1 Introduction

Due to the high thermal conductivity, high electron drift velocity, and large critical electric breakdown strength, wide band gap semiconductors (SiC and GaN) have been used in many electronic applications with high operating temperatures [1, 2]. Especially, they could work stably at temperatures up to 600 °C [3, 4], which requires the solder joint to possess a high melting point. One of the effective technology is the Transient Liquid Phase (TLP) bonding, which could achieve high-temperature soldering joints [5, 6]. It could be used in forming Cu–Sn–Cu joints [7, 8], as Sn could be completely exhausted and transformed into Cu-Sn intermetallic (Cu6Sn5 is 415 °C and Cu3Sn is 676 °C) [9]. It has been well investigated for the microstructure evolutions, morphologies, and growth grains of full Cu-Sn IMCs in the soldering process with different reflow durations [10, 11]. However, most works were mainly conducted at 200–350 °C, which cannot meet the demands of servicing at higher temperatures. It is necessary to investigate the performances for the Cu–Sn IMC joints subjected to the temperature higher than 350 °C. For instance, the Cu41Sn11 phase and two-phase eutectoid microstructures comprising of Cu41Sn11 and α(Cu) [Cu41Sn11&α(Cu)] were generated at the middle of the joint when the full Cu3Sn phase in the soldering joint was aged at 350–550 °C [12,13,14]. And, the Cu20Sn6 phase and two-phase eutectoid microstructures comprising of Cu20Sn6 and α(Cu) [Cu20Sn6&α(Cu)] were transited from full Cu3Sn phase when the soldering joints were aged during 590–640 °C [15]. Although there are some works have already investigated the microstructure evolution and mechanical properties of the full IMCs aged from full Cu3Sn joints at high temperatures, it is still essential to systematically study the growth and mechanical properties for IMCs in the soldering joints subjected to high temperatures.

Intermetallic compounds that are derived from the solder and the substrate could greatly affect the reliability of the solder joint. [16, 17]. Li adopted TLP soldering processes to investigate the growth kinetics of Cu/Sn/Cu solder, which activation energy was 84.59 ± 25.84 kJ/mol, and the growth process was volume diffusion-controlled for the formed Cu3Sn. [9]. In the solid–solid interfacial interaction, the thickness of the formed IMC is always proportional to the square root of the aging time and the growth of that is recognized as volume diffusion-controlled [18,19,20]. Hence, it is necessary to understand how the activation energies control the growth kinetics of the IMCs, especially for that formed between Cu3Sn phase and the Cu substrate in the high temperature aging process.

Moreover, with the trend of miniaturization and higher-density integration for the semiconductor device, the soldering joints consisting of full IMCs contain a small number of grains, or even only one grain. The investigation for the grain orientations of the IMCs in the joint could contribute to understanding the variations of mechanical properties to control the process of high-temperature phase formation. Previous studies showed that Cu6Sn5 grains had the preferred orientation of (0001)Cu6Sn5 on polycrystalline Cu and single crystal Cu substrates [21, 22]. In addition, various studies had focused on the orientation relationship between Cu3Sn and Cu. Shang et al. [23, 24] found that the reflowed Cu3Sn layer on single crystalline Cu exhibited an orientation relationship: (0–25)Cu3Sn//(042)Cu, [100]Cu3Sn//[100]Cu conducting the comprehensive transmission electron microscopy (TEM) study. Moreover, the preferred orientation of the Cu3Sn was [100], which was parallel to the Cu substrate regardless of its orientation [25]. However, there is lack of sufficient research to characterize the grain orientation for high-temperature IMCs, which is beneficial to understand the microstructure evolution and growth mechanism.

Intermetallic compounds developed between the solder and the substrate provide the mechanical integrity of the solder joint and enhance the reliability of electronic packages. Nanoindentation has been applied in analyzing the hardness of the IMCs joints, due to small-length of material scale and inexpensive experimental procedures [26, 27]. Micromechanical hardness had been investigated by many researchers with nanoindentation [28,29,30,31]. The average value of hardness for Cu6Sn5 and Cu3Sn were 6.65 ± 0.25 and 6.10 ± 0.40 GPa, respectively. Besides, the corresponding Young’s modulus for both phases were 114.30 GPa and 125.50 GPa, respectively. Furthermore, the shear strengths and fracture analysis for the Cu6Sn5 and Cu3Sn reflowed or aged from various processing technology had been investigated [32,33,34,35,36]. Those results demonstrated that the range of shear strength for IMCs related to the Cu-Sn solders was in the 40–50 MPa [35, 36]. In addition, excessive growth for IMCs affected the fracture mechanism for the soldering joint, which specifically transited from ductile fracture to brittle fractures [33, 35]. Nevertheless, these mechanical performances for the soldering joints were the results after reflow in temperature range of 200–350 °C, which may ignore the properties of Cu–Sn joints to be serviced at higher temperatures.

