Effect of Bi Content on the Microstructure and Mechanical Performance of Sn-1Ag-0.5Cu Solder Alloy

Abstract

The purpose of this work is to explore the impact of 0.5, 1.5, 2.5 and 3.5 wt.% Bi additions on the microstructure and mechanical performance of Sn-1Ag-0.5Cu solder alloy. Scanning electron microscope (SEM) and X-ray diffraction (XRD) were utilized to examine the microstructure of the present solders. Creep measurements have been used for the preliminary assessment of mechanical properties. The steady-state creep rate, έ_st, diminished as the Bi’s concentration increased and reached 2.5 wt.%, with this trend altering above this point. Furthermore, increasing the aging or testing temperature caused the έ_st values to increment for all the investigated solders. έ_st variations with different Bi content and aging temperature were observed by examining the Sn-Ag-Cu solders’ structural evolutions. The mean value of the activation energy of all investigated solder alloys was found to be ∼52 kJ/mol. This value is appropriate to that quoted for the dislocation climb through the core diffusion as the dominant operating mechanism. The XRD findings supported the microstructure and lattice parameters variations with both aging temperatures and bismuth concentrations.

1. Introduction

Owing to environmental issues about Pb toxicity, considerable endeavors have been done to get Pb-free solders that have a low melting temperature, low cost, good wettability, and desirable mechanical characteristics [1,2,3]. Among the various Pb-free solders available, Sn-Ag-Cu (SAC) solders have been recognized as the most attractive replacement of Pb-containing solders due to their modest melting point, excellent wetting ability, and remarkable mechanical features [4,5,6]. However, several obstacles are attributed to the SAC solders with a high Ag-content, such as the large brittle Ag₃Sn development as an intermetallic compound (IMC) that degrades the mechanical properties of the solder joints [7,8]. Diminishing the silver addition can enhance the mechanical characteristics of the SAC solder joints [9,10]. However, the SAC solders with low Ag-content reveal a high melting point, worse mechanical properties, and poor wettability as compared with widely used Sn-(3.0–3.9)Ag-(0.5–0.7) Cu solders [11]. The low Ag-content SAC solders’ performance can be developed by adding second reinforcing alloying elements such as Fe, Mn, Ti, Ni, Bi, Zn, and Al [12,13]. Alloying additions of metallic elements could refine the microstructure and limit the formation brittle Ag₃Sn (IMCs) [14,15,16].

Several research papers have documented the impact of minor addition of metallic elements on SAC soldiers’ microstructure and mechanical characteristics. Zhao et al. [17] investigated the impact of 1 and 3 wt.% Bi addition on Sn-3Ag-0.5Cu (SAC305) solders’ microstructures and mechanical properties. They explored that the Bi addition increased the tensile strength but decreased the elongation of SAC305 solders. Chuang et al. [18] deduced that titanium addition into SAC305 solder refined the microstructure of eutectic areas (β-Sn + Ag₃Sn). As titanium percent exceeds 1.0 wt.%, a small amount of coarse Ti₂Sn₃ IMC exists, and the elongation decreases. A notable contribution was made by Mahdavifard et al. [19], who studied the effect of Fe and Bi on the microstructure and mechanical characteristics of Sn-1Ag-0.5Cu solder. Bi addition to SAC105-Fe solder increased the interdendritic regions of β-Sn. Also, it decreased the Ag₃Sn and Cu₆Sn₅ IMCs, which enhanced the yield strength and ultimate tensile strength but decreased the total elongation. It was concluded by Cheng et al. [20] that minor Ni addition to SAC105/Cu increased Cu₆Sn₅ layer growth, but the Cu₃Sn growth was obstructed through the subsequent aging of the solid-state. Zn’s impact was examined by Song et al. [21] on microstructure evolutions and tensile properties of SAC105 solder. They reported the enhancement of tensile properties and the reduction of the total elongation. Kanlayasiri et al. [22] investigated the influence of indium addition on the tensile strength of SAC105 solder. The addition of 3.0 wt.% In increased the tensile strength of SAC105 solder by around 79%.

