LOW TEMPERATURE ALLOY DEVELOPMENT FOR ELECTRONICS ASSEMBLY PART II

Transcription

1 LOW TEMPERATURE ALLOY DEVELOPMENT FOR ELECTRONICS ASSEMBLY PART II Morgana Ribas, Ph.D., Sujatha Chegudi, Anil Kumar, Sutapa Mukherjee, Siuli Sarkar, Ph.D. Alpha, an Alent plc Company Bangalore, KA, India Ranjit Pandher, Ph.D., Rahul Raut, Bawa Singh, Ph.D. Alpha, an Alent plc Company South Plainfield, NJ, USA ABSTRACT In this paper, we present details of a very systematic study undertaken for the development of low temperature, leadfree eutectic alloy. Approaches used in alloy development, test methodologies and results are discussed here. Alloy properties targeted for improvements included: Strength, ductility, microstructure stability, thermal cycling and drop shock resistance. At the same time, desirable attributes such as alloy spread and melting temperature are maintained close to the eutectic Sn-Bi. This paper summarizes basic alloy properties, including mechanical, thermal and electrical properties, and paste attributes of a set of new Sn-Bi-X alloys, in which X is a micro-additive. Further, comprehensive reliability studies were undertaken for these new low temperature alloys. Thermal Cycling was performed from -40 C to 80 C with a 30 minute dwell time. Drop Shock studies were also under taken as per the JEDEC JESD22-B111 standard. Improvements obtained are compared to standard Sn-Bi systems and discussed here. Overall, Sn-Bi-X alloys present significant enhancements in metallurgical properties, soldering properties for SMT assembly, and in thermal and mechanical reliability Key words: Lead-free, solder alloy, low-temperature alloy, solder joint strength, thermal cycling, reliability. INTRODUCTION Lead-free Sn-Ag-Cu alloys are the standard alloys for electronics packaging. Introduction of these alloys has brought considerable improvement over eutectic Sn-Pb thermal cycling performance. On the other hand, Sn-Ag-Cu alloys have melting ranges between 217 C and 228 C, requiring reflow temperatures in the range of 245 C to 265 C. Although the electronics industry has adapted to these higher reflow temperatures, a set of very strong drivers is pushing forward the use of lower reflow temperatures. Major benefits of using low temperature alloys are: Assembly of heat sensitive packages and components. Long-term reliability, as low temperature solders reduce exposure to thermal excursion, warpage and other defects caused by excessive heat. Increased process flexibility by enabling multiple reflows of a single board and making possible to use a combination of higher and lower reflow temperatures. Reduced material costs by using low temperature alloy and solder paste, low Tg PCBs and low temperature compatible components. Reduced energy costs through lowering temperature processing. Higher throughputs by reducing reflow / processing cycle time. Reduced labor and equipment maintenance cost. For assembly of temperature-sensitive applications, such as handheld devices, low-temperature solder alloys are then preferred, although Sn-Ag-Cu alloys will continue to have a considerable presence. In this category are alloys that can be used in reflow soldering temperatures from 170 C to 200 C, resulting in lower thermal stresses and defects such as warping during assembly. In most cases, Sn-Bi systems are preferred as compared, for example, to Indium containing alloys which tend to be more costly. However, overcoming some of eutectic Sn42- Bi58 limitations, such as brittleness, poor thermal conductivity, poor fatigue life, and expansion on solidification has been a challenge [1-4]. Use of micro-additives has been successfully used to improve mechanical properties, solderability, and mechanical and thermal reliability of Sn-Ag-Cu alloys. Similar improvements have been achieved with low-ag (SAC) alloys with Bi, Ni, In and Cr small additions [5-8]. In the case of the Sn42-Bi58 eutectic alloy, it was observed that its thermal-mechanical fatigue properties can be improved by small Ag additions [3, 9]. For example, Sn42- Bi57.6Bi-Ag0.4 alloy has been commercialized, resulting in improved mechanical and thermal properties [10].