Herein, the Cu20Sn6 phase was formed through original Cu3Sn IMC in the soldering joints subjected to the high temperatures of 600, 620, and 640 °C in a short time. The growth kinetics, grain orientation of the Cu20Sn6 phase, and variations for mechanical properties for the soldering joints consisting of IMCs were systematically investigated to control the growth of the Cu20Sn6 phase and gain knowledge of mechanical performance for soldering joints consisting of IMCs during the high temperature.

2 Experimental procedure

2.1 Sample preparation

The pure Cu plates (99.99 wt%, 2 × 2 × 1 mm3) were used as the substrate. The pure Sn foils (99.99 wt%) were 25 μm thick as the solder layer. Before soldering, the Cu plates were in turn grounded and polished with abrasive papers and 0.5 μm diamond polishing paste and washed in acetone with ultrasonic vibration. And the Sn foil was also washed in acetone solution with an ultrasonic bath for 5 s.

After cleaning, Cu plates and Sn foils were placed in a self-designed fixture to build a Cu/Sn/Cu sandwich structure. As shown in Fig. 1, the Cu/Sn/Cu structure was reflowed at 320 °C for 24 h with the soldering pressure of 1 N in Ar gas flow condition. The detailed parameters for preparing uniform full Cu3Sn can be drawn from our previous work [37, 38]. As shown in Table 1, the samples consisting of full Cu3Sn IMCs were aged notably at 600, 620, and 640 °C with various aging times.

Fig. 1
figure 1

Diagram of Cu/Sn/Cu sandwich structure and applied pressure

Full size image
Table 1 The parameters for the full Cu3Sn joints aged at high temperatures
Full size table

2.2 Microstructure characterization

After the soldering joints were prepared, the specimens were embedded in epoxy resin and ground using #1000, #2000, and #3000 abrasive papers. Then, 1 μm diamond polishing agents were used to polish these samples. The IMC phase was observed by Optical microscopy (OM) because it is difficult to distinguish them with a similar contrast ratio under the observation of scanning electron microscope (SEM). The energy-dispersive X-ray spectroscopy (EDS) was equipped on a Hitachi SU8230 SEM to identify the phase composition. To further confirm the phase composition of the IMC, X-ray diffraction (XRD) was applied with the diffraction angle 2θ ranging from 20° to 80°. In addition, electron backscattering diffraction (EBSD) was employed to characterize the grain morphology and orientation of the IMC phase. Furthermore, the center of the samples was selected as the observed area with a working distance of 15 mm, an accelerating voltage of 21 kV, and a step size of 0.2 μm.

To measure the thickness of IMCs, the Image J software was used to post-process the images of phase distribution [39]. Firstly, the grey-level images were applied to distinguish the interfaces between the adjacent phase layers. Then, the selected IMC layer was transformed into a binary image using an appropriate threshold. After that, the areas of the extracted layer were calculated based on the scale of the image. Finally, the average thickness of the phase layer was determined by the area divided by the length of the image:

$$h=\frac{A}{l}$$
(1)

where A is the area of the layer, l represents the length of the image, and h is the average thickness of the formed phase laye.

2.3 Mechanical properties

Nanoindentation tests were launched at room temperature with a Nano Indenter XP system. Hardness and Young’s modulus were characterized using continuous stiffness measurement (CSM) technology. Each phase is tested by five indents with more than 5 μm distances between them.

The specimens were pressed into 500 nm depth with a strain rate of 0.05 s−1 by a standard Berkovich diamond indenter, which gives additional harmonic movements with an amplified of 2 nm and a frequency of 45 Hz. Nanoindentation hardness and modulus were determined by the load-depth (P-H) curve utilizing the standard Oliver and Pharr method [40, 41]:

$$H=\frac{{{P_{Max}}}}{{{A_c}}}$$
(2)

where H is hardness, Pmax is the maximum load of indentation, and Ac is the projected contact area of the indentation, respectively.