The microstructure and mechanical response of solder joints change as they age isothermally. The creep behavior degrades drastically when solder joints are exposed to aging over time. Such variations in the creep behavior are attributed to the microstructure evolution that develops during aging. Although there have been some creep reports on the low Ag-content SAC105 solders in previous studies, the impact of heat treatment and Bi addition on the microstructure and creep behavior of low Ag-content SAC105 solders were quite limited. The insufficiency of the investigation into this topic inspired the present study.

2. Materials and Methods

In this study, low Ag-content SAC105–xBi (x = 0.0, 0.5, 1.5, 2.5, and 3.5 wt.%) solders were successfully obtained by permanent mold casting. Pure metals (99.99%) of Sn, Ag, Cu, and Bi were brought to produce the alloys under investigation. Melting was conducted in alumina crucibles at 623 K for 120 min in an electric resistance furnace under a protective argon atmosphere. The molten alloys were poured into mild steel molds of dimensions 150 mm × 10 mm × 10 mm in the air at room temperature (298 K). The five solder alloys’ chemical compositions were verified utilizing inductively coupled plasma atomic emission spectroscopy (ICP-AES) (

Table 1. Chemical composition of the experimental SAC105-xBi solders verified by means of inductively coupled plasma atomic emission spectroscopy (ICP-AES), wt.%.

Solder	Element Content
	Ag	Cu	Bi	Sn
SAC105	0.97	0.48	0	balance
SAC105-0.5Bi	0.94	0.46	0.49	balance
SAC105-1.5Bi	0.93	0.43	1.45	balance
SAC105-2.5Bi	0.88	0.40	2.42	balance
SAC105-3.5Bi	0.85	0.39	3.40	balance

).

The as-cast ingots were swaged and cold drawn into (i) sheets of 0.6 mm in thickness (by rolling) for microstructure investigations and (ii) wires of 0.8 mm gauge diameter and 50 mm gauge length for tensile creep measurements. For 120 min at 453 K, the samples were solution heat-treated for homogenization and quenched into iced water (273 K). They were immediately aged at various temperatures (T_a = 353, 373, 393, and 413 K) for 120 min, followed by water quenching at 273 K.

The tensile creep tests were carried out using an Instron 3360 Universal Testing Machine (Model:GT-K01, Quanzhou, China) at various testing temperatures ranging from 318 to 348 K under a certainly applied load corresponding to a stress of 19.5 MPa. The details of the tensile testing machine used in the current work are described elsewhere [23]. The precision of the temperature measurements is within the range of ±1 K.

To study SAC105–xBi samples’ microstructure, they were ground and polished according to ordinary steps for solder alloys [24]. Before the microstructure examination, a solution comprising 10 mL nitric acid, 10 mL acetic acid, and 80 mL glycerol was employed to etch the samples for about 90 s. The microstructural features of the phases present in the samples were examined by a scanning electron microscope-SEM (JEOL JSM-6360LV-Akishima, Tokyo, Japan) equipped with energy dispersive spectroscopy (EDS) (Oxfordins Inca-200, Oxford Instruments, Abingdon, UK) operating at 20 kV. Furthermore, X-ray diffraction (XRD) measurements were carried out at room temperature using a Shimadzu D6000 X-ray diffractometer (Shimadzu Corporation, Tokyo, Japan) with Ni-filtered CuK_α radiation (λ = 1.5406 Å) worked at 30 kV and 40 mA. The XRD spectra were collected in the 20–80° 2θ range at a scan rate of 2° min⁻¹ and a step size of 0.02°.

3. Results and Discussion

Creep is typically associated with time-dependent plastic deformation of materials at an elevated temperature under constant uniaxial stress [25]. Most metals and alloys display normal creep curves at temperatures above about 0.5 T_m, where T_m is the absolute melting point [26]. Typical creep curves include three parts that appear after the initial instantaneous strain (ε_o) upon loading. The creep rate decreases rapidly in the first stage of creep, known as primary creep. The creep rate is approximately constant in the second stage of creep, referred to as steady-state or secondary creep. Once the creep rate has accelerated to the fracture point, it becomes a tertiary creep [27].

