III. RESULTS AND DISCUSSION

Figure 1(a) is the backscattered electron imaging (BEI) micrograph of the as-prepared Sn/Ni–8.0 at.%V couple. The solder and the substrate were held at 400 °C for 30 s before they were quenched in water. As shown in Fig. 1(a), a reaction layer of 1∼2 μm was formed. Figure 1(b) is the BEI micrograph of the couple reacted at 210 °C for 24 h, and two reaction layers can be observed. Compositional analysis results of the two layers are Sn–20.57 at.%Ni–63.84 at.%V and Sn–43.54 at.%Ni, indicating the T phase layer and the Ni₃Sn₄ layer, respectively. These results are similar to those previously reported by Chen et al.Reference Chen and Chen^3, Reference Chen, Chen and Chang⁴ Accordingly, it is concluded that the reaction paths in the solid/solid Sn/Ni–8.0 at.%V couples are Sn/T/Ni–V and then Sn/T/Ni₃Sn₄/Ni–V. It needs mentioning that although the T phase region appears like a single phase region and it is named the “T” phase for consistency with the nomenclature in the literature,Reference Chen and Chen^3, Reference Chen, Chen and Chang⁴ the results in this study and previous studiesReference Chen and Chen^3, Reference Chen, Chen and Chang⁴ have clearly demonstrated that the T phase region is indeed a two-phase mixture.

FIG. 1.

Backscattered electron imaging (BEI) micrograph of the Sn/Ni–8.0 at.%V couple: (a) as-prepared and (b) reacted at 210 °C for 24 h.

Figures 2(a) and 2(b) are the BEI micrographs of the etched results of the Sn/Ni–8.0 at.%V couples reacted at 250 °C for 40 min and for 24 h, respectively. Two continuous reaction layers and Ni₃Sn₄ particulates either attached to the second layer or dispersed in the Sn matrix are found. The EPMA measurements along the two different layers are shown in Table I. The positions of the measurement numbers as shown in Table I are marked on Fig. 2(a). Based on the compositional measurements, the gray phase adjacent to the Ni–8.0 at.%V is the T phase as reported by Chen et al.Reference Chen, Chen and Kao^2–Reference Chen, Chen and Chang⁶ The second phase layer adjacent to the Sn phase has lower Ni content and has not been reported previously.Reference Tu and Zeng^1–Reference Tsai, Lin and Kao¹² It should be mentioned that the contrast between the second phase layer and the Sn matrix is not clear, and the second phase layer can be easily overlooked without etching.

FIG. 2.

BEI micrograph of the etched Sn/Ni–8.0 at.%V couple: (a) reacted at 250 °C for 40 min and (b) reacted at 250 °C for 24 h.

TABLE I.

The electron probe microanalysis measurements along the two different layers in the Sn/Ni–8.0 at.%V couples reacted at 250 °C for 24 h.

Figure 3(a) is the TEM BF image and the diffraction pattern of the T phase in the Sn/Ni–8.0 at.%V couple reacted at 250 °C for 40 min. From the BF image, it is observed that the T phase has a very fine grain structure with a grain size of about 30 nm. The diffraction pattern of the T phase is shown in the right-side of Fig. 3(a). The diffraction spots are mixed with diffraction rings and are likely to be contributed from the ultrafine grains and an amorphous phase, respectively. Figure 3(b) is another TEM BF image of the T phase with higher magnification, together with two highlighted squares obtained by using the fast Fourier transform (FFT) of highlighted squares in Fig. 3(b). Both from Figs. 3(a) and 3(b), it can be seen that the T phase is composed of a crystalline phase region and an amorphous phase region. The crystalline phase region is the Ni₃Sn₄ phase while the amorphous phase region is a V-rich region. These results are consistent with the TEM studies by Chen et al.Reference Chen, Chen and Kao^2, Reference Chen, Chen and Chang^4, Reference Chen, Chen and Chang⁵

FIG. 3.

