■ Table 1: Common options for Level I and Level II interconnects materials for LED packaging and assembly
Establishing a solder processing temperature hierarchy with traditional tin/silver/copper (SAC) as the level 1 alloy has been a major challenge in SMT applications. The lower temperature options have been plagued with low operating temperature window (too low melting temperature) and poor mechanical strength of the material.
In this study we present process and reliability data for a new off-eutectic alloy with melting point range (192- 203°C). This alloy has the potential to fill a key gap in the alloy choices for Level 2 applications for LED, computer, server, and telecommunications applications. This solder alloy, being significantly different from the conventional lead-free solders, necessitates a completely different flux chemistry. Solder material properties and the solder paste processing through SMT assembly is discussed in detail. The alloy reliability (drop-shock and thermal cycling) was benchmarked against traditional Pb-free solders currently in the market.
Apart from traditional SMT assembly, alloy functional performance in LED applications was also investigated. Detailed results of thermo-mechanical reliability tests of the solder as well as the impact of flux chemistry on the LED performance over time are presented.
■ Figure 1. LED Al-MCPCB test vehicle.
A typical packaging/assembly process of the LED or any other semiconductor device involves several steps in a series. At each of these steps, some kind of electrical/thermal interconnect material is used. The most common materials used in interconnects are conductive adhesives, sintered materials, soldering alloys, epoxies, and polymers. Among these, conductive adhesive, sintering materials, epoxies and polymers go through an irreversible process phase transition during the process. However, solders go through a quasi-physical and mostly reversible phase transition during processing. Solders are the most common materials used in electronics assembly. The fixed and reproducible melting point of a solder is an advantage as far as formation of the interconnects is concerned.
However, the same property of solders becomes a liability if the assembly process involves multiple steps using the same solder. Solder used in the first assembly process will go through repetitive melting and freezing cycles during secondary, tertiary and additional process steps. The quality of the interconnect going through multiple melting/freezing cycles will degrade resulting in reduced life cycle of the final product or the product with compromised reliability. Therefore, there is a need for soldering materials with different melting/freezing temperatures.
■ Figure 2. Reflow profile for LEDs assembled with mid- temp solder paste.
Before the electronics industry transitioned to Pb-free solders, different compositions of SnPb solders were being used to create soldering temperature hierarchies. For example, 10Sn90Pb or 05Sn95Pb or some of their variants with some Ag addition were used as die attach materials (also called level I interconnect in LED packaging and assembly terminology) because of their high melting temperatures (299C for 10Sn90Pb and 310C for 05Sn95Pb). 63Sn37Pb or some of its variants with melting temperature around 183C were used for circuit board assembly (also called Level II -V in the LED Industry).
After the restriction of Pb in solders, the most common Pb- free solders that emerged were SnAg or SnAgCu (SAC) solders with melting temperatures in the range of 217- 228C. The only viable option for high-temperature Pb-free solder to replace high-Pb solders was 80Au20Sn. With 80% gold, Au20Sn is one of the most expensive solders. In addition, AuSn is a high modulus, relatively brittle material that results in high stress interconnects. Even though it has 80% gold, its thermal and electrical conductivity is slightly lower than SAC or SnAg eutectic solders. Despite its expense and other drawbacks, AuSn solder is still being used because of its easy processing on standard SMT lines. Competing technologies, such as conductive adhesives and sintered materials, require different processing and, in some cases, may even need a special equipment.
■ Figure 3. X-ray images showing voids in the solder joints.
Table I shows common options for Level I and Level II interconnect materials for LED packaging and assembly. Similar material sets are being used in other semiconductor packaging and electronic assemblies as well. Some of the properties such as high thermal conductivity and high reliability are even more important for packaging and assembly of high power electronics components such as power diode, MOSFET and IGBT etc.
SnZn-based alloys. However, zinc-containing alloys oxidize easily, show severe drossing in wave solder pots, are prone to corrosion, and have a shelf life that is measured in terms of days or weeks compared to months for eutectic Sn-Pb or SAC solders . Because of the highly corrosive behaviour of such solders, many end users avoid using Zn based solders. Therefore, a search for alternate mid- temperature solder, not based on SnBi or SnZn platforms continues. The metallurgists at Alpha Assembly Solutions has been trying to find a mid-temperature solder hierarchy for multi-level electronics assembly such as high power LEDs for outdoor
■ Figure 4. Shear strength of LED packages assembled on FR4 PCB after temperature cycling.
lighting, power electronics, and harsh environments. In this paper we present a mid-temperature solder that addresses all the shortcomings of SnBi-or SnZn- based systems. Major constituents of the alloy are Sn, Ag, Cu, Bi and In. Fraction of In and Bi is so that it does not precipitate any low-temperature melting phase.
