Typical acquired wetting balance curves that represent the change of measured force (F
r) versus time (t) and characterize the dynamic wetting behavior of the SAC305 solder on PCB substrates with different surface finishes are shown in Fig. 1 and 2 for RF800 and EF2202 flux, respectively. The results obtained from wetting balance tests are summarized in Fig. 3, where Fig. 3(a) presents the data obtained at the maximum value of force F
r (resultant force measured by the balance):
$$ F_{\text{r}} = F_{\text{w}} - F_{\text{a}} $$
(1)
This static force was achieved after specified time when an equilibrium state during wetting tests was reached. In Fig. 3(b), one may compare the values of wetting time t
0, where t
0 denotes the moment when the wetting force F
w is equal to buoyancy force (F
a—Archimedean push) and at that time, the measured wetting force is zero. Knowing F
r, we can extract the wetting force F
w from the measurement by adding the Archimedean push F
a:
$$ F_{\text{a}} = \uprho * \upnu * g $$
(2)
$$ F_{\text{w}} = \upgamma_{\text{LV}} * l * \cos \uptheta $$
(3)
$$ \upgamma_{\text{LV}} * l * \cos \uptheta = F_{\text{r}} + \uprho * \upnu * g $$
(4)
The obtained results show that the shortest wetting time t
0 was noted for the OSP finish (t
0 = 0.6 s with EF2202 flux and t
0 = 0.98 s with RF800 flux), while the longest one was in the case of the ENIG finish (t
0 = 1.36 s with EF2202 flux and t
0 = 1.55 s with FR800 flux) (Fig. 1, 2). In the case of the HASL LF finish, the wetting time reached the value of t
0 = 1.23 s with EF2202 flux and t
0 = 1.17 s with RF800 flux. The buoyancy force F
a has the value of F
a = −0.93 mN.
The measured force F
r of SAC305 with the HASL LF finish using RF800 flux (Fig. 1) was larger (F
r = 1.50 mN) than those obtained for the OSP (F
r = 0.46 mN) and ENIG (F
r = 0.76 mN) ones. With EF2202 flux (Fig. 2), the largest value of the recorded force F
r was obtained also for the HASL LF finish (F
r = 1.80 mN).
According to Fig. 3(a) and (b), the OSP finish with both fluxes shows the smallest values of the recorded force (F
r = 0.52 mN with RF800 flux and F
r = 1.03 mN with EF2202 flux) as well as the wetting time t
0 = 0.98 s with RF800 flux and t
0 = 0.6 s with EF2202 flux.
The contact angle values θ for examined combinations of materials in solderability tests are shown in Fig. 3(c) and are collected in Table 1. The final contact angle values were calculated on the basis of the data recorded in the wetting balance tests using software supplied with the Menisco ST88 apparatus. The value θ is representative of the wetting quality and it characterizes the solderability of the materials. The smaller the θ, the better both wettability and solderability. The lowest value of the contact angle for SAC305 is observed on the HASL LF finish (θ = 45°), while the highest one is formed on the OSP type (θ = 63°). No significant difference in the wetting behavior and contact angles was noted between the selected three fluxes.
Table 1 Contact angles formed with SAC305 solder on different PCB surface finishes (260°, 3 s)
Solderability quality is classified into four classes as follows (Ref 18, 20):
-
Class 1: excellent quality, contact angle 0-30°;
-
Class 2: good quality, contact angle 31-40°;
-
Class 3: admissible quality, contact angle 41-55°;
-
Class 4: uncertain quality, contact angle 56-70°.
The results of wetting balance tests show good agreement with visual observation of the degree of solder spreading over the PCB surface finishes. For the HASL LF finish and RF800 flux (Fig. 4a), the solder covers 2/3 of the test coupon surface with a corresponding contact angle θ of 45° and solderability class 3. In the case of the OSP finish and RF800 flux, the solder covers a much smaller area of only 1/4-1/3 of the surface (Fig. 4b), and it forms the higher contact angle value of 63° that corresponds to the solderability class 4.
Figure 5 shows the LM-DIC images of the microstructure of the cross-sectioned samples after wetting balance tests. During the 3-s immersion time, molten SAC305 solder reacts with the substrate to form the interfaces of complex structures. Their SEM/EDX observations confirmed the formation of the continuous IMC at the substrate/solder interface. Figure 6 presents the comparison of the microstructure of this interfacial region for all the types of finishes, each represented by two pictures. It could be easily noticed that the IMC shows typical scallop-shaped morphology. Its thickness is very similar for the OSP and HASL LF finishes, which is about 2 μm (Fig. 6a, b, d, e), while for ENIG, the IMC layer is thinner (~0.5 μm). Typically, for wave soldering process, such thin intermetallic layers are observed in the case of SnPb solder, while for lead-free ones, they are twice thicker. Spalling of the IMC was mostly observed in the case of the HASL LF finish (Fig. 6b, e), regardless of the flux type used. On the other hand, the EDX microchemical analysis showed that for both HASL LF and OSP finishes, the intermetallic layer at the solder/side interface is the Cu6Sn5 phase. Conversely, for the ENIG finish, the chemical composition of the IMC layer (6.7Ag, 54.5Sn, 20.0Ni, and 18.8Cu, at.%) corresponds to the (Ni,Cu)3Sn4 phase. This phase was deduced taking into account three facts: (1) the tin amount corresponds to the Ni3Sn4 phase, (2) no excess of copper, which could indicate the formation of Cu6Sn5, is detected, and (3) the IMC layer is very thin and therefore information on local chemical composition obtained by EDS analysis also comes from the matrix due to the beam broadening. This result confirms the validity of the thesis that among the two most popular intermetallics, Ni3Sn4 and Cu6Sn5, the first one should be formed primarily if the copper content is below 0.6 wt.%, which corresponds to 1.11 at.% (Ref 21). Moreover, only in the case of the ENIG finish, the enrichment in silver (up to 8.6 at.%) within (Ni,Cu)3Sn4 was observed. According to previous studies (Ref 22), such enrichment was noted for the Cu6Sn5 phase where the absorption of fine Ag3Sn particles on the IMC surface took place, particularly for the scallops of Cu6Sn5 formed by the ripening because the Ag3Sn nanoparticles decrease the surface energy and hinder the growth of the Cu6Sn5 phase.
During soldering with the ENIG finish, the original Ni-P plating transformed into a multilayer region where four layers can be distinguished (Fig. 6c, f). Two of them characterized by the largest thickness were of similar contrast and composition. The upper layer next to the IMC (marked with No. 1) was composed of 18.8-19.3 P, 1.3-1.5 Sn, and 79.2-80 Ni (at.%), while the bottom one (marked by No. 3 in Fig. 6c, f) contained 31.3 P and 68.7 Ni (at.%). Two other layers (No. 2 and No. 4 in Fig. 6c, f, respectively) were too thin for EDS measurements as in their case, the beam broadening and escape depth for x-rays hinder the analysis of the individual layer. Following the literature data, during processing, the Ni-P amorphous plating transforms into a layer containing 17.1 at.% P and 82.9 at.% Ni and a P-rich layer containing 28 at.% P and 72 at.% Ni, which is identified mostly as Ni3P or Ni12P5 phase, respectively (Ref 23, 24).