T phase in the Sn/Ni–8.0 at.%V couple reacted at 250 °C for 40 min: (a) Transmission electron microscope (TEM) bright field (BF) image and (b) TEM BF image with higher magnification rate.

Figures 4(a) and 4(b) are TEM BF images and the corresponding diffraction patterns of the second phase layer as shown in Figs. 2(a) and 2(b). It appears that there are darker particulates with size of 50–80 nm dispersed randomly in a brighter continuous matrix. The diffraction patterns on the upper-right corner of Fig. 4(a) are similar to those as shown in Fig. 3(a), which represents a mixture of crystalline phase and an amorphous phase. Moreover, as shown in the right-side of Fig. 4(b), two distinct diffraction patterns, obtained through the FFT, are further indication of the crystalline particulates and the amorphous matrix. The EPMA-determined composition of the darker particulate phase region is Sn–4.2 at.%Ni–17.3 at.%V and that of the brighter phase region is Sn–1.1 at.%Ni–0.5 at.%V. The brighter matrix region is primarily the Sn phase, with very limited solubility of Ni and V.

FIG. 4.

TEM BF image of the second phase layer in the Sn/Ni–8.0 at.%V couple: (a) reacted at 250 °C for 40 min and (b) reacted at 250 °C for 24 h.

To further analyze the reaction layers, the Sn matrix in the Sn/Ni–8.0 at.%V couple reacted at 250 °C for 24 h was removed. The removal was conducted by very careful polishing so that the Sn layer was completely removed but not overpolished so that the second reaction layer was preserved. Figure 5 is a diffractogram of the reaction layer using the grazing angle x-ray diffraction analyzer, showing both Sn phase and “VSn₂(V₂Sn₃)” phase. It should be mentioned that the phase diagram of the Sn–V binary system has been revised,Reference Chen, Gierlotka and Chen¹⁴ confirming that the V₂Sn₃ phase is better described as VSn₂. To avoid confusion, the terminology “VSn₂(V₂Sn₃)” is used in this study.

FIG. 5.

Diffraction pattern of the reaction layer using the grazing angle x-ray diffraction analyzer.

It is proposed that the T phase layer in the Sn/Ni–8.0 at.%V couple is formed by the fast diffusion of Sn into Ni(V) matrix.Reference Chen, Chen and Kao^2–Reference Chen, Chen and Chang⁶ At 250 °C, the Ni in the outer region of the T phase layer diffuses outward to the molten Sn. The Ni-depleted T phase region transforms into the loosely connected VSn₂(V₂Sn₃) phase, as shown in Fig. 4(a). Since Sn is molten at 250 °C, the liquid penetrates into the reaction layer and mixes with the VSn₂(V₂Sn₃) phase, and the two phases together form the second reaction layer. The reaction path evolutions in the molten Sn/solid Ni–8.0 at.%V couples are liquid/T/Ni–V, liquid/liquid + Ni₃Sn₄/T/Ni–V, liquid/liquid + Ni₃Sn₄/liquid + VSn₂(V₂Sn₃)/T/Ni–V, and liquid/liquid + Ni₃Sn₄/liquid + VSn₂(V₂Sn₃). In the study by Osenbach et al.Reference Osenbach, Amin, Bachman, Baiocchi, Bitting, Crouthamel, Delucca, Gerlach, Goodell, Peridier, Stahley and Weachock¹¹ which was conducted at 150 °C, the solid Sn could not fill in the voids when the loosely connected VSn₂(V₂Sn₃) phase was formed, and thus the reaction layer was porous.

However, no porous VSn₂(V₂Sn₃) phase layer or a reaction phase layer composed of Sn and VSn₂(V₂Sn₃) phases was observed in the solid/solid solder/Ni–8 at.%V couples reacted at 210 °C in this study or those at lower temperatures by Chen et al.Reference Chen, Chen and Kao^2–Reference Chen, Chen and Chang⁶ It is likely the outward Ni-flux from the reaction phase layers into the solid Sn phase is not as large as that into the molten Sn, and it can be balanced by the inward flux from the Ni–V substrate. Thus the reaction layers, the T phase and the Ni₃Sn₄ phase layer, are sustained without transformation to the porous SnV₂(V₂Sn₃) phase when there is remaining Ni–V substrate.