2 In this Phase II, we present basic alloy properties, solder paste performance, thermal cycling and drop shock performance of selected alloys. Phase I covered the basic alloy properties of the set of alloys under preliminary investigation and are discussed in Reference 11 [11]. In this report we evaluate and analyze C1-A, C1-B and C1-C new alloys, along with standard Sn42-Bi58 and Sn42-Bi57.6- Ag0.4 alloys. This set of alloys provide an insight into the role micro-additives play in improving solder alloy reliability and performance, and are discussed in the following sections. EXPERIMENTAL DETAILS Physical Properties of Alloy Tensile tests were conducted using an Instron model 5566 universal testing machine. Rounded specimens were prepared as per ASTM E8 tensile test standard (16 mm gauge length and 4.01 mm gauge diameter) and the tests were performed at room temperature and 10-3 mm/s strain rate for at least five specimens of each alloy. The average values of ultimate strength, yield strength and elongation are reported here. The melting range, coefficient of thermal expansion (CTE) and thermal conductivity are also reported. Liquidus and solidus temperature were measured using a differential scanning calorimeter (DSC), as per ASTM E794 standard. CTE was measured using a thermal mechanical analyzer, according to the RT- 500C standard. The thermal diffusivity of the alloys was measured using a Nanoflash. In this method the front side of the sample is heated by a light pulse and the resulting temperature signal versus time on the rear surface is measured using an infrared detector. The temperature distribution from one face to inside the sample depends on the thermal diffusivity (α) of the material, which is then used to calculate the thermal conductivity (K). So, Thermal Conductivity (K) =ρ α c p, where ρ is the density and c p the specific heat. Copper dissolution is assessed by measuring the time taken for a wire to break under load when immersed in solder. For this test, a 0.05 mm diameter copper wire was fluxed and dipped in molten solder, which was kept at 190 C. Copper starts dissolving in the solder and after a given time the wire has insufficient strength to support the weight attached to it. This procedure is repeated 4-5 times for each alloy and the rate of copper dissolution is calculated from the time that takes for the copper to dissolve. Application and Reliability of Solder Paste Test vehicles used in this study were reflowed in a seven zone heater reflow machine (Ominiflo7); soak at C for 90sec, 180 C peak and 60sec time above liquidus. Random solder balls (RSB) were evaluated using ceramic coupons, JIS spread and JIS hot slump test used copper coupons, whereas wetting, mid chip solder ball (MCSB), coalescence, and joint cosmetics evaluation was performed using CERF boards, the latter shown in Figure 1. BGA225_12Mil BGA256_20Mil BGA36_15Mi QFP120_90% l FFTC QFP208_90% QFP208_100% BGA56_12Mil BGA56_10Mil BGA256_15M il QFP120_100% Figure 1. Calibrated CERF board used to evaluate solder paste performance Thermal cycling tests were carried out in an air to air chamber at -40 C (30min) 80 C (30min) for 2,000 thermal cycles. The effect of thermal cycling on solder joints of the new alloys was evaluated by removing CERF boards from the chamber at regular intervals for IMC thickness, shear and pull tests. A scanning electron microscope (SEM) was used to analyze solder joint microstructure features on 0402 chip resistors and to measure intermetallic compound (IMC) thickness on 0603 chips. A DAGE 4000 bond tester was used to measure shear force on 1206 chip resistors (as per the JIS Z :2003 standard), and pull force on QFP208 leads (IPC standard). Drop Shock test followed JESD22-B111 standard. A M23 shock tester from Lansmont was used to evaluate the effect of impact and vibration on FusionQuads components. Tests were conducted as per JEDEC Condition B (1500 Gs, 0.5 millisecond duration, half-sine pulse). An example of this shock pulse is shown in Figure 2, along with additional details of the equipment and package. Shock pulse 1500g, 0.5 msec Amkor FusionQuads Details Body Size (mm) 14 x 14 Ext. Leads (0.5mm pitch) 100 Internal Leads 116 Die Size (mm) 7.0 Package Size (mm) 8.0 Figure 2. Details of package and equipment used in drop shock test

