn or Cu/OSP) or Ni3Sn4 (for ENIG) intermetallics (Figures 1 and 2). These form in the bulk solder and at the pad-solder interface. Therefore, faster cooling rates decrease the rate of formation of these intermetallics. In both cooling rate scenarios, an uneven pad-solder intermetallic structure exists.

Figures 1 and 2. Slower cooling rates in lead-free samples demonstrated increased formation of Ag3Sn, Cu6Sn5 (for Imm/Sn or Cu/OSP) or Ni3Sn4 (for ENIG intermetallics.

• For tin/lead, the behavior is different and similar. It is similar to lead-free in that the microstructure changes from homogeneous to more segregated when the cooling rate changes from -2.5°C/sec to 0.5°C/sec. (Figures 3 and 4). It is different from lead-free in the lack of bulk-solder intermetallics and the formation of a uniform intermetallic layer at the pad/solder interface.

Figures 3 and 4. Tin/lead behavior is similar to lead-free because the microstructure changes from homogenous to segregated when cooling rates change from -2.5°C to 0.5°C/sec. This is different from lead-free in the lack of bulk-solder intermetallics and the formation of a uniform intermetallic layer at the pad/solder interface.

• Intermetallic thickness grows at the solder-pad interfaces proportional to TAL and the aging process. After 500 hours of aging, tin/lead samples cooled at -2.5°C/sec showed an increased intermetallic layer by 93%, while lead-free samples showed an increased intermetallic layer by 13% when compared to time zero. When comparing both alloys, the lead-free intermetallic layer increases by 50% compared to tin/lead assemblies after aging.
• For tin/lead assemblies, the intermetallic is characterized by a thin, uniform layer averaging 1.5 µm at cooling rates of -2.5°C/sec. For lead-free assemblies, however, the intermetallic is characterized by a non-uniform layer of an average of 3.8 µm. This represents an increase of 153% when compared to the same conditions using tin/lead.
• Faster cooling rates result in stronger joints as observed in the shear testing of samples with a Cu-OSP board finish, while opposite results were observed for ENIG boards for tin/lead and lead-free samples. For Imm/Sn boards, a direct comparison was not possible because the failure mode on the shear testing was pad lifting, which was different from the above cases (ductile failure).
• Aging studies at 125°C resulted in decreased shear-strength values. Pad adhesion, solder migration (coarsening) and increased intermetallic thickness at the solder-pad interface impact shear strength. After aging, the median shear strength decreases by 23% and 17% for tin/lead and lead-free samples respectively, when cooled at -2.5°C/sec.

Conclusion

The profile type, linear heating ramp vs. RSS, was not a major factor in final-joint formation. This finding has ramifications in profile development for lead-free. In general, joint formation and resulting quality is dependent greatly upon all reflow process parameters, especially flux-profile compatibility.

Denis Barbini, Ph.D., Advanced Technologies manager, Vitronics Soltec, Inc., may be contacted at (603) 772-7778; e-mail: info@us.vitronics-soltec.com. Ursula Marquez, process and research engineer, Vitronics Soltec, Inc., may be contacted at (603) 772-7778; e-mail: info@us.vitronics-soltec.com.