Creep curves series for SAC105-xBi solders were depicted for specimens at various aging temperatures (T_a = 353, 373, 373, and 413 K) for 120 min. Curves for specimens aged at 353 and 413 K are displayed in Figure 1 as representative creep curves. Inspection of these sets of curves indicates that the studied solders’ creep rate values are reduced with Bi-content elevation. The SAC105 solder shows the lowest creep resistance among the solders, while the SAC105-2.5Bi solder introduces the highest one. Furthermore, the creep rate values are increased with increasing aging temperature. It is possible to calculate the slope of the obtained creep curves (Figure 1) as a derivative of the creep strain and creep time. With this solution, the steady-state creep rate, έ_st, values can be obtained. Figure 2 demonstrates the steady-state creep rate dependence, έ_st, on the Bi content at different aging and testing temperatures. It is worth noting that for all aging temperatures, έ_st values are diminished with Bi concentration up to 2.5 wt.% then increased above this percentage. Moreover, έ_st values are shifted monotonically towards higher values with increasing aging and/or testing temperature.

According to the binary phase diagram of Sn-Ag, Sn-Cu, and Ag-Cu systems (shown in Figure 3, Figure 4 and Figure 5, respectively), three possible phases may be established: (i) Ag reacts with Sn to form Ag₃Sn (Figure 3). (ii) Due to the reaction between Sn and Cu Cu₆Sn₅ can be formed (Figure 4). However, when the Cu concentration is high enough, Cu₃Sn will not form unless it is under elevated temperatures, which the current study did not identify. (iii) The Ag-rich phase and Cu-rich phase are established in (Figure 5). There is no reaction between Ag and Cu that results in the formation of IMCs [28,29]. Figure 6 presents a SEM micrograph that demonstrates the microstructure of the SAC105 solder after it has aged at 353 K. Interdendritic regions consist of a dot-like Ag₃Sn phase and an irregular polygon of the Cu₆Sn₅ phase, also dark-gray β-Sn dendrites, all generate the microstructure. EDS analysis of the Ag₃Sn phase verified its chemical composition, as seen in Figure 6a. A magnified view exhibited in Figure 6b showed that the Ag₃Sn phase develops dot-like morphologies. According to EDS analysis, the irregular polygon phase was identified as the Cu₆Sn₅ phase (Figure 6c). Our observations align well with those established by other researchers [19,30,31], who detected the growth of Ag₃Sn and Cu₆Sn₅ phases in the SAC solders. The Ag₃Sn and Cu₆Sn₅ phases in the β-Sn matrix possess much higher strength than the bulk Sn-material. It was reported that [32] the fatigue life of the SAC solders has been improved by three to four times over the Sn-Pb eutectic solders due to the presence of Ag₃Sn and Cu₆Sn₅ phases. It seems that the Cu₆Sn₅ and interspersed Ag₃Sn precipitates that block and pin the dislocation motion contribute to the higher fatigue resistance.

To evaluate the influence of Bi addition on the microstructure of SAC105 solder alloy, SEM was utilized. Figure 7 exhibits typical SEM images of microstructures of the SAC105-xBi (x = 0, 0.5, 1.5, 2.5, and 3.5 wt.%) solders aged at 353 K. Notably, the addition of Bi to the SAC105 solder produced significant microstructural changes, as shown in Figure 7a–e. All specimens contain a basic microstructure, which includes β-Sn, Ag₃Sn, and Cu₆Sn₅ phases. As it is seen, the addition of Bi can promote the formation of the eutectic regions inside the β-Sn matrix and reduces the large β-Sn dendrite cell generation. The reason may be that Bi particles may act as the nucleation sites for the formation of the second phase in the eutectic colony during solidification. These fine Ag₃Sn and Cu₆Sn₅ phases perform a predominant role in strengthening, reflecting the reduction of the έ_st values through Bi-content increment (see Figure 2). Our observations are well consistent with the earlier work of other investigators [17,19,33].