3 RESULTS AND DISCUSSION Effect of Alloying Additions on Alloy Properties Basic alloy properties can provide insightful information about solder joint resistance and thermal-mechanical fatigue prior to any further reliability testing. For example, alloy evaluation provides interesting clues about various properties such as mechanical properties, creep/thermal fatigue and thermal conductivity. Figure 3 shows SEM images of the studied alloys microstructure. The various alloying additions result into different levels of microstructure refinement. Addition of Ag results in small refinement of Sn42-Bi58 known eutectic microstructure. However, secondary alloying additions of the new alloys refine the microstructure further. case of C1-C, thermal conductivity is increased further (~24% higher than Sn42-Bi58) due to its secondary alloying addition. Table 1. Melting point, coefficient of thermal expansion (CTE), thermal conductivity (K) and Cu dissolution rate Alloy Melting Cu CTE K Temp Dissol. (ppm/ C) (W/mK) ( C) (µm/min) Sn42-Bi Sn-Bi-Ag C1-A C1-B C1-C (a) (c) (b) (d) Whereas lowering Bi content from Sn42-Bi58 has been shown to increase elongation [4], it also increases the liquidus temperature resulting in off-eutectic alloys. As for drop-in replacement of standard Sn-Bi58 eutectic alloy reducing Bi is not an option, as the new alloys should balance Bi brittleness with good mechanical properties. Table 2 shows a summary of the tensile properties of commercial and new low temperature eutectic alloys. Differences observed in ultimate tensile strength (UTS) are within the test accuracy for most of these alloys. However, it is possible to say that Ag addition and Ag+X addition of C1-C result in higher tensile strength than Sn42-Bi58. Yield strength of Sn42-Bi57.6-Ag0.4 and C1-C is also slightly higher than Sn42-Bi58. C1-A and C1-B have lower yield strength. As expected, Ag addition results in higher ductility, especially in C1-A and C1-C. Figure 3. Microstructure refinement of eutectic Sn-Bi by alloying additions (a) Sn42-Bi57.6-Ag0.4, (b) C1-A, (c) C1-B and (d) C1-C Table 1 shows a summary of thermal properties and copper dissolution behavior of the low temperature alloys studied. As shown, small alloying additions to Sn42-Bi58 alloy as in Sn42-Bi57.6-Ag0.4, C1-A, C1-B and C1-C have little effect on Sn42-Bi58 melting point. These alloys are all eutectic or near-eutectic, i.e., liquidus and solidus temperatures are identical or very close, and, as such, are preferred for processing and as a drop-in replacement of the standard Sn-Bi58 eutectic alloy. Minimal CTE mismatch is also preferred in order to minimize stresses at the solder joint under thermal fatigue. Except for C1-B that has slightly higher CTE, the alloying additions performed resulted in CTE of about 18 ppm/ C. Higher thermal conductivity is always desirable from a device functional point of view. There is no linear relation between thermal conductivity of individual elements and the alloy they form, as the conductivity also depends on the type and amount of intermetallics formed. However, small alloying additions with high thermal conductivity (e.g., Ag and Cu) are expected to increase thermal conductivity of the final alloy. As such, addition of Ag increases in 18.9% the thermal conductivity of Sn42-Bi58 alloy. Similarly, alloys C1-A, C1-B and C1-C, which also contain small Ag alloying additions have higher thermal conductivity. In the Table 2. Tensile properties of the alloys Alloy UTS YS Elongation (MPa) (MPa) (%) Sn42-Bi Sn-Bi-Ag C1-A C1-B C1-C Toughness was estimated through the impact energy measured in the Charpy test, for which the results are presented in Figure 4. Sn42-Bi57.6-Ag0.4 absorbs more impact energy as compared to Sn42 Bi58. Nonetheless, secondary alloying additions of C1-A, C1-B and C1-C resulted in further toughness improvement. These results are particularly interesting as they suggest that further improvement in drop shock performance of Sn42-Bi58 and Sn42-Bi57.6-Ag0.4 is possible. Effect of Alloying Additions on Paste Performance Sn42-Bi58 and Sn42-Bi57.6-Ag0.4 solder powders are commercially available, whereas small batches of type C1- A, C1-B and C1-C powders had to be manufactured for this study. ALPHA CVP520 paste flux was used to prepare the solder pastes analyzed here. Paste attributes and printing performance evaluated included tack, viscosity, coalescence, spread/wetting, JIS hot slump, flux residue, MCSB and joint cosmetics (including black residue). A summary of this evaluation is shown in Tables 3 and 4.

