Characterization of SnAgCu and SnPb solder joints on low-temperature co-fired ceramic substrate

The Authors

Z.W. Zhong, School of MAE, Nanyang Technological University, Republic of Singapore

P. Arulvanan, Singapore Institute of Manufacturing Technology, Republic of Singapore

Hla Phone Maw, Singapore Institute of Manufacturing Technology, Republic of Singapore

C.W.A. Lu, Singapore Institute of Manufacturing Technology, Republic of Singapore

Abstract

Purpose – The purpose of this paper is to present the results of experiments performed to attach silicon dies (chips) to low-temperature co-fired ceramic (LTCC) substrates with Ag or AgPd pads using SnAgCu or SnPb solder and the results of the characterization of the solder joints.

Design/methodology/approach – LTCC substrates were fabricated by stacking and laminating four green tapes with the top layer screen-printed with Ag or AgPd paste to form pads. Silicon die sizes of 1 × 1 mm and 2 × 2 mm with electroless nickel immersion gold plated were soldered to 2 × 2 mm pads on the LTCC substrates using SnPb or SnAgCu solder. The solder joints were then characterized using X-ray, die shear, energy dispersive X-ray and scanning electron microscopy techniques.

Findings – The joints made by AgPd pads with SnAgCu solder provided the best results with the highest shear strength having strong interfaces in the joints. However, the joints of Ag pads with SnPb solder did not provide high-shear strength.

Originality/value – The findings provide certain guidelines to implement LTCC applications. AgPd pads with SnAgCu solder can be considered for applications where small silicon dies need to be attached to LTCC substrates. However, Ag pads with SnAgCu solder can be considered for lead-free solder applications.

Article Type:

Research paper

Keyword(s):

Solders; Joining materials; Substrates.

Journal:

Soldering & Surface Mount Technology

Volume:

Number:

Year:

2007

pp:

18-24

Copyright ©

Emerald Group Publishing Limited

ISSN:

0954-0911

1 Introduction

As integrated circuit (IC) fabrication advances rapidly, microelectronic packaging faces new challenges (Zhong, 1999; Palm et al., 2001). Solder joints can provide mechanical and electrical connections and enhance heat spreading efficiency, and therefore are widely used in microelectronic packaging (Yang et al., 2004a). However, lead contained in tin-lead solders is harmful to the environment (Moscicki et al., 2005).

Environmental concerns and legal constraints (Bigas and Cabruja, 2006a; Kamal and Gouda, 2006) have led to replacing tin-lead solders with lead-free solders (Chan et al., 2003) or adhesives (Palm et al., 2003) in manufacturing of microelectronic devices. There is also a strong demand to find lead-free alternatives for automotive and other high-temperature applications (McCluskey et al., 2006). Incorporation of lead-free techniques is better than resistance to change (Ciocci and Pecht, 2006).

However, a great deal of tests on lead-free solder joints are needed before they can replace tin-lead solders (Koo and Jung, 2005). Sufficient material and process data for lead-free solder interconnections are lacking and therefore research is being conducted worldwide.

The quality of lead-free solder joints is compared to that of tin-lead solder joints (Rocak et al., 2007). Traditional failure analysis techniques are widely applicable to analyses of lead-free and lead-containing solder joints (Castello et al., 2006). Potential failure modes and causes of lead-free solder joints are studied (Plumbridge, 2006).

Electro-deposition of Sn-Ag has been developed to improve the flip chip technique (Bigas and Cabruja, 2006b), and a pitch of 40-50 μm can be obtained. For high-quality fabrication of flip chip Sn-Ag balls, the material should be deposited uniformly over the whole wafer surface without voids and contamination (Helneder et al., 2005). Thermomechanical fatigue (TMF) tests of Sn-Ag based solder joints were performed with different heating rates, and surface damage and residual mechanical strength of the solder joints were characterized to evaluate the effect of TMF heating rate (Lee and Subramanian, 2007).

Investigation of mechanical properties of Sn-Cu is also a significant work for microelectronics packaging (Zhu et al., 2007). Sn-0.5Cu, Sn-3.5Ag, Sn-5Sb and Sn-9Zn, were rapidly solidified by a melt-spinning technique (Kamal and Gouda, 2006) to study the effects of rapid solidification.