When 3.5Bi was added into the SAC105 solders, the steady-state creep rate values were found to increase abnormally (see Figure 2). These unusual observations may be elucidated as follows. The growth of larger particles at the smaller particles’ expense with a higher interfacial enthalpy is known as Ostwald ripening. The process’s driving force diminishes the total interface area between the dispersed particles and the matrix. Depending on the particle size variation, the free energy of a particle alters. This alteration leads to solubility variation at the particle/matrix interface (Gibbs–Thomson effect). The result is that the matrix surrounding the larger particles contains less solute than the matrix area around smaller particles, which sets up solute concentration gradients in the direction of the smaller particles to the larger particles. Consequently, the smaller particles shrink, and the larger particles grow, even while the particles’ overall volume fraction is still invariable [34,35]. The 3.5Bi addition elevates the size and spacing between Ag₃Sn and Cu₆Sn₅ precipitates, which are still smaller for the Bi-free solders (Figure 7e). These observations may be referred to the Bi rich phase accumulation in the β-Sn matrix. Figure 8 exhibiting the microstructure of the SAC105 solder containing 3.5Bi. A magnified view disclosed in Figure 8a showing the distribution of the Bi-rich phase within the β-Sn matrix. The Bi-rich phase’s chemical composition was also characterized by EDS analysis, as shown in Figure 8b. The microstructure revealed that the Bi contained in the SAC105-3.5Bi solder participated in the “enrichment” in the β-Sn dendrite cells. Our results are compatible with the previous findings of Zhao et al. [17], who concluded that fine particles of Bi rich phase are precipitated when the Bi content is more than 3 wt.%. This is in line with the previous work of Huang et al. [33], who concluded that it is not possible to observe fine Bi particles in the Sn–Ag-based solders containing 1 and 2 wt.% Bi. Since the Van der Waals’ forces cause Bi particles to become entangled with each other, bismuth concentrations reach approximately 3.5% wt.%; this is around the time that the Ag₃Sn and Cu₆Sn₅ phases start to precipitate, resulting in lower solder reliability. This is consistent with the previously reported findings by Sayyadi and Naffakh-Moosavy [7]. They found that the SAC-Bi solder alloys’ mechanical properties were reduced by increasing the Bi concentration from 2.5 to 5 wt.%.

The results shown in Figure 2 demonstrate that the έ_st values are increased with increasing aging temperature, T_a, from 353 to 413 K for all five investigated solders. This might be ascribed to the coarsening and coalescence of Ag₃Sn and Cu₆Sn₅ precipitates. Such coarsening and coalescence increase precipitates size accompanied by the reduction of their numbers, thus the matrix gets clean, and the dislocation barriers will widely be reduced, leading to higher creep rates. SEM investigations strongly support the above interpretation based on coarsening and coalescence of Ag₃Sn and Cu₆Sn₅ precipitates. Typical SEM images of the SAC105 solders reinforced with various Bi additions aged at 413 K are presented in Figure 9. It has been reported [33] that the aging process of the SAC solders results in a growth in IMCs size, accompanied by a reduction in their numbers. When the dislocation moves, it has two possible courses of action. It can either bend around and bypass the precipitates or cut through them. The first alternative is possible only when the slip plane is continuous from the matrix through the precipitates. When the stress to move a dislocation in precipitates is comparable to that in the β-Sn matrix. When there is an interface or a change in orientation, it is impossible to cut a precipice. Under such an instance, dislocations have to bend around them and bypass. In this instance, the mechanism works similarly to a Frank–Reed source. For Ag₃Sn and Cu₆Sn₅ precipitates, which may be regarded as incoherent precipitates of large size and a small number, dislocations will bow and leave a loop of stress field around the particle [36]. The yield stress, σ_o, of the solder can be determined from the relationship [37]:

$${\sigma }_o= \sqrt{\frac{Gb\sigma }{\pi \left(1-v\right)L }}$$

(1)