4 220 Table 4. Viscosity, tack, wetting and JIS hot slump performance of the solder pastes Alloy Viscosity Tack (gf) Wetting JIS Hot Slump Impact Energy (mj) Sn-Bi58 M Sn-Bi- Ag0.4 M mm slump 0.3mm slump 205 Sn-Bi58 Sn-Bi- 0.4Ag Figure 4. Charpy impact energy results C1-A C1-B C1-C C1-A M mm slump Tack force was good for all solder pastes (above 100gf), especially C1-B and C1-C. The JIS hot slump evaluates the ability of a solder paste deposit to maintain its shape under ambient controlled heating process. Sn42-Bi58, Sn42- Bi57.6-Ag0.4 and C1-C have similar hot slump properties, bridging at 0.3 mm gaps. C1-A and C1-B has slightly inferior hot slump, bridging at 0.4 mm gaps. Viscosity was very similar for all the solder pastes, except for C1-A and C1-C that presented lower viscosity. C1-B M C1-C M mm slump 0.3mm slump Coalescence and spread/wetting were found acceptable for all solder pastes; no dewetting was observed. RSB was observed for all solder pastes, but in slightly higher amount in the new alloys, which can be attributed to wider particle size distribution of the trial batches. Flux residue was clear and transparent for all pastes, although black residue was observed in all Sn-Bi alloys. Overall, the cosmetics of joints with alloying additions do not differ from those using Sn42-Bi58 alone. Thermal Cycling Performance Cross-sections of chip components after -40 C (30min) 80 C (30min) for 2,000 thermal cycles have not revealed cracks in any of the alloys studied. IMC thickness measurements shown in Figure 5 reveal formation of a μm thin and uniform reactive layer at the solder/pcb interface. The IMC thickness gradually increases upon thermal cycling. After 1500 cycles it increases from about 0.5 to 1.3 μm, reaching 1.5 μm after 2000 cycles. Table 3. Solder pastes performance summary 2.0 Attributes Tack Coalescence RSB Spread Flux Residue Black Residue Comments All samples showed good tack (>100 gf) Good for all samples Observed (preferred or acceptable) Good spread. was observed Residue is clear and transparent Observed black residue in all solder pastes IMC Thickness (µm) Number of Cycles Sn-Bi58 Sn-Bi-Ag0.4 C1-A C1-B C1-C Figure 5. Effect of thermal cycling on IMC thickness The effect of thermal cycling on chip shear and lead pull tests of the various solder pastes is presented in Figure 6. Considering the standard deviation (~1kgf), there is little difference in the initial shear forces. Although no cracks were observed on solder joints after 2000 cycles, except for

5 C1-C, all alloys appear to have a smaller degree of solder joint degradation that is observed through a reduction in their shear force. The effect of alloying additions on the pull force was more noticeable. Similarly, the pull force of the initial solder pastes is basically identical. However, after 2000 thermal cycles the force (standard deviation kg) to pull the leads from the solder joints of C1-B, C1-C and Sn42-Bi57.6-Ag0.4 is clearly higher than of Sn42-Bi58 alloy. 13 Failures, % Variable Sn-Bi-Ag0.4 C1-C Drop Shock Test Results Weibull Shape Scale N AD P > Number of Drops 1000 TV + Fusion Quads Shear Force (kg) (a) Figure 7. Drop shock performance of selected alloys SUMMARY AND CONCLUSIONS The results presented here show that Sn42-Bi58 properties can be improved further by addition of micro-additives. The degree of improvement depends on the nature of the alloying addition and how it interacts with the basic alloy. For the set of eutectic alloys shown here, micro-additions resulted in microstructure refinement, higher thermal conductivity and good paste performance (i.e., tack, coalescence, RSB, spread/wetting). Pull Force (kg) Number of Cycles Sn-Bi58 Sn-Bi-Ag0.4 C1-A C1-B C1-C (b) Figure 6. Effect of thermal cycling