The ceramic nanoparticles are chemically non-reactive in the solder joints. They do not form any chemical compounds neither with the solder nor with the substrate during the soldering process. Therefore, it is extremely challenging to find their traces in the solder joints by means of EDX analysis. One of the most popular indirect methods to indicate that the main part of the ceramic NPs remains in the solder is to inspect the morphology and microstructure of the part on the solder joints where the highest concentration of the nanosized inclusions is expected. In the present study, the ceramic nanoparticles mixed with the flux to reach the maximal concentration of those between the solder and substrate. According to the theory of adsorption of a surface-active material, the ceramic nanoparticles should maximize the number of adsorbed particles on the surface of the intermetallic compounds decreasing their surface energy and growth rate (Yakymovych et al. 2016). Therefore, the discontinuous scallop-type shape of the Cu6Sn5 interface IMC layer between the solder and substrate, which is common for the SAC and SA solder joints, should change into a more planar-type shape.

The SEM micrographs and the corresponding EDX analysis of the as-solidified samples show that the interfacial Cu6Sn5 grains, which are extended into the solder matrix, become to be smaller for the joints with nanosized ceramic additions (Fig. 2a–g). Addition of the nanoparticles as adsorption elements changes the driving force for the growth of the interfacial IMCs to the more planar direction and decreases the surface energy of the IMCs layer which leads to a reduction of the growth velocity of that plane. The morphology of the interfacial Cu6Sn5 layer has shown a change from scallop-like to prism-like for the samples with ZrO2 nanoinclusions (Fig. 2b–d). A very thin and planar Cu3Sn IMC layer is detected between the Cu6Sn5 IMC layer and the Cu substrate similar to our current study of the hybrid Sn–3.5Ag solder joints with Ni NPs. This IMC layer observed by conventional SEM is almost undistinguishable from the unaided eye due to a very thin Cu3Sn IMC. That could be a reason, why it was not detected in the as-reflowed hybrid SAC305 solder joint with nano-sized Al2O3, SiO2, TiO2, and ZrO2 (Aspalter et al. 2020). The changes in the thickness of the interfacial IMCs layer in the as-solidified hybrid Sn–3.5Ag solder joints are presented in Fig. 3 in percentage to compare it with the literature data both for the nanocomposite (Yakymovych et al. 2016) and hybrid SAC305 solder joints (Aspalter et al. 2020). The thickness of the interfacial Cu6Sn5/Cu3Sn layer in the Sn–3.5Ag solder joint equals 5.60 µm, 4.60 µm in the nanocomposite SAC305 solder joint, and 3.30 µm in the hybrid SAC305 solder joint.

SEM micrographs of the as-solidified Cu/nanocomposite flux/Sn–3.5Ag/nanocomposite flux/Cu hybrid solder joints with 0.0 wt% (a), 0.2 wt% nano-ZrO2 (b), 0.5 wt% nano-ZrO2 (c), and 1.0 wt% nano-ZrO2 (d)

Thickness changes of the interfacial (Cu, Ni)–Sn IMCs layer in the as-solidified solder joints with nanosized ceramic inclusions

The main purpose of Fig. 3 is to establish if we have a similar tendency by adding ceramic NPs using two different approaches of introducing the ZrO2 NPs additions into the solder joint, namely either via mixing with solder paste or with flux placed between solder and substrate. It shows that the thickness of the IMCs layer decreased in both cases, while the maximal effect was reached at lower concentrations by using the nanocomposite flux + solder foil compared to the nanocomposite Sn–3.5Ag solder joints. This effect is less pronounced in the joints with the higher amount of the ZrO2 NPs additions due to their agglomeration and segregation resulting in a decrease in the surface energy and a reduction in the number of the surface-active nanoparticles. It is hardly possible to perform the qualitative comparison due to a lack of additional information such as the proportion of the nanoinclusions concentrated at the interface solder/substrate. However, it was also expected that the present method should localize the most amount of those in this area. The interfacial Cu3Sn layer is practically not affected by additions of the ceramic nanoinclusions.

A part of the samples was thermally aged at the temperature of 453 K for 96, 192, and 312 h, respectively (Fig. 4).

SEM micrographs of the thermal aged Cu/nanocomposite flux/Sn–3.5Ag/nanocomposite flux/Cu hybrid solder joints with 0.2 wt% (a), 0.5 wt% (b), and 1.0 wt% (c) of nanosized ZrO2 particles at 423 K for 96 h (I), 192 h (II), and 312 h (III)

The continuous scallop-type shape of the interfacial IMC layer obtained in the as-reflowed samples was transformed to a more planar shape with increased aging time and increased amount of the nanoinclusions up to 1 wt%. It should be also noted that the thickness of not only Cu6Sn5 IMC but also of the Cu3Sn IMC was significantly increased during the aging and remained the flat-type shape.

