Improved Reliability and Mechanical Performance of Ag Microalloyed Sn58Bi Solder Alloys

: Ag microalloyed Sn58Bi has been investigated in this study as a Pb-free solder candidate to be used in modern electronics industry in order to cope with the increasing demands for low temperature soldering. Microstructural and mechanical properties of the eutectic Sn58Bi and microalloyed Sn57.6Bi0.4Ag solder alloys were compared. With the addition of Ag microalloy, the tensile strength was improved, and this was attributed to a combination of microstructure reﬁnement and an Ag 3 Sn precipitation hardening mechanism. However, ductility was slightly deteriorated due to the brittle nature of the Ag 3 Sn intermetallic compounds (IMCs). Additionally, a board level reliability study of Ag microalloyed Sn58Bi solder joints produced utilizing a surface-mount technology (SMT) process, were assessed under accelerated temperature cycling (ATC) conditions. Results revealed that microalloyed Sn57.6Bi0.4Ag had a higher characteristic lifetime with a narrower failure distribution. This enhanced reliability corresponds with improved bulk mechanical properties. It is postulated that Ag 3 Sn IMCs are located at the Sn–Bi phase boundaries and suppress the solder microstructure from coarsening during the temperature cycling, hereby extending the time to failure.


Introduction
The implementation of legislation entitled "Restriction on the use of certain Hazardous Substances (RoHS)" in the European Union in July 2006 marked the beginning of the phasing out of conventional Sn37Pb (expressing 37 wt.% Pb and the balance is Sn concentration, similarly hereafter) eutectic solder in the electronic industry. RoHS has been since adopted in countries and regions with dense electronics manufacturing: Korea, Japan, China, and a range of US states [1]. In response, most major electronics manufacturers have stepped up their search for alternatives to Pb-containing (Sn37Pb) solders (i.e., Pb-free solders). Sn-Ag, Sn-Cu, and Sn-Ag-Cu (SAC) Pb-free eutectic or near eutectic systems have emerged as the front runners in the replacement of Sn37Pb solder [2]. Among them, near-eutectic SAC alloys have established themselves as the interconnect materials of choice for the electronic packaging industry [3][4][5][6][7]. Various SAC alloy systems have been proposed by Japanese (SAC305, short for Sn3.0Ag0.5Cu), EU (SAC387, short for Sn3.8Ag0.7Cu), and US (SAC396, short for Sn3.9Ag0.6Cu) consortiums [8], with the SAC305 alloy widely used by the industry as the most promising candidate for reliable Pb-free solder due to its reliability under thermal cycling [9,10].
Comparing conventional Sn37Pb eutectic solder (183 • C), the higher melting temperature of SAC305 (217-221 • C) limits its application when facing miniaturization challenges associated with emerging ultra-mobile computing, wearable devices, and the Internet of Things (IoT) markets [11]. Therefore, further studies have been carried out globally by industrial as well as academic consortiums on new Pb-free alternatives [12][13][14][15]. Sn58Bi solder is gaining considerable attention because of its low melting temperature of 138 • C [16,17], with other advantages of Sn-Bi-based solder associated with Tensile tests were performed at room temperature at a strain rate of 0.75 mm/min (5 × 10 −4 s −1 ) on an H25KS tensile tester(Tinius Olsen, Horsham, PA, USA). A ZHV30 Vickers hardness tester (ZwickRoell, Ulm, Germany) was used to determine the hardness of as-cast solder alloys. For each composition, five tests were carried out to achieve a statistical result. A phase identification of the various solder was carried out using an X'pert X-ray diffractometer (XRD, Phillips, Amsterdam, Netherlands) operated at 40 kV, and Cu-Kα radiation was used at diffraction angles (2θ) from 10° to 90° with a scanning speed of 1.2°/min. Solder pastes were produced utilizing a rosin-based flux. Commercial Sn37Pb solder paste was used in this study as a control. As shown in Figure 2, dummy components comprising 16 of 1206 chip resistors were soldered with these pastes on an FR4 printed circuit board (PCB) with a Cu organic solderability preservative (OSP) surface finish. The PCB was 1.55 mm thick and consisted of two layers. A standard right angle 37-pin D-type male connector was used to connect the PCB to an external data logger. A WT-180/40/5 environmental test chamber (Weiss Technik, Heuchelheim, Germany) was used to generate thermal cycling for all test vehicles. A 0-100 °C cycle was imposed, which had a ramp rate of 6.7 °C/min on heating and 4 °C/min on cooling. The dwell times were 10 min at both the high and low extremes of the cycle. During ATC, the solder joints were monitored continuously with a 128/256 STD event detector (Analysis Tech, Wakefield, MA, USA) set at a resistance limit of 1000 Ω. Data were recorded and stored using the WinDatalog software (V3.4.0, Analysis Tech, Wakefield, MA, USA). In accordance with the IPC-9701A industry test guideline, a spike of 1000 Ω for 0.2 ms followed by nine additional events within 10% of the cycles to the initial event was marked as a failure [38]. Tensile tests were performed at room temperature at a strain rate of 0.75 mm/min (5 × 10 −4 s −1 ) on an H25KS tensile tester(Tinius Olsen, Horsham, PA, USA). A ZHV30 Vickers hardness tester (ZwickRoell, Ulm, Germany) was used to determine the hardness of as-cast solder alloys. For each composition, five tests were carried out to achieve a statistical result. A phase identification of the various solder was carried out using an X'pert X-ray diffractometer (XRD, Phillips, Amsterdam, Netherlands) operated at 40 kV, and Cu-Kα radiation was used at diffraction angles (2θ) from 10 • to 90 • with a scanning speed of 1.2 • /min.

