2020 Volume 61 Issue 10 Pages 1900-1906
Recently, our group employed surface fine crevice structures produced on Cu metal substrates using either laser irradiation or the reduction-sintering of oxide powders to induce region-selective super-spread wetting as a means of joining these substrates. The present work expanded the scope of this method by joining Cu and Fe substrates. Laser irradiation was found to permit the joining of Cu and Fe but generated voids at the join interface. In contrast, the reduction-sintering of mixed oxide powders allowed joining with essentially no voids. Thus, the latter technique is superior to laser irradiation when joining dissimilar metals.
Fig. 10 A cross-sectional SEM image of the joint between Fe and Cu substrates having surface fine crevice structures produced by reduction sintering of a CuO–Fe2O3 mixture and EDX elemental maps for Sn, Fe and Cu.
Super-spread wetting is a phenomenon in which a liquid metal spreads over a complicated surface structure having many fine pores via capillary action.1–8) Tanaka and coworkers1–3) originally discovered that this process occurs on the surface fine porous structure produced by atmospheric oxidation-reduction method. Taking advantage of super-spread wetting, they successfully joined Cu wire to Cu substrate with minimal formation of solder fillets, which can otherwise result in spatial limitations on printed circuit boards. In previous research by our own group, we devised two methods of producing surface fine crevice structures that promoted the region-selective super-spread wetting of liquid materials. These techniques consisted of the laser irradiation of Cu and Fe surfaces4–7) and the reduction-sintering of an oxide powder on a Cu surface.8) The super-spread wetting of these two forms of surface fine crevice structures on Cu surfaces showed the potential to allow the joining of Cu plates.
Recently, the joining of dissimilar metals has become increasingly of interest and has been applied in industry because of the various benefits this process provides, including weight reductions, low manufacturing costs and enhanced mechanical and thermal properties. Super-spread wetting is thought to be applicable to the joining of Cu substrates to one another and of two dissimilar metals. The present study therefore investigated the joining of Cu and Fe via the super-spread wetting of liquid Sn on these two metals following the formation of surface fine crevice structures. The aim was to confirm the feasibility of joining dissimilar metals so as to expand the applications of super-spread wetting. At first, the super-spread wetting on the surface fine crevice structure by laser irradiation was tried to be applied to the dissimilar metal joining. Then, that by the reduction-sintering of oxide powders was challenged by newly proposing the utilization of mixed oxide to the creation of the surface fine crevice structure. Here, we focus on the composition distribution at the interface between the two materials, which can affect the compound formation resulting in the generation of voids. In regard to the above two kinds of trials, it is significant note that the compositions that consist in the surface fine crevice structure created by the reduction-sintering of oxide powders could be controlled by using the powders of desired compositions, while that created by laser depends on the substrate materials.
A 15 × 20 × 2 mm Cu substrate and a 5 × 10 × 1 mm Fe substrate were employed as the lower and upper substrates, respectively, to be joined. The purities of the Cu and Fe substrates were 99.96% and 99.5%, respectively. Both substrates were first polished with abrasive paper (#600 and #1000) and then cleaned by ultrasonic agitation in ethanol. Laser irradiation was conducted to produce surface fine crevice structures over a 11 × 10-mm region of the Cu substrate and the entire surface of one side of the Fe substrate, as schematically illustrated in Fig. 1. A Q-switched neodymium-doped yttrium aluminum garnet (Nd:YAG) laser (Miyachi Corporation, ML-7062) with a pulse rate of 6.0 kHz, wavelength of 1064 nm and output power of 50 W was used in this study. The laser beam was applied to each specimen in a reticular pattern at a scanning rate of 10 mm/s and