There are several works related to the analysis of such changes in growth kinetics based on empirical kinetic laws. It is expected that the interfacial IMCs layers grow continuously during isothermal aging by the following reactions:

$$}_} }_ + 9} \to 5}3};$$

Arafat et al. (2020) have modeled intermetallic growth between Cu and Sn aged at different temperatures. It is supposed that the much slower growth rate of the interfacial IMCs in the aged solder joints at temperatures lower than 373 K, and especially of Cu3Sn, is mainly related to the substantially slower diffusion of Cu atoms to the IMC growth fronts at these temperatures.

The Cu/Ag3.5–Sn/Cu solder joints thermal aged at 453 K for 96 h show a proportional increase in the thickness of both interfacial Cu–Sn IMCs. Further aging has led to much slower growth of the interfacial Cu6Sn5 layer even being practically constant in the time period between about 100 h and 200 h (Fig. 5). A decrease in the growth rate of the interfacial Cu6Sn5/Cu3Sn layer in the hybrid SAC305 solder joints with minor nanosized ZrO2 inclusions was indicated after 137 h of the isothermal aging at 453 K by Aspalter et al. (2020). At that time, the growth rate of the interfacial Cu3Sn layer remained constant for the entire time period. Ajay Kumar and Dutta (2018) pointed out that the interfacial Cu3Sn grows mainly by consumption of the Cu substrate and Cu6Sn5 IMC layer, while the Cu6Sn5 grows essentially at the Sn/Cu6Sn5 interface and to a lesser extent at the Cu3Sn/Cu6Sn5 interface. This tendency is stronger pronounced in the solder joints doped with ZrO2 because ceramic nanosized inclusions decrease the growth kinetics of the Cu6Sn5 IMCs. In contrast, the growth rate of the Cu3Sn IMC remains to be similar to the undoped solder joint.

The average thickness of the interfacial Cu6Sn5 (triangles) and Cu3Sn (circles) IMCs in the thermally aged samples at 453 K

The relationship between the IMC layer thickness and aging time can be expressed by the following equation:

$$\left( - x_ } \right) = (D_}}} t)^$$

(1)

where x0 is the IMC thickness of the as-solidified joint and xt is the IMC thickness at aging time t. The effective diffusion coefficient of the IMC during thermal aging Deff can be determined from the slope of the linear dependence between the IMC thickness, (x0 − xt), and the square root of the aging time, t1/2 (Fig. 6):

$$\left( - x_ } \right) = a + b \times t^$$

(2)

The aging time dependence of the average thickness of the interfacial IMCs layer in the aged solder joints at 453 K

The calculated parameter b of Eq. (2) and the effective diffusion coefficient are presented in Table 2.

The increase of the ceramic nanoinclusions up to 1.0% into the flux continuously decreases the effective diffusion coefficient of the interfacial Cu–Sn IMCs layer showing that this is an effective way to reduce the thickness of the isothermally aged solder joints, which is very important for thin microjoints.

Unfortunately, we have not found any literature data on the effective diffusion coefficient of the interfacial IMCs layer in both undoped and doped Sn–3.5Ag solder joints. The obtained results are compared with the literature data for the hybrid SAC305 solder joint with ZrO2 and SiO2 nanoadditions, where these two ceramic powders have shown the most profound effect compared to the other ceramic powders used in that study, namely Al2O3 NPs and TiO2 NPs (Fig. 7). As shown from Fig. 7, the addition of 0.5 wt% of ceramic nanoparticles into the flux decreased the average thickness of the interfacial Cu–Sn IMC layer in the thermally aged Sn-based solder joints, while this effect is more profound in the investigated Sn–3.5Ag solder joint comparing with the literature data for the SAC305 solder joint (Aspalter et al. 2020). This effect remains constant within the entire aging time for the investigated solders supposed that the thermal constraints do not have a significant impact on the adsorbed NPs on the Cu6Sn5 surface. Furthermore, Fig. 7 shows that the addition of the ZrO2 nanoparticles seems to have a more profound effect on the interfacial IMC layer in the thermally aged hybrid SAC305 solder joints than SiO2 (Aspalter et al. 2020). However, additional investigations focused on the hybrid Sn–3.5Ag + SiO2 joints are needed to confirm it.

The aging time dependence of the difference in the average thickness of the interfacial IMCs layer in the solder joints with 0.5 wt% ceramic nanoinclusions at 453 K (filled circle—SAC305 + ZrO2 NPs; filled square—SAC305 + SiO2 NPs)