From the slope of the unloading curve (stiffness) a reduced modulus, Er, is measured which accounts for the elastic recovery of the sample and the indenter.

$${E_r}={\left\{ {{{\left. {\frac{{1 - {\nu ^2}}}{E}} \right|}_{sample}}+{{\left. {\frac{{1 - {\nu ^2}}}{E}} \right|}_{indenter}}} \right\}^{ - 1}}$$
(3)

where E is Young’s modulus and ν is Poisson’s ratio. With the given properties of the diamond indenter (Eindenter=1140 GPa and νindenter = 0.07), the indentation modulus (ENI) is defined as

$${E_{NI}}={\left. {\frac{E}{{1 - {\nu ^2}}}} \right|_{sample}}$$
(4)

which is primarily tested here. With given Poisson’s ratio of the test material, Young’s modulus (Esample) can be calculated.

To investigate the shear strength of the aged joints, the aged samples were embedded in the self-made shearing clamps, as shown in Fig. 2. The shearing tests were carried out on the solder joints using a uniaxial microforce testing system with a shearing rate of 4 mm/min. The fracture surfaces were observed by scanning electron microscope (SEM).

Fig. 2
figure 2

Schematic illustration of shear testing of the solder joints

Full size image

3 Results and discussion

3.1 Microstructure evolution

Figure 3a–d showed the OM images of the IMC layer transited from full Cu3Sn joints with various treatment times at 600 °C. Figure 3a displayed a new phase layer, which is located at both sides of the original Cu3Sn joints after aging for 3.5 min. When the aging time reached 4.0 min, as shown in Fig. 3b, the thickness of IMC gradually increased by consuming Cu3Sn. EDS was used to identify the newly formed phase to characterize the major elements of Fig. 3b (Points A and B). Based on the EDS results (Fig. 4) and Cu/Sn binary phase diagram [42] (illustrated in Fig. 5a), the new phase layer formed in the Cu/Cu3Sn interface was identified as ζ-Cu20Sn6 phase. Furthermore, micro X-ray diffraction was used to identify the formed ζ-Cu20Sn6 phase, which was shown in Fig. 5b. The lattice parameter of Cu20Sn6 was a = 7.33 Å, b = 7.33 Å, c = 7.86 Å [43]. The Cu20Sn6 phase layer (Fig. 3c), compared to Fig. 3a and b, became thicker when the aging time was 4.5 min. At the aging time of 5 min, as expressed in Fig. 3d, the whole joint was completely occupied by Cu20Sn6. According to the phase transition in Fig. 3a–d and the analysis of Cu/Sn binary phase diagram in Fig. 5a, the ζ-Cu20Sn6 was formed above 590 °C by the following reaction:

$${\text{6 C}}{{\text{u}}_{\text{3}}}{\text{Sn+2Cu}} \to {\text{C}}{{\text{u}}_{{\text{20}}}}{\text{S}}{{\text{n}}_{\text{6}}}$$
(5)
Fig. 3
figure 3

OM images of the phase layer from full Cu3Sn joints aged at 600 °C for the following times: a 3.5 min, b 4.0 min, c 4.5 min, d 5.0 min

Full size image

However, based on the Cu–Sn phase diagram, ζ-Cu20Sn6 was just a thermodynamically stable phase without Cu or Sn elements, otherwise, it would transit to other Cu–Sn IMCs or solid solution [15] in the range of 590–640 °C. Theoretically, the ζ-Cu20Sn6 phase is metastable when it is cooled down to room temperature, it would extremely sluggish decompose to Cu3Sn and Cu41Sn11 through the eutectoid reaction at 582 °C. As a result, Cu41Sn11 was hardly observed by both XRD and EDS.

Fig. 4
figure 4

EDS results for full Cu3Sn joint aged at 600 °C with aging time of 4.0 min; a: A in Fig. 3b, b B in Fig. 3b

Full size image
Fig. 5
figure 5

a Cu–Sn binary phase diagram [42], b XRD result pattern of the joint for 600 °C with aging time of 4.0 min