scanning interval of 0.01 mm. It has been shown that Cu and Fe oxides such as CuO, Cu2O, FeO, Fe3O4 and Fe2O3 are generated in response to such irradiation.4,6) After the laser irradiation, the Cu and Fe substrates were heated under H2 (99.995% purity at a flow rate of 20 mL/min) at 773 K for 1 h to reduce the formed oxides that could be the primary cause of decreasing the wettability of the solid metal surfaces by liquid metal. We confirmed that surface fine crevice structures, that were produced on the Cu and Fe by the above method (Fig. 2), display super-spread wetting by preliminary wetting experiments using Sn as the liquid metal. Sn was applied as the liquid material because it is the main component of solder alloys and its melting temperature is 505 K.9) According to the Cu–Sn10) and Fe–Sn11) binary phase diagrams, liquid Sn would also be expected to react with solid Cu and Fe to form intermetallic compounds (IMCs) at the interfaces during joining experiments. In each trial, a Cu substrate was placed underneath a Fe substrate, such that the laser-irradiated surfaces were in contact with one another. We confirmed that the 0.44 g/mm2 of pressure is required for pressing the upper Fe substrate so that it does not move during the wetting of Sn through the preliminary tests. Therefore, a loading pressure of 0.44 g/mm2 was applied by placing a Cu block on top of the Fe substrate, to ensure good contact at the interface during the infiltration of the molten Sn. An approximately 0.1 g of Sn was placed on the laser-irradiated area of the lower Cu substrate with some distance from the Fe substrate, as illustrated in Fig. 1, to avoid direct contact with the Cu/Fe interface, after which the entire assembly was transferred into a furnace and residual air was evacuated. A flow of H2 was subsequently introduced into the furnace at a flow rate of 20 mL/min, and the sample was heated to 773 K at a heating rate of 10 K/min. Once the temperature reached 773 K, the sample was cooled to room temperature at a rate of 10 K/min.
Schematic illustration of the process used during the Cu–Fe joining experiments. The lower Cu and upper Fe substrates are brought into contact with one another, each having surfaces with fine crevice structure.
Surface and cross-section images of the surface fine crevice structures formed on (a) a Cu substrate and (b) a Fe substrate by laser irradiation.
Figure 3 presents images showing the appearance of a test sample after a joining experiment. The Sn completely wet the irradiated surface of the lower Cu substrate via the super-spread wetting phenomenon. Even when the sample was inverted (Fig. 3(b)), the substrates did not separate, indicating that the Cu and Fe substrates were securely joined by super-spread wetting into the surface fine crevice structures that had been formed. A cross-section of the joint was observed by scanning electron microscopy (SEM) and analyzed by energy-dispersive X-ray spectroscopy (EDX), with the results presented in Figs. 4 and 5. Although a large void can be seen at the joint, it is believed that the Sn completely penetrated the surface fine crevice structures of the Cu and Fe substrates and filled the entire gap along the laser-irradiated surfaces at the high temperature applied during the experiment. This is evident because the contour of the void on the upper substrate (the Fe side) is similar to that on the lower side (the Cu side). Our previous study showed no evidence for the formation of large voids at Cu–Cu homogenous metals joining produced using the same procedure.4,5) Voids such as these at the interface of solder joints would likely significantly degrade the mechanical robustness of the interconnection and consequently affect the drop reliability and joint life.12)
Photographic images showing (a) the sample after joining of Cu and Fe substrates with surface fine crevice structures produced by laser irradiation and (b) the same sample inverted.
A cross-section of the join between the lower Cu and upper Fe substrates having fine crevice structures produced by laser irradiation.
A cross-sectional SEM image of the joint between Fe and Cu substrates having surface fine crevice structures produced by laser irradiation and EDX elemental maps for Sn, Fe and Cu.