Microstructure
Solder pastes were produced utilizing a rosin-based flux. Commercial Sn37Pb solder paste was used in this study as a control. As shown in Figure 2, dummy components comprising 16 of 1206 chip resistors were soldered with these pastes on an FR4 printed circuit board (PCB) with a Cu organic solderability preservative (OSP) surface finish. The PCB was 1.55 mm thick and consisted of two layers. A standard right angle 37-pin D-type male connector was used to connect the PCB to an external data logger.  Tensile tests were performed at room temperature at a strain rate of 0.75 mm/min (5 × 10 −4 s −1 ) on an H25KS tensile tester(Tinius Olsen, Horsham, PA, USA). A ZHV30 Vickers hardness tester (ZwickRoell, Ulm, Germany) was used to determine the hardness of as-cast solder alloys. For each composition, five tests were carried out to achieve a statistical result. A phase identification of the various solder was carried out using an X'pert X-ray diffractometer (XRD, Phillips, Amsterdam, Netherlands) operated at 40 kV, and Cu-Kα radiation was used at diffraction angles (2θ) from 10° to 90° with a scanning speed of 1.2°/min. Solder pastes were produced utilizing a rosin-based flux. Commercial Sn37Pb solder paste was used in this study as a control. As shown in Figure 2, dummy components comprising 16 of 1206 chip resistors were soldered with these pastes on an FR4 printed circuit board (PCB) with a Cu organic solderability preservative (OSP) surface finish. The PCB was 1.55 mm thick and consisted of two layers. A standard right angle 37-pin D-type male connector was used to connect the PCB to an external data logger. A WT-180/40/5 environmental test chamber (Weiss Technik, Heuchelheim, Germany) was used to generate thermal cycling for all test vehicles. A 0-100 °C cycle was imposed, which had a ramp rate of 6.7 °C/min on heating and 4 °C/min on cooling. The dwell times were 10 min at both the high and low extremes of the cycle. During ATC, the solder joints were monitored continuously with a 128/256 STD event detector (Analysis Tech, Wakefield, MA, USA) set at a resistance limit of 1000 Ω. Data were recorded and stored using the WinDatalog software (V3.4.0, Analysis Tech, Wakefield, MA, USA). In accordance with the IPC-9701A industry test guideline, a spike of 1000 Ω for 0.2 ms followed by nine additional events within 10% of the cycles to the initial event was marked as a failure [38]. A WT-180/40/5 environmental test chamber (Weiss Technik, Heuchelheim, Germany) was used to generate thermal cycling for all test vehicles. A 0-100 • C cycle was imposed, which had a ramp rate of 6.7 • C/min on heating and 4 • C/min on cooling. The dwell times were 10 min at both the high and low extremes of the cycle. During ATC, the solder joints were monitored continuously with a 128/256 STD event detector (Analysis Tech, Wakefield, MA, USA) set at a resistance limit of 1000 Ω. Data were recorded and stored using the WinDatalog software (V3.4.0, Analysis Tech, Wakefield, MA, USA). In accordance with the IPC-9701A industry test guideline, a spike of 1000 Ω for 0.2 ms followed by nine additional events within 10% of the cycles to the initial event was marked as a failure [38].