Full size image

3.2 Growth kinetics of Cu20Sn6 phase

Figure 6 displays the OM images of the samples from the full Cu3Sn joints aged at different aging times and temperatures (600, 620, 640 °C). Table 2 showed the thickness of the Cu20Sn6 phase transited from Cu3Sn. And Fig. 7a illustrated the thickness changeable of the Cu20Sn6 with aging time. It was found that the thickness of the Cu20Sn6 phase increased with aging time. Moreover, when Fig. 6a (4.54 μm) is compared to Fig. 6g (6.67 μm), and Fig. 6f (6.04 μm) to Fig. 6l (9.05 μm), it can be concluded that the growth rate of the ζ-Cu20Sn6 phase is temperature dependent. Besides, as shown in Fig. 6a (4.54 μm), 6e (4.24 μm), and 6i (2.71 μm), the ζ-Cu20Sn6 phase layer became thinner with the increased aging temperature at the early stage (grain nucleation stage), which corresponded to the higher aging temperature would bring the smaller grain size of IMCs [12, 15, 44]. In addition, as marked in the Cu–Sn phase diagram with a blue bar for solidus (Fig. 5a), the Cu20Sn6 phase would transit to other IMCs or solid solutions if it was further aging. This may explain why the Cu20Sn6 would exist in a short time with increasing temperature. The growth rate of the Cu20Sn6 phase was becoming lower and lower in all aging temperatures, which inspired us to fit the growth kinetics by paralic law.

Fig. 6
figure 6

OM images of the phase layer distribution transited from the full Cu3Sn joints aged at various temperatures with different aging times

Full size image
Table 2 The thickness for Cu20Sn6 phase transited from Cu3Sn
Full size table

In order to assess the growth kinetic of the Cu20Sn6 IMC, a power-law relationship was considered to fit the growth kinetic profile [45] as follows:

$$h=K{t^n}$$
(5)

where h is the thickness of the IMC phase at aging time t, K indicates the growth rate coefficients or diffusion coefficient related to the aging temperature, what’s more, n denotes the kinetic exponent. Based on empirical evidence from previous work, the growth mechanism for the phase was determined by the value of n [46–48]. When the growth curves for the IMC obey the parabolic law, the growth kinetics is controlled by grain boundary diffusion-controlled with n is 1/3, and it is bulk diffusion-controlled with n is 1/2. [46, 47]. However, if the curves of the thickness changing with aging time t follow a straight-line law, n is equal to 1 and the growth kinetics is dominated by the interfacial reaction rate [48].

Fig. 7
figure 7

The thickness of Cu20Sn6 during the aging process as function of the aging time: a Curve of thickness variation, b fitted curves as function of thickness

Full size image

In order to obtain the activation energy and growth diffusion, the fitted line for the average thickness of the Cu20Sn6 IMC layer was plotted versus the aging time at different aging temperatures by Eq. (5) (Fig. 7b). The results revealed that the growth exponent n at 600, 620, and 640 °C was 0.48, 0.51 and 0.49, respectively, which nearly equal to 1/2. As a result, we reasonably estimated the growth of the Cu20Sn6 phase may be dominated by the bulk diffusion-controlled mechanism. So, we use n = 1/2 insert into Eq. (5) to calculate the growth coefficients of the Cu20Sn6 layers:

$${h^2}={K^2}t=kt$$
(6)

where k is a temperature-dependent growth rate constant.

To calculate the values of the growth rate coefficient (k) of new phase Cu20Sn6, the experimental data of thickness corresponding to the various aging times were fit by Eq. (6). As shown in Fig. 8a, the square of the Cu20Sn6 thickness (h2/µm2) was plotted against the various aging time (t/min) at three temperatures of 600, 620, and 640 °C. From the slope of three fitting lines, the growth rate coefficients of k were shown in Table 3.

Fig. 8
figure 8

a Plot of the square of the Cu20Sn6 thickness (h2/µm2) corresponding to the aging time (t/min), b lnk values versus 1/T on a logarithmic curve

Full size image
Table 3 Growth rate coefficients for the newly formed Cu20Sn6 phase at three aging temperatures
Full size table

Besides, the growth rate coefficients of k in Eq. (6) can also be written in Arrhenius Eq. 

$$k={k_0}\exp ( - \frac{Q}{{RT}})$$
(7)

where k0 is the growth rate constant, Q is the activation energy, R is the universal gas constant, and T is the absolute temperature.

Taking the logarithm of Eq. (7) to yield the following equation:

$$\ln k=\ln {k_0} - \frac{Q}{{RT}}$$
(8)

As displayed in Fig. 8b, lnk values were plotted against the 1/T in a logarithmic curve based on the temperature-dependent. According to the curves, the slope and intercept of the fitting line were \({\text{-}}\frac{Q}{R}\) and lnk0, respectively. After calculation, the activation energy of the Cu20Sn6 phase at the aging temperatures from 600 to 640 °C was 45.25 kJ/mol, and the corresponding growth rate constant of k0 was 3.6 × 104 µm2/min.