Figure 5 shows the elemental maps for Sn, Fe and Cu at the joint as acquired using EDX. Sn was evidently present throughout the entire region between the Fe and Cu substrates, although a concentration difference exists at the Fe substrate side. In addition, there was a variation in the Fe concentration at the Fe side. At the Fe substrate side, the regions of high Fe concentrations were not in accordance with those of Sn, while the regions in which the Fe and Sn levels were elevated tended to overlap. The concentration of Fe becomes relatively low at the region apart from the vicinity of the Fe substrate, while the Cu shows a high concentration at the Cu substrate side (below the void) and a moderate concentration at the Fe substrate side (above the void). These concentration distributions suggest that several interfacial reactions occurred at the solid Cu/liquid Sn/solid Fe interfaces. According to the Cu–Sn10) and Fe–Sn11) binary phase diagrams several IMCs could potentially form at the Cu and Fe sides and Fig. 4 shows the positions at which spot EDX analyses were performed while Table 1 displays the results. The dark regions with complicated shapes observed at the Fe substrate side correspond to the remains of the surface fine crevice structure formed on the Fe. On the other hand, Cu substrate side shows less concavo-convex shape than Fe substrate side. It is implied that the outmost surface of the Cu and Fe surface fine crevice structure dissolved into the liquid Sn, changing in the concavo-convex shape and volume from original surface fine crevice structure shown in Fig. 2. Moreover this difference in the degree of structure shape change between Fe and Cu can perhaps be attributed to the different solubilities of Fe and Cu in liquid Sn.10,11) That is, the Cu dissolved in the liquid Sn to a greater extent than the Fe such that the surface fine crevice structure on the Cu was not maintained. The compositions close to the Cu and Fe sides were also different, as shown in Table 1. Based on the concentrations of Cu, Sn and Fe, and phase diagrams for Fe–Sn and Cu–Sn systems, it is supposed that the layer with the thickness ranging from approximately 20 µm to 45 µm generated at the Cu side (position 7) correspond to the Cu3Sn IMC layer containing trace amounts of other compounds such Cu6Sn5 or Fe–Sn IMCs. Whereas, the IMC layer with the thickness of 50 µm generated at the upper Fe side (position 1) correspond to the FeSn2 IMC layer with trace amounts of other compounds such FeSn or Cu–Sn IMCs. At positions 2–6, it is very difficult to confirm the IMCs based on the present EDX data because the concentrations of Sn, Cu and Fe do not conform to specific IMCs. It is likely that several IMCs coexisted at these positions and in other areas as well. The Cu/Sn and Fe/Sn reactions that are believed to have produced these IMCs (excluding the dissolution of Fe or Cu) are:
| \begin{equation} \text{6 Cu} + \text{5 Sn} \to \text{Cu$_{6}$Sn$_{5}$}, \end{equation} | (1) |
| \begin{equation} \text{3 Cu} + \text{Sn} \to \text{Cu$_{3}$Sn}, \end{equation} | (2) |
| \begin{equation} \text{Cu$_{6}$Sn$_{5}$} + \text{9Cu} \to \text{5Cu$_{3}$Sn}, \end{equation} | (3) |
| \begin{equation} \text{Fe} + \text{Sn} \to \text{FeSn} \end{equation} | (4) |
| \begin{equation} \text{Fe} + \text{2 Sn} \to \text{FeSn$_{2}$}. \end{equation} | (5) |
The volume shrinkages accompanying the above reactions were calculated based on the relationships between the volumes of reactants and products, as shown in Table 2. The shrinkages associated with the formation of Cu6Sn5, Cu3Sn, FeSn and FeSn2 were determined to be 5.13, 24.89–27.4, 4.73 and 3.66%, respectively, and so the value for Cu3Sn was much higher than those for Cu6Sn5, FeSn and FeSn2 because the density of Cu3Sn is much higher than those of its component metals. In addition, variations in temperature will affect the volume changes of these materials via thermal expansion and shrinkage to different extents. This effect would have contributed to the volume shrinkages of the various interfacial IMCs during cooling, and so mismatches in both the volume shrinkage accompanying the formation of IMCs and their coefficients of thermal expansion could have promoted void formation. Therefore, inhibiting void formation when using the super-spread wetting technique to join dissimilar metals is necessary to form homogeneous IMCs at the interface. A joining method that satisfies this condition is described in the following section.
A 10 × 20 × 2-mm Cu substrate and a 10 × 20 × 2-mm Fe substrate were prepared by polishing with abrasive paper (#600 and #1000) and cleaned by ultrasonic agitation in ethanol. Subsequently, a mixed oxide powder was used to produce a surface fine crevice structure capable of promoting even super-spread wetting. The reduction-sintering of a mixture of oxide powders was expected to form a fine structure having a uniform composition and morphology on the different substrate materials. CuO (98% purity) and Fe2O3 (95% purity), both of which are stable oxides of the substrate materials, were selected for the preparation of the powder slurry. Both powders have particle sizes ranging from approximately 1 µm to 3 µm and both have a spherical particle morphology. The CuO and Fe2O3 were ground using a mortar and pestle for 15 min after being mixed at a molar ratio of 2:1, because this ratio gave a Cu:Fe molar ratio of 1:1. The CuO–Fe2O3 powder mixture was then added to a 30 mass% ethanol solution to produce a slurry that was applied to 5 × 10-mm regions of the Cu and Fe substrates via doctor blading. Prior to this process, a 100-µm-thick metal frame with a 5 × 10-mm hole was placed on the substrate surface, after which the hole was filled with the CuO–Fe2O3 slurry and evenly flattened by sliding a metal blade over the surface of the frame to remove excess slurry. The samples were then sintered and simultaneously reduced by heating at 973 K for 1 h under a H2 flow rate of 20 mL/min.
