Sn3.2Ag0.7Cu5.5Sb Solder Alloy with High Reliability Performance up to 175C Open Access

While SnAgCu has been the prevailing choice for SMT assembly and some packaging solder alloy at electronic industry, the adaptability toward next generation automotive applications, including automotive LED application, is challenged due to questionable service temperature range capability [1–10]. For automotive applications, high reliability is a MUST not only under moderate temperature, but also under high service temperature condition. A number of new solder alloys have been attempted by the industry, with major emphasis on enhancing the high temperature capability. However, only limited success has been achieved up to this point, except for one novel solder alloy, with high reliability demonstrated on both moderate and high service temperature conditions. In this work, this new alloy was characterized on its physical properties, mechanical properties, and its soldering performance including wetting and voiding. The reliability performance in TCT, TS, and thermal aging tests will be discussed.

For high reliability with a wide service temperature, one alloy 90.6Sn3.2Ag0.7Cu5.5Sb was developed, denoted as Indalloy276, as shown in Table 1. Also included in this Table are 96.5Sn3Ag0.5Cu (SAC305), and 90.9Sn3.8Ag0.7Cu3Bi1.45Sb0.15Ni, denoted as SACBSbN, also known as Innolot.

The melting behavior of 276 measured by Differential Scanning Calorimetry (DSC) was shown in Figure 1.

The melting temperature of 276 is about 11°C higher than SACBSbN, as shown in Table 1.

In this work, the reliability of solder joints for Si die attached onto Alloy42 leadframe was the primary interest, and the solder joint die shear strength with increasing number of thermal cycling or thermal shock treatment, or with increasing time of high temperature aging, was used to reflect the alloy reliability. Here the CTE mismatch is 13.7 ppm/K, as shown in Table 2. However, during the trial test, the silicon die often cracked before the solder joints failed, thus not able to reflect the joint strength degradation. In order to be able to monitor the solder joint strength degradation, Cu die on Alloy42 was used, with a CTE mismatch of 12.2 ppm/K. Here the CTE mismatch is comparable with that of Si on Alloy42 combination, and Cu die does not crack during shear test.

The ambient temperature mechanical properties alloys listed in Table 1 are shown in Table 3. The shear test curves for solder joints of Cu die (3mm × 3mm) attached onto Alloy42 substrate measured at 140°C and 165°C are shown in Figure 2 and Figure 3, respectively. The solder materials used were solder pastes using Flux B (FB).

As shown in Table 3 and Figure 2 and Figure 3, both SACBSbN and 276 are higher in yield stress and UTS (Ultimate tensile stress) than SAC305. SACBSbN has the lowest ductility, and the rapid drop in stress failure pattern shown in Figure 2 and Figure 3 indicated it was brittle failure. SAC305 was higher in ductility than 276 at ambient temperature. But, the relative ranking reversed at elevated temperature, 140°C and 165°C, with 276 being more ductile than SAC305, as shown in Figure 2 and Figure 3.

Two reflow profiles under air atmosphere were used in this study, as shown in Figure 4 with a peak temperature of 245°C and Figure 5 with a peak temperature of 255°C. Unless otherwise specified, the 255°C peak temperature profile was used.

The voiding was monitored by X-ray. For samples processed with 255°C peak temperature profile, the voiding results are shown in Figure 6. Flux B (FB, also known as 10.1HF) showed less voiding than Flux A (FA, also known as 8.9HF). SAC305 showed the lowest voiding, with 276 and SACBSbN being comparable in voiding amount. Cu die on OSP showed lower voiding than Cu die on Alloy42.

The shear strength of as-reflowed solder joints measured at room temperature for various combination of materials and solder alloys are shown in Figure 7. All solder pastes were reflowed with Peak255 profile.

Several trends were observed. First, Cu-Cu resulted in higher shear strength than Cu-Alloy42, presumably due to better wetting on Cu than on Alloy42. Secondly, 276 showed a higher shear strength than both SAC305 and SACBSbN. Thirdly, Flux A (FA) showed a higher strength than Flux B (FB). However, the relative ranking of flux on shear strength could not be observed in Figure 8 to be discussed next, suggesting a possible data scattering effect.

The shear strength of as-reflowed solder joints measured at 25°C, 65°C, 105°C, 140°C, and 170°C were shown in Figure 8.