In our previous work, the growth of Cu41Sn11 was controlled by volume-diffusion during the aging temperature (390–450 °C) and the diffusion activation energy of Cu41Sn11 was about 108.19 kJ/mol [49]. Besides, the growth kinetic for the Cu41Sn11 phase during the aging temperature (520–590 °C) was mainly dominated by the grain boundary diffusion-controlled with the activation energy being 72.86 kJ/mol [44]. In comparison, the activation energy of Cu41Sn11 was higher than Cu20Sn6 in this work. So, the original Cu3Sn was prone to transit to IMC at higher temperatures.

3.3 Grain orientation for Cu20Sn6 phase

Figure 9a–e displayed the EBSD results of the Cu3Sn joints aged at 600 °C for 5.0 min. As displayed in Fig. 9a, the full joint was mostly consisting of Cu20Sn6. Based on the grain distribution in Fig. 9b, the grain morphology of Cu20Sn6 presented the typical columnar crystal shape, while the Cu3Sn were equiaxial grains. [15]. In addition, small grains of the Cu20Sn6 were gradually swallowed by the its larger columnar crystals [12, 15]. Herein, the normal direction (ND, Z0) was set perpendicular to the cross-section of the joints, while the rolling direction (RD, Y0) was perpendicular to the Cu surface. Figure 9c showed the IPF of the Z0 direction (ND) for the joints corresponding to Fig. 9a–b, in which indicated that the grain orientations of Cu3Sn and Cu20Sn6 exhibited different trends. The IPFs for Cu3Sn and Cu20Sn6 aged at 5 min were shown in Fig. 9d–e. As shown in Fig. 9d, it was found that the [100] and [203] directions of the Cu3Sn phase were parallel to RD [44, 50], while the [010] direction was parallel to ND. Similarly, the grain orientations along the growth direction of the Cu20Sn6 phase were [0001] direction [51], which was not obvious in the RD figure of Fig. 9e. However, the grains of Cu20Sn6 in Fig. 9c displayed green and purple colors, which corresponded to the [01–11] and [11–21] directions parallelling to the ND in Fig. 9e. Accordingly, it was not obvious that the preferred orientations of Cu20Sn6 along the growth directions (RD), while it formed[001]Cu3Sn//[01–11]Cu20Sn6//[11–21]Cu20Sn6//ND) relationship in the growth plane during the late stage for the phase transformation.

Fig. 9
figure 9

EBSD results of Cu3Sn joint aged at 600 °C for 5 min; a Phase distribution, b Grain distribution, c Z0 direction of IPF for joints, d IPF of remaining Cu3Sn, e IPF of Cu20Sn6

Full size image

3.4 Mechanical properties of formed IMCs

3.4.1 Nanoindentation properties of Cu3Sn and Cu20Sn6 IMCs

Figure 10 exhibits the representative nanoindentation load-displacement curves of the Cu3Sn and Cu20Sn6 phases in the joints. Figure 11 shows the results of hardness and Young’s moduli of Cu3Sn and Cu20Sn6 IMCs by CSM technique. The average hardness for the Cu3Sn phase is approximately 7.12\(\pm\)0.12 GPa, while the Cu20Sn6 phase is approximately 8.69\(\pm\)0.47 GPa. Besides, Young’s moduli of the Cu3Sn phase is 143.80\(\pm\)4.51 GPa, and the Cu20Sn6 phase is 166.43\(\pm\)7.32 GPa. It was obvious that the hardness and Young’s moduli of Cu3Sn were lower than the Cu20Sn6 phase.

Fig. 10
figure 10

Nanoindentation data presented as load versus displacement curves for selected IMC Cu3Sn and Cu20Sn6

Full size image
Fig. 11
figure 11

Hardness and Young’s modulus for Cu3Sn and Cu20Sn6 phase

Full size image

3.4.2 Shear strengths and fracture surfaces for the aged joints

Figure 12 exhibited uniaxial engineering shearing stress-strain curves, shearing strengths, and the fracture surface morphologies for the joints aged from 3.5 to 5.0 min at 600 °C. Figure 12a displayed shearing stress-strain curves for the full Cu3Sn joints aged at 600 °C from 3.5 to 5.0 min when the Cu–Sn joints transited to Cu20Sn6. As displayed in Fig. 12b, the shearing strength of the joint aged for 3.5 min is 35.71 MPa, while it was 48.49 MPa at 4.0 min, 56.51 MPa at 4.5 min and 60.77 MPa at 5.0 min, which indicates that the shearing strength for the soldering joints increased with the increased aging time. Moreover, the shearing strength of Cu20Sn6 phase (aging time 5.0 min) was higher than the almost full Cu3Sn phase (aging time 3.5 min).