Microstructure
The microstructure of as-solidified Sn58Bi and Sn57.6B i0.4Ag solder alloys are shown in Figures 3  and 4. The microstructure of Sn58Bi alloy is made up of two phases, Sn phase (black area) and Bi phase (white area). This has been identified in XRD patterns shown in Figure 5. As the content of Ag microalloy was low (0.4 wt.%), the XRD patterns of the two solder alloys were similar. The microstructure of as-solidified Sn58Bi and Sn57.6B i0.4Ag solder alloys are shown in Figures 3 and 4. The microstructure of Sn58Bi alloy is made up of two phases, Sn phase (black area) and Bi phase (white area). This has been identified in XRD patterns shown in Figure 5. As the content of Ag microalloy was low (0.4 wt.%), the XRD patterns of the two solder alloys were similar.   Sn57.6Bi0.4Ag showed slightly more refined microstructure than Sn58Bi, and this was attributed to microalloying of Ag. In the microstructure of Sn57.6Bi0.4Ag, an extra rod-like phase The microstructure of as-solidified Sn58Bi and Sn57.6B i0.4Ag solder alloys are shown in Figures 3 and 4. The microstructure of Sn58Bi alloy is made up of two phases, Sn phase (black area) and Bi phase (white area). This has been identified in XRD patterns shown in Figure 5. As the content of Ag microalloy was low (0.4 wt.%), the XRD patterns of the two solder alloys were similar.   Sn57.6Bi0.4Ag showed slightly more refined microstructure than Sn58Bi, and this was attributed to microalloying of Ag. In the microstructure of Sn57.6Bi0.4Ag, an extra rod-like phase The microstructure of as-solidified Sn58Bi and Sn57.6B i0.4Ag solder alloys are shown in Figures 3 and 4. The microstructure of Sn58Bi alloy is made up of two phases, Sn phase (black area) and Bi phase (white area). This has been identified in XRD patterns shown in Figure 5. As the content of Ag microalloy was low (0.4 wt.%), the XRD patterns of the two solder alloys were similar.   Sn57.6Bi0.4Ag showed slightly more refined microstructure than Sn58Bi, and this was attributed to microalloying of Ag. In the microstructure of Sn57.6Bi0.4Ag, an extra rod-like phase Sn57.6Bi0.4Ag showed slightly more refined microstructure than Sn58Bi, and this was attributed to microalloying of Ag. In the microstructure of Sn57.6Bi0.4Ag, an extra rod-like phase could be observed, as shown in Figure 6. According to the phase diagram of Sn-Bi-Ag, as shown in Figure 7, Metals 2019, 9, 462 5 of 10 this phase was confirmed to be Ag 3 Sn IMCs precipitated from the Sn matrix. However, an Ag 3 Sn phase was not detected in XRD, presumably due to its limited content.
Metals 2019, 9, x FOR PEER REVIEW 5 of 10 could be observed, as shown in Figure 6. According to the phase diagram of Sn-Bi-Ag, as shown in Figure 7, this phase was confirmed to be Ag3Sn IMCs precipitated from the Sn matrix. However, an Ag3Sn phase was not detected in XRD, presumably due to its limited content.
(a) (b)   Figure 8 shows the typical tensile stress-strain curves of Sn58Bi-based solder alloys. UTS, YS (represented by 0.2% proof stress), elongation, Young's modulus, and microhardness of each alloy are listed in Table 1 as well as presented in Figure 9. It can be clearly observed that UTS, YS, and Young's modulus of Sn58Bi-based solder alloy were all slightly increased by microalloying 0.4 wt.% Ag. UTS of Sn57.6Bi0.4Ag alloy was 58.7 MPa, which was 6.7% higher than that of Sn37Pb eutectic alloy (55 MPa) [39] and 2.6% higher than that of Sn58Bi (57.2 MPa). It is worth noting that the YS of Sn58Bi solder alloy was improved by 1.8%, to 46 MPa, with the addition of 0.4 wt.% Ag. Also, the Young's modulus was raised by 14.7% as compared with Sn58Bi. While the tensile strength increased, the Ag microalloy had a negligible effect on the microhardness of Sn58Bi. The elongation, however, was decreased by 10.6% with Ag addition, compared with that of original Sn58Bi. could be observed, as shown in Figure 6. According to the phase diagram of Sn-Bi-Ag, as shown in Figure 7, this phase was confirmed to be Ag3Sn IMCs precipitated from the Sn matrix. However, an Ag3Sn phase was not detected in XRD, presumably due to its limited content.