where G is the solder shear elastic modulus, b is the Burgers vector, σ is the stress at the particle surface, ν is the Poisson’s ratio, and L is the inter-particle spacing between the two particles. According to Equation (1), as the inter-particle spacing, L, increases during aging, the yield stress, σ₀, will decline. Therefore, the steady-state creep rate values, έ_st, are increased with the coarsening and coalescing of Ag₃Sn and Cu₆Sn₅ precipitates. Our experimental results are consistent with those reported by Wu al. [38], who pointed out that the thickness of Ag₃Sn and Cu₆Sn₅ precipitates increases as the aging temperature increases. Sona and Prabhu [28] reported that Ag₃Sn and Cu₆Sn₅ IMCs grow during aging, and the strength of the SAC solder alloys decreases with the IMCs size increase. Coarsening of the microstructure may be the reason for this strength reduction after aging in the solder alloys.

A further representation for the influence of testing temperature, T_t, on the steady-state creep rate values, έ_st, at different aging temperatures is displayed in Figure 10. Based on dislocation density and average dislocation spacing, the increase in the steady-state creep rate, έ_st, values with increasing testing temperature, T_t, for all investigated solders can be clarified. A decrease in the dislocation density occurs as the testing temperature increases, providing an increase in the average distance of dislocations. A moving dislocation can then bypass the precipitates by moving in clean pieces between the precipitated and coarsened particles. Accordingly, the steady-state creep rate, έ_st, values rise with testing temperature increment. The other data, which has been described in other literature [39,40], is very similar to the results found in this study.

Estimating the creep process’s apparent activation energy value is attempted by applying the Arrhenius-type relation that relates the creep rate, έ_st, to the testing temperature, T_t, as follows [41]:

έ_st = A σⁿ exp(−Q/RT_t)

(2)

where A is a structure-dependent constant, σ is the applied stress, n is the stress exponent, Q is the creep activation energy, and R is the universal gas constant. The slopes of the straight lines relating ln έ_st against (10³/T_t) were estimated, giving the creep activation energy, Q, for all investigated solders aged at various temperatures (Figure 11). The mean value of activation energy was found to be ∼52 kJ/mol. This activation energy value is consistent with that quoted for dislocation climb through core diffusion as the dominant operating mechanism [42,43].

X-ray diffraction (XRD) analysis was performed to confirm the tetragonal β-Sn phase existence as a matrix (JCPDS 04-0673). Figure 12a,b shows the XRD charts for the investigated solder alloys aged at 353 K and 413 K for 120 min as a representative example. The Cu₆Sn₅ (JCPDS card no. 65-2303) and Ag₃Sn (JCPDS card no. 44-1300) IMCs appeared as the main phases overall the solder patterns. Bismuth was observed as a separated phase (JCPDS 89-2387) for all Bi-containing alloys because it did not form IMCs in the interfacial reactions. Olofinjana et al. [44] reported Bi peaks as a primary phase and their development in a series of SAC-Bi solders. Kim et al. [45] illustrated the existence of both the Cu₆Sn₅ and Ag₃Sn phases in Sn-Ag-Cu alloys. Figure 12 reveals that the increment of IMCs peaks intensities with the elevation of Bi content up to 2.5 wt.%. The higher bismuth concentration (3.5 wt.%) declares the reduction in the Cu₆Sn₅ and Ag₃Sn intensities but increases Bi diffraction peaks intensities at both aging temperatures. As a result, the steady-state creep rate values reach their maxima for SAC-2.5 Bi alloy at all aging temperatures, which can be rendered to Bi precipitates formed in the matrix (up to 2.5 wt.%), decreases the size of the IMCs, and causes higher strength through the prevention of the dislocations movement, delay plasticity, and decrease ductility [7,46].

The estimation of the lattice parameters, a and c for the β-Sn rich phase with the tetragonal crystal structure was carried out by substitution d-spacing values for two diffraction planes ((200) and (101)) in the following equation [45]:

$$\frac{1}{d^2}=\frac{h^2+k^2}{a^2}+\frac{l^2}{c^2}$$

(3)

where h, k, and l represent the Miller indices for 200 and 101 diffraction planes.

Table 2. The calculated crystalline lattice parameters a, c, c/a ratio, and the unit cell volume V for β-Sn matrix at different aging temperatures and various bismuth concentrations.