Figure 6(a) is a BEI micrograph of the Sn/Ni–8.0 at.%V couple reacted at 400 °C for 120 s, and two reaction layers are observed. Figure 6(b) is micrograph of the same couple further heat-treated at 210 °C for another 24 h. It is found that the T phase and the Ni₃Sn₄ phase grow, while the Sn + SnV₂(V₂Sn₃) layer shrinks. The results indicate that the Ni flux from the Ni–V substrate in the solid/solid Sn/Ni–V couples can “heal” the Ni-deficient layer produced by the liquid/solid reaction at higher temperatures.

FIG. 6.

BEI micrograph of the Sn/Ni–8.0 at.%V couple: (a) reacted at 400 °C for 120 s and (b) reacted at 400 °C for 120 s and further heat-treated at 210 °C for another 24 h.

These results support the proposed mechanism in which the T phase is not converted into the porous VSn₂(V₂Sn₃) phase in the solid/solid reactions if there is unreacted Ni–8 at.%V substrate. In contrast to the samples used by Osenbach et al.Reference Osenbach, Amin, Bachman, Baiocchi, Bitting, Crouthamel, Delucca, Gerlach, Goodell, Peridier, Stahley and Weachock¹¹ which had a limited supply of Ni, the samples used in this study and those by Chen et al.Reference Chen, Chen and Kao^2–Reference Chen, Chen and Chang⁶ have relatively thicker Ni–8.0 at.%V layer. Since the Ni–V layer is not completely consumed, the porous VSn₂(V₂Sn₃) phase is thus not observed. This mechanism is further supported by the observation that no porous reaction layer is observed at the solder/Ni–V Under bump metallurgy (UBM) interface in the initial stage of reactionsReference Liu, Tu, Sheng, Tung, Frear and Elenius^7–Reference Osenbach, Amin, Bachman, Baiocchi, Bitting, Crouthamel, Delucca, Gerlach, Goodell, Peridier, Stahley and Weachock¹¹ when there are unreacted Ni–V substrates. According to all experimental results determined in this study and those in the literature,Reference Tu and Zeng^1–Reference Tsai, Lin and Kao¹² the reaction paths in the solid/solid Sn/Ni–8.0 at.%V couples can be further expanded to longer reaction time and are Sn/T/Ni–V, Sn/Ni₃Sn₄/T/Ni–V, and Sn/Ni₃Sn₄/VSn₂(V₂Sn₃). The evolution of the solid/solid and liquid/solid paths in the Sn/Ni–8.0 at.%V reaction couples are shown in Figs. 7(a) and 7(b), respectively.

FIG. 7.

The evolution of the reaction path in the Sn/Ni–8.0 at.%V couple: (a) solid/solid reaction and (b) liquid/solid reaction.

The electrical resistance in the Sn–Pb/UBM(Al/Ni(V)/Cu) couples was determined by Osenbach et al.Reference Osenbach, Amin, Bachman, Baiocchi, Bitting, Crouthamel, Delucca, Gerlach, Goodell, Peridier, Stahley and Weachock¹¹ For reaction at 150 °C up to 463 h, the electrical resistance is almost constant, but the electrical resistance increases almost 500 times to 900 ohms at the end of the reaction for 800 h. According to the reaction paths shown in Figs. 7(a) and 7(b) and the electrical property measurements,Reference Osenbach, Amin, Bachman, Baiocchi, Bitting, Crouthamel, Delucca, Gerlach, Goodell, Peridier, Stahley and Weachock¹¹ it can be concluded that the formation of the amorphous phase region in the solder/Ni(V) interface in the early stage does not cause significant change of the electrical resistance. The formation of the porous Sn–V phase layer is the main cause for the degradation of electrical properties.