In general, the shear strength decreased with increasing temperature, except for the initial increase in shear strength at 65°C. The latter could be attributed to annealing or stress relaxation effect. The shear strength of 276 and SACBSbN were comparable, and both were higher than SAC305. The effect of flux type was unclear, suggesting an insignificant effect on shear strength.

The reliability of solder joints was assessed by preconditioning the solder joints with various TA (thermal aging) or TCT (thermal cycling test) or TST (thermal shock test) treatment, followed by measuring the shear strength.

Figure 9 showed solder joint shear strength measured at room temperature after the mild TA (thermal aging) at 125°C up to 2016 hours.

For automotive applications, 125°C aging condition is considered mild condition. Under this condition, 276 is comparable with SACBSbN in shear strength, and the two alloys are equal or higher than SAC305. Here the flux factor also is insignificant.

For the more stressed TA treatment at 175°C, after 2088 hours SACBSbN showed a shear strength about 40% of 276 for FA system, and about 75% of 276 for FB system, as shown in Figure 10.

For mild TCT (−40/125°C) condition, the shear strength against cycling number is shown in Figure 11. After 2238 cycles, the shear strength of 276 is comparable with SACBSbN, and both of them are higher than SAC305.

For harsh TCT (−40°C/175°C) condition, SAC305 dropped to less than 50% of initial strength after 1000 cycles. SACBSbN and SAC305 exhibited a shear strength about 1/20-1/11 of 276, as shown in Figure 12. Flux-wise, FA is comparable with FB.

For the harsh TST conducted at −55°C/155°C, the results were shown in Figure 13. The cycling detail was 10 min dwelling time, ~24 min per cycle.

With increasing TST cycling number, 276 retained strength very well, followed by SACBSbN, with SAC305 declined most rapidly. After 3000 cycles, the shear strength of 276 is about 8 times of SACBSbN and SAC305. Joints with FA showed slightly higher shear strength than that of FB.

The cross-sectional view of samples using FB after TST (−55/155°C) and TCT (−40/175°C) treatment was shown in Figure 14. The dark arrow signs showed the cracks in the joints. In most incidences, the crack appeared to initiate from the edge of the joints. In the TCT treated samples, both SAC305 and SACBSbN showed obvious cracks after 1000 cycles, 276 showed signs of crack after 1500 cycles, and minor cracks after 2343 cycles.

Figure 15 showed some close-up look of joints after TCT treatment. The crack in SAC305 is very obvious.

For TST treated samples, SAC305 showed cracks after 850 cycles, while 276 and SACBSbN still remained intact.

For samples using FA, Figure 16 showed cross-sectional view of samples after 2343 TCT cycles (−40/175°C). Clear delamination observed for SAC305 and SACBSbN, while that of 276 still remained intact.

Figure 17 showed cross-sectional view of samples using FA after TST (−55/155°C) up to 2250 cycles. Both SAC305 and SACBSbN showed severe cracks after 850 cycles, while 276 remained intact at 850 cycles, and showed initial cracks after 2250 cycles.

The samples after TCT (−40/175°C) or TST (−55/155°C) treatment were examined by X-ray for delamination and voiding, followed by Dye-and-Pry treatment for crack or delamination.

Figure 18 showed samples using FB after TCT 2343 cycles. The light-colored regions on substrate marked with red arrows were surviving joined area. SAC305 and SACBSbN were barely connected, while joints of 276 largely remained intact. For SACBSbN, the peripheral dark region of X-ray picture was attributed to the warped Cu die.

Figure 19 showed samples using FB after TST 3000 cycles. SACBSbN virtually fully separated. SAC305 barely remained connected, while joints of 276 largely remained intact.

The performance of samples using FA and FB was fairly similar. Figure 20 showed samples using FA after TST 3000 cycles. SAC305 virtually was fully separated. SACBSbN barely remained connected, while joints of 276 largely remained intact.

The reliability data is summarized and listed in Table 4.

Table 4 showed that under mild test condition, 276 was comparable with SACBSbN, and both performed better than SAC305.

However, under harsh condition, 276 was much better than SACBSbN, which was equal or better than SAC305.

The reliability summarized in Table 4 can be correlated with the mechanical properties.

Both 276 and SACBSbN are alloys based on SnAgCu, but reinforced with precipitate hardening and solution hardening, with the use of additives including Sb, Ni, and Bi. The high rigidity resulted from reinforcement should result in higher creep resistance, which promised a better thermal fatigue life than SAC305 under stressed condition [11, 12], as reflected in Figure 9 to Figure 13.