One of the options is to use conductive adhesive or sintered Ag for Level I (Die attach) and Pb-free solder for Level II attachment. Advantages of this option are high temperature stability of the Level I interconnect and high thermal conductivity of the sintered Ag material. However, the disadvantage is often a need for unique, very different processing conditions from typical SMT. For Level I assembly this includes needing specialized equipment depending on the type of conductive adhesive or sintered Ag paste used.
There are limited options if one has to use solders for both L1 and L2 assemblies. One such option is to use SAC305/SAC405 for Level I and a SnBi-based solder (such as 42Sn58Bi, 42Sn57.6Bi0.4Ag or SBX02) for Level II assembly [1 – 6]. There are two major problems with this material set. First, SnBi-based solders have poor mechanical strength
■ Figure 5. Shear strength of LED packaged assembled on Al core MCPCB after temperature cycling.
and are very brittle. Second, all of these solders’ solidus temperature is ~138C which may be too low for certain applications such as high-power LEDs in outdoor lights or in automotive head lamps where operating temperatures might be too close the solder solidus temperature. In the recent past, there have been several attempts to improve the mechanical properties and thermo-mechanical reliability of SnBi based solders. While significant progress has been made to improve their mechanical properties, there is not much success in increasing the solidus temperature above SnBi eutectic.
Therefore, there is a need for alternate low temperature Pb- free solder with higher solidus temperature than SnBi eutectic and lower liquidus temperature than SAC solders. This would be used in the second level assembly when the level I is assembled with SAC305/SAC405/SnCu/SnAg etc. There have been some efforts in the electronics industry to develop solders with melting points in the 190- 200C range. Most of these attempts have been focussed on
■ Figure 6. Change in luminous flux on LEDs on AL- MCPCB as a function of temperature cycles.
A detailed study of the new mid-temperature solder alloy was taken to assess its reliability as compared to SAC305. Since practically all the Level II assembly is done on SMT assembly lines, solders were prepared in the paste form with appropriate fluxes for both pastes. Before undertaking reliability testing, general paste performance tests were run to make sure the paste performance conforms to SMT assembly requirements. Figure 1 shows the aluminum core MCPCB test vehicle designed for commercially available mid-power white LEDs used in this study. An identical pad design was used for PCB with FR4 core. Commercially available packaged white LEDs, Lumileds Luxeon 3535L, were used in these experiments.
All the parts assembled with mid-temp solder paste were reflowed under as follows: Soak 80 s at 145-170 C, TAL: 80 s above 200 C and Peak: 215 C in air. Figure 2 shows a record of the profile at one of the component sites.
■ Figure 7. Change in luminous flux on LEDs on FR4-PCB as a function of temperature cycles.
The shear strength of components
assembled with Mid- temp and SAC alloys was used as a measure of the mechanical integrity of the solder joints. To assess the reliability of the solder interconnects the assembled parts were subjected to temperature cycling. In addition, any impact of thermal cycling on the LED performance was also evaluated.
THERMAL CYCLING TEST:
• Test according to IPC 9701-A standard
• -40oC (10min) to 125oC (10min) for 2000 cycles.
• Lumileds Luxeon 3535L
• Finish: ENIG
• Performance outputs:
• Solder joint dhear strength
• Luminous flux
• LED junction temperature and thermal resistance.
■ Figure 8. Junction temperature of LEDs assembled on AL- MCPCB using mid temp solder and SAC305, as a function of temperature cycling.
SOLDER JOINT SHEAR TEST
To assess the mechanical integrity of the solder joints, joints were sheared and peak shear force was recorded for each of the sheared components. A Dage 4000 shear tester was used to shear the entire Luxeon package. Eight components were sheared from each set. The shear height was set low enough to make sure that level II interconnects were sheared and to minimize the rolling effect. Samples were removed from thermocycling after 500, 1000, 1500 and 2000 cycles and sheared to record any drop in solder joint strength.
Results and discussion
Following assembly of the LEDs, all the solder joints were x-rayed to assess voiding and general integrity. All the joints showed acceptable voiding. Figure 3 shows typical x-ray images of the assembled LEDs.