Lattice Parameter	Aging Temperature (K)	SAC105	Error %	SAC105-0.5Bi	Error %	SAC105-1.5Bi	Error %	SAC105-2.5Bi	Error %	SAC105-3.5Bi	Error %
a (Å)	353	5.9153	1.44	5.93	1.697	5.822	0.139	5.8235	0.127	5.880	0.840
	373	5.8453	0.246	5.83	0.017	5.815	0.257	5.7576	1.258	5.867	0.618
	393	5.8579	0.462	5.8944	1.088	5.776	0.928	5.776	0.943	5.841	0.171
	413	5.8692	0.655	5.7796	0.88	5.730	1.716	5.800	0.530	5.850	0.337
c (Å)	353	3.2355	1.682	3.1646	0.545	3.067	3.597	3.1689	0.409	3.160	0.691
	373	3.1984	0.516	3.1853	0.104	3.128	1.685	3.1918	0.309	3.099	2.599
	393	3.2058	0.748	3.0818	3.146	3.184	0.080	3.2522	2.206	3.136	1.431
	413	3.1909	0.280	3.1506	0.986	3.180	0.042	3.2129	0.973	3.126	1.734
c/a	353	0.5469	0.233	0.5336	2.205	0.526	3.462	0.5441	0.281	0.537	1.518
	373	0.5471	0.270	0.5463	0.122	0.537	1.430	0.5543	1.588	0.528	3.198
	393	0.5472	0.285	0.5228	4.188	0.551	1.018	0.5630	3.180	0.536	1.599
	413	0.5436	0.372	0.5451	0.106	0.555	1.704	0.5539	1.512	0.534	2.065
V (Å)3	353	113.21	4.645	111.28	2.860	104.006	3.866	107.472	0.662	109.256	0.986
	373	109.28	1.013	108.26	0.070	105.818	2.191	105.81	2.197	106.685	1.389
	393	110.01	1.683	107.07	1.026	106.275	1.768	108.50	0.288	107.007	1.092
	413	109.91	1.599	105.24	2.721	104.463	3.443	108.087	0.093	107.032	1.069

explores that under both aging conditions and Bi-content, the lattice parameters, c/a ratio, and the unit cell volume V of the b-Sn phase slightly change for the investigated alloys. Calculation errors were recorded as separated columns in that Table. The findings reveal that the presented specimens’ lattice parameters are close to those previously determined for the Sn-lattice (JCPDS 04-0673). SAC-2.5Bi alloy has relatively low (a) values and has relatively high values of the other parameters (e.g., c and c/a ratios). The previous study [47] explored that Sn-phase’s lattice parameters changed with Bi-content, but they still close to binary alloy lattice parameters. Once the Bi atoms were dispersed into the matrix, they stopped the dislocations’ mobility. This change improved the strength and hardness simultaneously, as explained by Mahdavifard et al. [19].

Consequently, the unit cell volume was determined through the equation [45]:

$$V=a^{2}c$$

(4)

The recorded V values (Table 2) are near the standard value (108.18 ų) to β-Sn phase. These values for the unit cell volume agree with those in the literature [19,47].

4. Conclusions

The creep behavior of Sn-1Ag-0.5Cu solders has been affected by Bi addition and aging temperature at different testing temperatures through the current study. In light of these findings, the following conclusions can be drawn:

(1)

The steady-state creep rate values were decreased continuously as Bi concentration was elevating up to 2.5 wt.%; above this weight ratio, the trend runs in the opposite direction.

(2)

The steady-state creep rate values for all Bi-containing solders increased as the aging and/or testing temperatures raised.

(3)

The steady-state creep rate alterations with Bi ratio and aging temperature were attributed to the formation, growth, coarsening, and coalescence of Ag3Sn and Cu₆Sn₅ precipitates.

(4)

According to the activation energy mean value, the dislocation climb through the core diffusion was the predominant creep mechanism.

(5)

The unit cell volume, the lattice constants a and c of the Sn-matrix vary depending on the change in the bismuth content and aging temperature variations