As shown in Figure 2 and 3, both 276 and SACBSbN showed a higher maximum stress than SAC305. This suggests that both alloys may deform to a less extent than SAC305 under typical stressed condition, hence may promise a better joint integrity under mild condition.

However, there is a significant difference between 276 and SACBSbN. 276 is rigid and ductile, while SACBSbN is rigid but brittle, as shown in Figure 2 and Figure 3. Under the harsh test condition where ΔT was high, the dimension mismatch between parts and substrate became very significant due to CTE mismatch. This significant dimension mismatch would cause a brittle joint to crack quickly, as seen on SACBSbN. The challenge was more tolerable for a ductile joint, as shown by 276. Accordingly 276 showed a much better reliability than SACBSbN under harsh condition, including high testing temperature and large ΔT.

The nature of solder ductility or brittleness were exemplified by the joint fracture pattern as shown in Figure 21, 22, and 23 for as-reflowed SAC305, SACBSbN, and 276, respectively.

The ductility can be closely related to the microstructure of solder joints.

In SACBSbN, presence of blocky Ag3Sn plates or rods noted by the arrows can be observed easily, as shown in Figure 24. On the contrary, 276 showed a much finer and homogeneous microstructure. This non-homogeneity microstructure of SACBSbN strongly suggested that it will have a greater difficulty to exhibit a ductile behavior than other joints with a more homogeneous structure.

The effect of homogeneity of microstructure on reliability can also be exemplified by examining the solder joints after 850 cycles of −55/155°C thermal shock treatment, as shown in Figure 25 and Figure 26 for SACBSbN and 276, respectively. Figure 25 showed presence of a higher concentration of a variety of IMC particles for SACBSbN, compared with the lower concentration of IMC particles for 276 joint shown in Figure 26.

Similar impact of non-homogeneity can also be observed in TCT test (−40/175°C). Figure 16 showed the joints after 2343 cycles. SACBSbN exhibited a greater number of IMC particles than 276, and cracks can be seen easily in SACBSbN joint.

Apparently, to achieve high reliability under a wide service temperature environment, a balanced ductility and rigidity for solder alloy is critical for success.

A novel lead-free solder alloy 90.6Sn3.2Ag0.7Cu5.5Sb, designated as Indalloy276, was developed targeting for high reliability with a wide service temperature capability. The alloy exhibited a melting temperature range of 223 to 232°C, reflowable at profile with peak temperature 245°C and 255°C, with ambient temperature Yield Stress 60MPa, UTS 77 MPa, and ductility 28%, and a higher stress than both SAC305 and SACBSbN, the latter two alloys were used as controls. When tested at 140°C and 165°C, the die shear stress of 276 was comparable with SACBSbN but higher than SAC305, and the ductility was higher than both SACBSbN and SAC305, with SACBSbN exhibited distinct brittle behavior. When aged at 125°C and 175°C, the die shear strength of 276 was comparable or higher than both controls. When pretreated with a harsh condition, TST (−55°C/155°C) for 3000 cycles, the die shear strength of 276 was 8 times of that of SACBSbN and SAC305. When pre-conditioned at TCT (−40°C/175°C) for 3000 cycles, the die shear strength of 276 was 11 to 20 times of SACBSbN and SAC305, depending on the flux type used. Both 276 and SACBSbN are alloys based on SnAgCu, but reinforced with precipitate hardening and solution hardening, with the use of additives including Sb, Ni, and Bi. 276 exhibited a finer microstructure with less particles dispersed, while SACBSbN exhibited more particles with some blocky Ag3Sn plates or rods. 276 is rigid and ductile, while SACBSbN is rigid but brittle. Under the harsh test condition where ΔT was high, the dimension mismatch between parts and substrate became very significant due to CTE mismatch. This significant dimension mismatch would cause a brittle joint to crack quickly, as seen on SACBSbN. The challenge was more tolerable for a ductile joint, as shown by 276. Accordingly 276 showed a much better reliability than SACBSbN under harsh condition, including high testing temperature and large ΔT. Overall, to achieve high reliability under a wide service temperature environment, a balanced ductility and rigidity for solder alloy is critical for success.

The authors would like to acknowledge the significant support of Christine LaBarbera on the microstructure characterization of solder joints.