Package shear test data
■ Figure 9. Junction temperature of LEDs assembled on FR4- PCB using mid-temp solder and SAC305, as a function of temperature cycling.
Figure 4 shows the normalized shear strength of the LED packages assembled on FR4 boards using mid-temp and SAC305 solders. After 2000 cycles the shear strength of both mid-temp and SAC305 joints dropped to about 75% of the original strength. Similar shear strength of the LEDs assembled on Al-MCPCB is shown in figure 5. The shear strength of SAC joints on Al-MCPCB drops down to about 73% of the original. Similarly, shear strength of the Mid- Temperature solder joints of the same LEDs on Al- MCPCB drop to only about 82% of the original. The mechanical strength of LED solder joints on AL-MCPCB show slightly different behavior as compared those on FR4 board. This is likely due to higher CTE mismatch between Al and Alumina (LED substrate material) as compared to that between FR4 and Alumina .
After 2000 cycles, there is less than 5% drop in the average luminous flux. This drop is same for LED with Mid- Temperature solder and SAC305 in Level II interconnects.
Even though the shear strength of the joints drops after temperature cycling, the failure mode remains cohesive. All the joints show a failure in the bulk solder with no evidence of cracks at the interface. Therefore, any change is shear strength is a result of the change in the bulk strength of the solder, which is usually related to the change in solder microstructure.
■ Figure 10. Luminous Flux of LEDs assembled on AL- MCPCB and FR4-PCB using mid-temp solder as a function of temperature cycling.
In addition to the mechanical strength of the interconnect material and its behavior under temperature cycling, the LEDs’ optical performance was also evaluated. Level II interconnects have smaller impact on LED performance as compared to level I interconnects. However, if there is significant degradation of the interconnect quality then the level II layer can have an impact on LED performance. Therefore, luminous flux, junction temperature and the total thermal resistance of the LED stack were measured as well.
Figure 6 shows the change in luminous flux after temperature cycling of LEDs assembled with the mid- temperature solder and SAC305 solder on Al-MCPCB. Mid-temperature alloy and SAC show similar performance.
Figure 7 shows the change in luminous flux under temperature cycling of the LEDs assembled with the mid- temp solder and SAC305 solder on FR4 PCB. The mid- temperature alloy samples showed similar performance to SAC305. This means mid-temp solder can replace SAC305 solder for LED Level II interconnects without compromising performance. The actual cause of the drop in luminous flux upon aging is not clear from this data. The level II interconnect may or may not have a role in the drop in luminous flux upon temperature cycling.
■ Figure 11. Thermal resistance of LEDs assembled on AL- MCPCB using mid temp solder and SAC305, as a function of temperature cycling.
Figures 8 and 9 show the LED junction temperature of LEDs assembled on Al-MCPCB and FR4-PCB respectively. LEDs operated at their specified maximum operating current (300mA). The junction temperature of LEDs on AL-MCPCB is about 48C while of those on FR4- PCB is about 115C. This is not surprising because the thermal conductivity of Al is 240 W/m ∙ k while that of FR4 is < 1 W/m ∙ k (through plane). In both the cases, the junction temperature does not change after temperature cycling which means the level II interconnects’ mechanical and/or electrical performance don’t change. The behavior of mid-temp solder joints is identical to SAC305 joints.
Higher junction temperatures over the lifetime of an LED will decrease its luminous efficiency. Figure 10 shows the luminous flux of LEDs assembled on Al-MCPCB and FR4- PCB with mid-temperature solder. The luminous flux of LEDs assembled on AL-MCPCB is about 15% higher than those on FR4-PCB.
Figures 11 and 12 show the thermal resistance of the stack between the LED and the heat sink and its variation over 2000 temperature cycles. Thermal resistance is calculated using the measured junction temperature, luminous output and known applied electrical power. Details of these measurements are shown in other publications [8,9]. This thermal resistance includes contributions from all the layers in the LED substrate, including: LED die, die attach layer (level I), level II interconnect, PCB the and thermal interface material used to attach the PCB to the heat sink . The average thermal resistance of LEDs on AL- MCPCB is about 40 C/W while it is about 140C/W for those on FR4-PCB.
■ Figure 12. Thermal resistance of LEDs assembled on FR4- PCB using mid-temperature solder and SAC305, as a function of temperature cycling.