Fig. 12
figure 12

Shearing test results for the full Cu3Sn joints aged at 600 °C from 3.5 to 5.0 min, a Shearing stress-strain curves, b variations of shearing strength for the aged joints after shearing test, c–f fracture surface morphologies for the aged joints after shearing test

Full size image

Figure 12c–f shows the fracture surface morphologies corresponding to the phase composition displayed in Fig. 3a–d. The fracture surface morphologies presented a distinct rock candy pattern in Fig. 12c, which indicate that the fracture mechanism for Cu3Sn was intergranular fracture [52]. However, when the aging time was 5 min (Fig. 12f), the fracture mechanism for the sole Cu20Sn6 phase is cleavage fracture as the surface morphologies showed an obvious river pattern. It implied that the joints consisting of sole Cu3Sn phase or Cu20Sn6 phases were brittle fractures. The fracture morphologies of the joints which consisted of Cu3Sn and Cu20Sn6 phases, shown in Fig. 12d and e, presented mixed fracture of intergranular fracture (Rock candy) and transgranular fracture (River pattern). The stress-strain curves of the joints aged at 3.5 and 5.0 min exhibited distinct brittle characteristics, which corresponded to the analysis of the fracture mechanism (Fig. 12c and f). Nevertheless, when the joints contained the Cu3Sn and Cu20Sn6 phase at an aging time of 4.0 and 4.5 min, a stress-invariant platform occurred in the stress-strain curves. This phenomenon was similar to the yielding phenomenon in the plastic material that the dislocation was prevented by the pinning effect. Because of the brittle characteristic of Cu3Sn and Cu20Sn6 phases, the stress-invariant platform was attributed to the dislocations, which were hindered by the atoms that occurred at the interface of Cu3Sn and Cu20Sn6. It may cause more energy resulting in higher strength for the joints consisting of Cu3Sn and Cu20Sn6 compared to solely the Cu3Sn phase.

4 Conclusion

Herein, it had been investigated the growth kinetics, grain orientations, and mechanical properties of the full IMC solder joints transited from full Cu3Sn joint to full Cu20Sn6 joint at an aging temperature higher than 600 °C. The results are as follows:

  1. (1)

    When the aging temperature is higher than 590 °C, the ζ-Cu20Sn6 phase was generated by the reaction of \({\text{6 C}}{{\text{u}}_{\text{3}}}{\text{Sn+2Cu}} \to {\text{C}}{{\text{u}}_{{\text{20}}}}{\text{S}}{{\text{n}}_{\text{6}}}\).

  2. (2)

    The growth kinetics exponent for the Cu20Sn6 phase at the aging temperatures of 600, 620, and 640 °C was n = 1/2, which meant the growth mechanism for the Cu20Sn6 phase was dominated by the bulk diffusion-controlled. Furthermore, the activation energy (Q) and growth rate constant (k0) were 45.25 kJ/mol and 3.6 × 10^4 µm2/min, respectively.

  3. (3)

    The preferred orientation for the final formed Cu20Sn6 phase was [0001] direction although it is not very obvious. Moreover, the [010]Cu3Sn was parallel to the [01–11]Cu20Sn6 and [11–21]Cu20Sn6 in ND, which formed a stable growth plane in the cross section of the sample.

  4. (4)

    The average hardness for Cu3Sn and Cu20Sn6 phase was approximately 7.12 ± 0.12 GPa and 8.69 ± 0.47 GPa, respectively. Besides, the corresponding Young’s moduli of the Cu3Sn phase was 143.80 ± 4.51 GPa, while it was 166.43 ± 7.32 GPa for Cu20Sn6. It indicated that the Cu20Sn6 phase had better performance on hardness and Young’s moduli.

  5. (5)

    The fracture mechanism for Cu3Sn was an intergranular fracture, while it was a cleavage fracture for Cu20Sn6. Besides, when the joints consist of Cu3Sn and Cu20Sn6, the fracture mechanism was presented as a mixed fracture of intergranular and transgranular fracture. Last but not least, the shearing strength for the soldering joints increased with increasing aging time from 35.71 (almost full Cu3Sn) to 60.77 (Cu20Sn6) MPa.