Lead-free solders are multi-components (Ghosh, 2004). The best lead-free alloys are those based on the SnAgCu eutectic system (Harrison et al., 2001), and therefore are the most promising materials to replace tin-lead solders (Helneder et al., 2005).

Lead-free solder alloys, such as SnAgCu, have a liquidus in the range of 220°C and higher, resulting in higher reflow temperatures typically above 240°C. This can negatively affect product reliability due to higher residual stresses in assemblies (Wood, 2003). Higher levels of voids are reported in lead-free solder joints than eutectic SnPb joints. Voids degrade electrical, mechanical, and thermal properties of solder joints (Setty et al., 2005). Interfacial preexisting voids can accelerate electromigration failures, which can reduce over a half of the solder joint lifetime with a 20 percent volume fraction of interfacial voids (Tang and Shi, 2001).

Intensive research on lead-free solders can lead to a better understanding of these new materials in terms of their properties and reliability (Viswanadham et al., 2004). Investigation of ball-grid-array (BGA) components with SnAgCu solder balls found an intermetallic of SnNiCu on NiAu pads after a thermal cycling test (Anand and Mui, 2002). Lead-free solder joints have fine and stable microstructures due to formation of small-dispersed particles and consequently higher shear strength values than tin-lead solders (Ye et al., 2000). Microstructures of Sn and plate like Ag₃Sn were observed with dendritic Sn phase for SnAgCu alloys including Cu₆Sn₅ in SnAgCu eutectics (Mutoh et al., 2002). The Ni₃Sn₄ intermetallic was found uniform and thin in Sn96Ag3.5Cu0.5, while it was thick and less regular in Sn95.5Ag4.0Cu0.5 under dual reflow conditions using electroless Ni-P for BGA assemblies (Minogue, 2002).

A precipitate of Cu₆Sn₅ having a hexagonal crystal structure with a grain size of 0.3 μm mainly distributed in the Sn matrix, and finely dispersed voids along the intermetallic layer were found (Ye et al., 2001). A phosperous-rich layer was found on all electroless-nickel-immersion-gold (ENIG) plated surfaces, and Ni/Au finish produced numerous small voids around the intermetallic formation (Dunford et al., 2002, 2003). The intermetallic compounds consisting of CuSn (Roubaud et al., 2002) and the CuAuNiSn compound (Shiau et al., 2002) were found. The failure mechanisms were also studied (Stam and Davitt, 2001). Four no-clean, SnAgCu alloy-based solder pastes consisting of different flux systems were also evaluated experimentally (Manjunath et al., 2006). There are also recent reports on reliability of SnAgCu solder (Arulvanan and Zhong, 2006; Braun et al., 2006; Roellig et al., 2007).

Meanwhile, the low-temperature co-fired ceramic (LTCC) technology has demonstrated great potential to meet the requirements for higher density and performance packages, especially in the high-end radio frequency packaging industry (Zhang et al., 2003). The ability to co-fire many layers simultaneously has the advantage of reducing both the process cost and process variability (Shapiro et al., 2002). The LTCC technology can co-fire all the required layers at a temperature below 950°C (Birol et al., 2005). Oxide type powders and low-melting glass frits may be mixed with ceramic materials to reduce the sintering temperature to below 900^°C (Jung et al., 2005).

LTCC is the next generation technology for electronics packaging (Jantunen et al., 2003). A typical LTCC circuit contains surface conductors for connecting components and internal and via-fill conductors for making up the internal components of the circuit (Lopez et al., 2004). The multilayer ceramic packaging technology consists of green tape preparation, metallization layout, lamination and cofiring (Yang et al., 2004b). LTCC tapes shrink by more than 10 percent (Matters-Kammerer et al., 2006). There are efforts to target zero shrinkage (Rabe et al., 2005). Applications of LTCC also include microwave modules, opto-electronics modules, meta-material antennas, automotive, military, medical and sensor packaging, pressure sensors, etc. (Cleveland et al., 1992; Gongora-Rubio et al., 2001; Baker et al., 2005; Peterson et al., 2005; Goldbach et al., 2006).