As seen in figures 11 and 12, the thermal resistance of the LED assembled with the mid-temperature solder is comparable with those assembled with SAC305. Also, there is no noticeable change in the thermal resistance of LEDs assembled with either of the solders after 2000 temperature cycles. This is further evidence that level II solder joints formed with mid-temperature solder are as good as those made with SAC305 and are stable through temperature cycling up to 2000 cycles.
Observations and conclusions
A mid-temperature lead-free solder alloy has been developed. The melting temperature of this alloy falls in between common Pb-free SnAgCu solders and low- temperature Pb-free SnBi-based solders. This mid-temperature solder provides an alternate solder for level II interconnects where level I interconnects use a SAC solder. Unlike SnBi solder, this new mid-temperature solder is suitable for high-temperature operating conditions such as outdoor and automotive lighting. This mid-temperature alloy performs as well as SAC305 in all reliability tests typical of LED lighting modules. There is some drop in shear strength of the joints after temperature cycling but all the joints break in the bulk of the solder and show no evidence of any cracks at the interface.
Level II interconnect materials have limited impact on LED performance. All the LEDs measured showed a drop in luminous flux after temperature cycling. However, there was no change in LED junction temperature or the thermal resistance of the stack. Therefore, neither any change in the LED dies nor any change in thermal/electrical interconnect is responsible for the drop in luminous flux. The only other possible cause for the drop in luminous flux is potential degradation of the phosphor. Further experiments are underway to verify that.
The authors would like to thank Joseph Adinolfi for light output measurement of LEDs, testing and analysis.
1. M. Ribas, S. Chegudi, A. Kumar, R. Pandher, S. Mukherjee, S. Sarkar, R. Raut and B. Singh, “Low Temperature Alloy Development for Electronics Assembly”, IPC APEX, San Diego – USA, February 2013.
2. Morgana Ribas, Sujatha Chegudi, Anil Kumar and Ranjit Pandher, “Development of Low- Temperature Drop Shock Resistant Solder Alloys for Handheld Devices” IEEE 15th Electronics Packaging Technology Conference (EPTC 2013), Singapore, December 2013.
3. M. Ribas, S. Chegudi, A. Kumar, R. Pandher, S. Mukherjee, S. Sarkar, R. Raut and B. Singh, “Low Temperature Alloy Development for Electronics Assembly – Part II”, SMTA International, Fort Worth – USA, October 2013.
4. M. Ribas, S. Chegudi, A. Kumar, R. Pandher, R. Raut, S. Mukherjee, S. Sarkar and B. Singh, “Thermal and Mechanical Reliability of Low- Temperature Solder Alloys for Handheld Devices”. IEEE 16th Electronics Packaging Technology Conference (EPTC), Singapore, December 2014.
5. Morgana Ribas, Anil Kumar, Ranjit Pandher, Rahul, Raut, Sutapa Mukherjee, Siuli Sarkar., and Bawa Singh, “Comprehensive Report on Low Temperature Solder Alloys for Portable Electronics” Proceedings SMTA International Technical Conference, Rosemont, IL, USA September – October 2015.
6. Emmanuelle Guéné, “Development and Characterization of Solder Pastes Based on Two Alternative Alloys: Bismuth-Tin-Silver (BiSn42Ag0.4-1%) for Low Temperature and Tin-Antimony (SnSb5-8.5) for High Temperature“, Proceedings SMTA International Technical Conference, Rosemont, IL, USA September – October 2015.
7. Carol Handwerker, Ursula Kattner and Kil-Won Moon, “Fundamental Properties of Pb-Free Solder Alloys” in Chapter 2 in Pb-free Soldering, Bath, J. (Ed), Springer 2007.
8. Nicholas Herrick and Ranjit Pandher, “Thermal, Optical, and Electrical Performance of LED Die Attach”, SMTA LED A.R.T. Conference, Atlanta, GA, Nov 2016.
9. Nicholas Herrick, Amit Patel, Gyan Dutt and Ranjit Pandher, “High Performance Electronic Interconnect Materials Characterization – Techniques & Challenges”, SMTAI International Technical Conference, Rosemont, IL, USA, September 2017.
10. Ranjit Pandher, Ravi Bhatkal and Kurt-Jürgen Lang, “Impact of Substrate Materials on Reliability of High Power LED Assemblies”, Proceedings SMTA International Technical Conference, Rosemont, IL, USA, September 2016.