on (a) chip shear and (b) lead pull components For handheld electronics, Drop shock performance of the alloy is a critical attribute. Further, better mechanical strength, ductility and ability to absorb impact are expected to contribute to improved shock and vibration properties. This can be achieved by use of micro-additives, as it has been demonstrated in case of SACX and SACX Plus, low- Ag alloys using Bi and Ni micro-additions [5, 6]. Preliminary results show that indeed drop shock performance of Sn42-Bi58 alloy can be improved by alloying additions. For example, as presented in Figure 7, the average number of drops to failure increases from 175 for Sn-Bi-Ag0.4 to 259 for C1-C, i.e., there is a 50% improvement. However, specific additives shown a higher degree of improved performance. For example, C1-C has higher mechanical strength, elongation and impact energy/toughness. Such enhancement of Sn42-Bi48 mechanical properties ultimately resulted in better thermal cycling and drop shock properties. None of the alloys have shown any cracks after 40 C (30min) 80 C (30min) for 2,000 thermal cycles. Further, after 2000 thermal cycles C1-C has higher shear force than any of the other alloys. Similar effect of the alloying additions was also observed on the pull force test; in which C1-B, C1-C and Sn42- Bi57.6-Ag0.4 show higher pull force than Sn42-Bi58 alloy after 2000 thermal cycles. Additionally, preliminary drop shock results show that C1-C has a 50% improvement on the average number of drops to failure compared to Sn42- Bi57.6-Ag0.4 alloy. So far, alloying additions used in C1-C seem to be the most promising to improve mechanical properties and reliability of Sn42-Bi58 alloy. Nonetheless, we continue performing a comprehensive reliability testing of other sets of alloys, which includes thermal cycling and drop shock evaluation. After completion, the results of these tests will be compiled in a final report and published. REFERENCES [1] J. Glazer, Metallurgy of low temperature Pb-free solders for electronic assembly. Int. Mater. Rev., 40 (1995), 65. [2] Z. Mei, H. A. Holder and H. A. Vander Plas, Low- Temperature Solders. Hewlett-Packard Journal, Article 10, August (1996). [3] F. Hua, Z. Mei, J. Glazer and A. Lavagnino, Eutectic Sn-Bi as an Alternative Pb-Free Solder. Proceedings of IPC (1999).

6 [4] H. Takao, A. Yamada and H. Hasegawa, Mechanical Properties and Solder Joint Reliability of Low-Melting Sn- Bi-Cu Lead Free Solder Alloy. R&D Review of Toyota CRDL, 39 (2004), 49. [5] R. Pandher, B. G. Lewis, R. Vangaveti and B. Singh, Drop Shock Reliability of Lead-Free Alloys Effect of Micro-Additives. 57th Electronic Components and Packaging Technology - ECTC, Reno (2007). [6] R. Pandher and R. Healey, Reliability of Pb-free Solder Alloys in Demanding BGA and CSP Applications. 58th Electronic Components and Packaging Technology - ECTC, Orlando (2008). [7] P. Lall, C. Bhat, M. Hande, V. More, R. Vaidya, R. Pandher, J. Suhling, K. Goebel, Interogation of System State for Damage Assessment in Lead-free Electronics Subjected to Thermo-Mechanical Loads. 58th Electronic Components and Packaging Technology (ECTC), Orlando, [8] P. Lall, D. Iyengar, S. Shantaram, R. Pandher, D. Panchagade, J. Suhling, Design Envelope and Optical Feature Extraction techniques for Survivability of SnAg Lead-free Packaging Architectures under Shock and Vibration. 58th Electronic Components and Packaging Technology (ECTC), Orlando, [9] M. McCormack, H. S. Chen, G. W. Kammlott and S. Jin, Significantly Improved Mechanical Properties of Bi- Sn Solder Alloys by Ag-Doping, J. Electron. Mater., 26 (1997) 954. [10] ALPHA CVP-520 Solder Paste Product Guide. [11] 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, Las Vegas (2013).