This paper presents the results of LTCC fabrication conducted with optimized parameters and experiments performed to attach 1 × 1 mm and 2 × 2 mm silicon dies (chips) to the LTCC substrates with Ag and AgPd pads using two solder alloys, namely, Sn95.8Ag3.5Cu0.7 (lead-free solder) and Sn63Pb37 (tin-lead solder). A shear test was carried out to evaluate the shear strength of the solder joints. The microstructure of the solder joints was characterized using X-ray, scanning electron microscopy (SEM) and energy dispersive X-ray (EDX) techniques.

2 Materials and methods

The LTCC substrate fabrication started with preparing four layers of green blank tapes to a 150 × 150 mm size. The tapes were attached to a frame for punching of tooling holes using a punching machine.

The punched tapes were transferred to a printing machine. Ag and AgPd pastes were printed onto the top layer of the tapes using a custom-designed screen. The printed tapes were dried for 10 min at 70°C. The tapes were removed from the frames and were stacked using an alignment fixture. The stacking was accomplished with the pattern-printed tape on top and three blank tapes beneath it.

The aligned tapes were vacuum-sealed and laminated using an isostatic laminator. Laminated tapes were cut to coupons of a size of 25 × 25 mm using a green tape cutter on a hot bed. Debinding was carried out in a box furnace at 460°C for three hours with the laminated tape resting on a setter.

After debinding, the tapes were transported to a belt furnace and were sintered using an optimized profile with a sintering ramp rate of 20°C/min, a peak temperature of 850°C and dwell time of 10 min. Measurements performed on the sintered substrates showed that the targeted dimensions were achieved with a density of 2.5 g/cm³ and with 15.4 percent shrinkage in the lateral (X and Y) directions and 25 percent shrinkage in the thickness (Z) direction.

The silicon dies to be used were fabricated from blank silicon wafers sputtered with a layer of TiW (thickness < 0.5 μm) and a Cu layer (thickness < 1 μm). An ENIG process was used to plate 2 μm thick nickel with 0.1 μm gold on the surface. The plated wafers were diced to the required sizes using a semi-automatic wafer-dicing machine.

Attachment of the dies to the LTCC substrate was carried out using the surface mount technology. Two types of solder paste were chosen for the experiment. The specifications of the solder pastes used are given in Table I. Solder paste printing stencils were fabricated with a thickness of 0.19 mm. The solder pastes were printed on the Ag and AgPd pads of the LTCC substrates, and reflow was carried out using a reflow oven.

The strength of the solder joints was measured by a die shear test carried out using an Instron tensile tester (Model-5548). The crosshead speed used for the test was 0.5 mm/min.

X-ray imaging was performed to estimate the amount of voids present in the solder joints. Cross sectioning of the solder joints on selected coupons was conducted to analyze and determine the characteristics of the SnAgCu and SnPb solder joints. SEM and EDX analyses were performed to analyze the intermetallic compounds of the solder joints. Failure analysis was also carried out using optical microscopy and SEM inspection of the sheared joints.

3 Results and discussion

Typical X-ray images of the void distribution in the SnAgCu and SnPb solder joints between the 2 × 2 mm dies and the pads of AgPd or Ag are shown in Figure 1. The analysis revealed that the average void content in the SnPb solder on the Ag pad was approximately 10 percent of the total area of the joint, while the average void content found in all other combinations was around 20 percent of the total area of the joints.

Figure 2 shows typical X-ray images of the void distribution in the SnAgCu and SnPb solder joints between the 1 × 1 mm dies and the pads of AgPd or Ag. The analysis revealed that the average void content in the solder joints on the Ag pads was less than 10 percent of the total area of the joints, while the average void content in the solder joints on the AgPd pads was around 15 percent. The reduction in void content in the joints for 1 × 1 mm dies, compared to that for 2 × 2 mm dies, could be attributed to reduction in total lateral-area of the solder joints allowing the volatile compound to escape easily without forming voids during reflow.

The shear test for 1 × 1 mm dies was not suitable using the equipment available for this study. Therefore, extensive characterization and failure analysis were conducted for the 2 × 2 mm dies, not for the 1 × 1 mm dies.

Optical stereo-zoom inspection and SEM inspection were carried out after the shear test to analyze the fracture surfaces and failure modes. The images of fracture surfaces are shown in Figure 3. The shear load to fracture the 2 × 2 mm die-solder joints and failure modes are summarized in Table II.