Mechanics
(a) (b)   Figure 8 shows the typical tensile stress-strain curves of Sn58Bi-based solder alloys. UTS, YS (represented by 0.2% proof stress), elongation, Young's modulus, and microhardness of each alloy are listed in Table 1 as well as presented in Figure 9. It can be clearly observed that UTS, YS, and Young's modulus of Sn58Bi-based solder alloy were all slightly increased by microalloying 0.4 wt.% Ag. UTS of Sn57.6Bi0.4Ag alloy was 58.7 MPa, which was 6.7% higher than that of Sn37Pb eutectic alloy (55 MPa) [39] and 2.6% higher than that of Sn58Bi (57.2 MPa). It is worth noting that the YS of Sn58Bi solder alloy was improved by 1.8%, to 46 MPa, with the addition of 0.4 wt.% Ag. Also, the Young's modulus was raised by 14.7% as compared with Sn58Bi. While the tensile strength increased, the Ag microalloy had a negligible effect on the microhardness of Sn58Bi. The elongation, however, was decreased by 10.6% with Ag addition, compared with that of original Sn58Bi.   Table 1 as well as presented in Figure 9. It can be clearly observed that UTS, YS, and Young's modulus of Sn58Bi-based solder alloy were all slightly increased by microalloying 0.4 wt.% Ag. UTS of Sn57.6Bi0.4Ag alloy was 58.7 MPa, which was 6.7% higher than that of Sn37Pb eutectic alloy (55 MPa) [39] and 2.6% higher than that of Sn58Bi (57.2 MPa). It is worth noting that the YS of Sn58Bi solder alloy was improved by 1.8%, to 46 MPa, with the addition of 0.4 wt.% Ag. Also, the Young's modulus was raised by 14.7% as compared with Sn58Bi. While the tensile strength increased, the Ag microalloy had a negligible effect on the microhardness of Sn58Bi. The elongation, however, was decreased by 10.6% with Ag addition, compared with that of original Sn58Bi.

Reliability
Reliability data was fit to a two-parameter Weibull distribution: where F(t) is the unreliability rate, t is the time (number of cycles), η is the scale parameter (or characteristic life), and β is the shape parameter (or slope). The two-parameter Weibull distribution is shown in Figure 10. A summary of test results is given in Table 2. N63.2, the number of cycles to 63.2% failures [4], was used to express the characteristic life in this study.