The shear test performed on 2 × 2 mm dies revealed that the joints made by SnAgCu solder could withstand higher shear load than the joints made by SnPb solder. The AgPd pads soldered with SnAgCu solder provided the highest joint strength and the Ag pads soldered with SnPb solder provided the lowest joint strength in Table II. Overall, AgPd pads provided higher shear strength values than Ag pads.

Figures 4 and 5 show typical graphs of shear load versus crosshead displacement obtained during the shear test on the SnAgCu and SnPb solder joints between the 2 × 2 mm dies and the pads of AgPd and Ag, respectively. The SnAgCu solder joint on the AgPd pad required higher energy to cause failure compared to the SnPb solder joint on the Ag pad, which required lower energy to cause failure, indicating poor interface strength of the joint.

The SEM images of cross sections of the solder joints show the structures of the solders and the interfaces. The solder joints and the interfaces of AgPd-SnPb, AgPd-SnAgCu and Ag-SnAgCu were diffused homogeneously as shown in Figures 6–8, and this resulted in higher shear strength values. Although the average void content in the SnPb solder on the Ag pad was less than that in other solder-pad combinations as shown in Figure 1, the SEM analysis revealed that the lower shear strength of the Ag-SnPb joints was due to wetting problems of the solder and the Ag pad, as shown in Figure 9.

Intermetallic compounds were analyzed using the electron dispersive X-ray (EDX) technique. Table III summarizes the findings of the intermetallic compounds observed using the EDX technique for the various solder-pad combinations.

These results showed that joints made by AgPd pads with SnAgCu solder provided the best results with the highest shear strength and strong interfaces in the joints. This was attributed to the good formation of the intermetallic compound at the die and solder interface and the AgSn intermetallic compound at the pad and solder interface. The worst joints were made with the combination of Ag pads and SnPb solder providing the lowest shear strength with wetting problems or poor intermetallic compounds at the interface between the Ag pad and SnPb solder. For applications where small silicon dies need to be attached to LTCC substrates, AgPd pads with SnAgCu solder can be considered. For lead-free applications, Ag pads with SnAgCu solder can be considered.

4 Conclusions

Small silicon dies were soldered onto Ag and AgPd pads of LTCC substrates using SnAgCu or SnPd solder. Extensive characterization and failure analysis were carried out on the solder joints with 2 × 2 mm dies using X-ray, die shear, SEM and EDX techniques. The results showed that joints made by AgPd pads with SnAgCu solder provided the best results with the highest shear strength having strong interfaces in the joints. However, the joints of Ag pads with SnPb solder did not provide high-shear strength. For applications where small silicon dies need to be attached to LTCC substrates, AgPd pads with SnAgCu solder can be considered. For lead-free applications, Ag pads with SnAgCu solder can be considered.

Figure 1X-ray images of void distribution in the SnAgCu and SnPb solder joints between the 2 × 2 mm dies and the pads of AgPd or Ag

Figure 2X-ray images of void distribution in the SnAgCu and SnPb solder joints between the 1 × 1 mm dies and the pads of AgPd or Ag

$ImageOptical images of fractured surfaces after the shear test on the 2 × 2 mm die-solder joints$
Figure 3Optical images of fractured surfaces after the shear test on the 2 × 2 mm die-solder joints

Figure 4Shear load versus crosshead displacement obtained during the shear test on a SnAgCu solder joint between a 2 × 2 mm die and an AgPd pad

Figure 5Shear load versus crosshead displacement obtained during the shear test on a SnPb solder joint between a 2 × 2 mm die and an Ag pad

Figure 6Cross section of a SnAgCu solder joint on an AgPd pad

Figure 7Cross section of a SnAgCu solder joint on an Ag pad

Figure 8Cross section of a SnPb solder joint on an AgPd pad

Figure 9Cross section of SnPb solder on an Ag pad

Table ISpecifications of the two types of solder paste used

$ImageShear load to fracture and failure modes of the joints$
Table IIShear load to fracture and failure modes of the joints

Table IIIIntermetallic compounds (IMCs) in the solder joints

References

Anand, A., Mui, Y.C. (2002), "Lead-free solder evaluations for ball attach process", paper presented at the 4th IEEE Electronic Packaging Technology Conference, Singapore, .