Reliability
Reliability data was fit to a two-parameter Weibull distribution: where F(t) is the unreliability rate, t is the time (number of cycles), η is the scale parameter (or characteristic life), and β is the shape parameter (or slope). The two-parameter Weibull distribution is shown in Figure 10. A summary of test results is given in Table 2. N63.2, the number of cycles to 63.2% failures [4], was used to express the characteristic life in this study.

Reliability
Reliability data was fit to a two-parameter Weibull distribution: where F(t) is the unreliability rate, t is the time (number of cycles), η is the scale parameter (or characteristic life), and β is the shape parameter (or slope). The two-parameter Weibull distribution is shown in Figure 10. A summary of test results is given in Table 2. N63.2, the number of cycles to 63.2% failures [4], was used to express the characteristic life in this study.  Both Sn58Bi and Sn57.6Bi0.4Ag solders showed an approximately doubled characteristic lifetime compared with conventional Sn37Pb solder. Upon microalloying Ag, lifetime of Sn58Bi was marginally increased. Their lifetimes were ranked as follows: Sn57.6Bi0.4Ag > Sn58Bi >> Sn37Pb. Compared with Sn37Pb, Sn58Bi and Sn57.6Bi0.4Ag showed much higher shape parameters, indicating solder joints failed within a relatively short time span.
The simultaneous combination of these two mechanisms, microstructure refinement and precipitation hardening, improves the mechanical properties of Sn57.6Bi0.4Ag, and this manifests itself in an enhanced reliability performance. The improved reliability may also result from an initial finer microstructure, which coarsened at a slower rate under the stresses imposed by the thermal cycling. The Ag3Sn intermetallic particle as a result of microalloying seems to have a dispersion-strengthening effect that enhances the solder joints too. Similar effects of Ag microalloying on solder reliability have been reported by Collins et al. [10]. In addition, it is postulated that during thermal cycling, Ag3Sn intermetallic compounds are located at the Sn-Bi phase boundaries, and they suppress the solder microstructure from coarsening through a grain  Both Sn58Bi and Sn57.6Bi0.4Ag solders showed an approximately doubled characteristic lifetime compared with conventional Sn37Pb solder. Upon microalloying Ag, lifetime of Sn58Bi was marginally increased. Their lifetimes were ranked as follows: Sn57.6Bi0.4Ag > Sn58Bi >> Sn37Pb. Compared with Sn37Pb, Sn58Bi and Sn57.6Bi0.4Ag showed much higher shape parameters, indicating solder joints failed within a relatively short time span.
The simultaneous combination of these two mechanisms, microstructure refinement and precipitation hardening, improves the mechanical properties of Sn57.6Bi0.4Ag, and this manifests itself in an enhanced reliability performance. The improved reliability may also result from an initial finer microstructure, which coarsened at a slower rate under the stresses imposed by the thermal cycling.
The Ag 3 Sn intermetallic particle as a result of microalloying seems to have a dispersion-strengthening effect that enhances the solder joints too. Similar effects of Ag microalloying on solder reliability have been reported by Collins et al. [10]. In addition, it is postulated that during thermal cycling, Ag 3 Sn intermetallic compounds are located at the Sn-Bi phase boundaries, and they suppress the solder microstructure from coarsening through a grain boundary pinning effect, therefore extending the time to failure for these alloys.

Conclusions
It is concluded from this study that by adding small amount of Ag (0.4 wt.%) into Sn58Bi solder alloy: 1.
Mechanical properties, including tensile strength, yield strength, and Young's modulus are improved as a result of the combination of microstructural refinement and precipitation hardening.

2.
Ductility is deteriorated as a result of the formation of brittle Ag 3 Sn IMCs.

3.
Board level reliability of solder joints is enhanced during ATC testing. Ag 3 Sn IMCs are assumed to be located at the Sn-Bi phase boundaries, and this leads to a suppression of coarsening in the solder microstructure.