3.1.2 Joining of Cu and Fe substratesIn these trials, we investigated Cu–Fe joining via super-spread wetting through a sintered Cu–Fe layer assuming that the same IMC was formed on each substrate. Figure 1 also illustrates the sample preparation for the trials in this section. A CuO–Fe2O3 slurry was applied to an 11 × 10-mm region of a 15 × 20-mm Cu substrate while one entire side of a 10 × 5 mm Fe substrate was coated, both via doctor blading. The samples were treated by reduction-sintering under the same conditions described in Section 3.1, after which the Fe substrate was placed over top of the Cu substrate so that the sintered Cu–Fe layers were in contact. An approximately 0.1 g of Sn was applied to the sintered layer of the underlying Cu substrate. This joining experiment was carried out applying the same conditions as the Cu–Fe joining experiment with laser irradiated structures, as described above.
3.2 Results and discussion 3.2.1 Surface fine crevice structures by the reduction-sintering of a mixture of CuO and Fe2O3 powders on Cu and Fe substratesX-ray diffraction (XRD) was employed to conduct the phase identification of the initial powder mixture and the samples after reduction-sintering. Figure 6(a) presents the XRD pattern of the mixture, and shows that all peaks clearly matched those expected for CuO and Fe2O3. The patterns obtained from the sintered layer on the Cu (Fig. 6(b)) and on the Fe (Fig. 6(c)) indicate the absence of metal oxides. The CuO–Fe2O3 mixture applied to the Cu and Fe substrates was therefore completely reduced to metallic Cu and Fe in the reducing atmosphere. Figure 7 shows cross-sectional SEM images and EDS maps of the sintered layers obtained on the Cu and Fe substrates. The SEM images confirm that the Cu–Fe sintered layer adhered well to both the substrates. The sintered layer having a thickness of approximately 100 µm displays a large number of micropores with pore sizes ranging from approximately 0.3 µm to 4.4 µm connected to one another within a sintered Cu–Fe particle network. In addition, the EDX mapping results indicate that both Cu and Fe particles were uniformly distributed within the sintered layers. The sintered layers created on the Cu and Fe substrates exhibit uniform chemical composition distributions and the same morphologies. Therefore, it was predicted that the same chemical reaction for forming IMCs occur on the different metal substrates during the joining process through the surface fine crevice structures that was produced by reduction-sintering of the CuO–Fe2O3 powder mixture.
XRD patterns obtained from (a) the as-prepared CuO–Fe2O3 powder mixture and the sintered layers on (b) Cu and (c) Fe substrates.
Cross-sectional SEM and EDX images of the surface fine crevice structures produced by the reduction sintering of CuO–Fe2O3 powder mixtures on the (a) Cu and (b) Fe substrates.
Figure 8 shows the sample after the joining of the Cu and Fe substrates using the sintered Cu–Fe layer. It is apparent that the Sn completely spread throughout the entire sintered region, indicating that super-spread wetting occurred via the surface fine crevice structure produced by the reduction-sintering of the CuO–Fe2O3 powder mixture. Figure 8(b) confirms that the Cu and Fe substrates were securely joined and did not separate even when the sample was lifted vertically. Figure 9(a) provides an image of a cross-section of the joint demonstrating that the Sn completely penetrated and filled the capillary pore spaces of the sintered Cu–Fe layers by the super-spread wetting, resulting in well-joining with almost no void formation. The elemental mappings of EDX data in Fig. 10 establish that both Cu and Fe were uniformly distributed between the Fe and Cu substrates. To permit a more detailed assessment, EDX spot analyses were carried out at the eight positions marked in Figs. 9(a)–(c), and the resulting elemental concentrations are summarized in Table 3. From the results at positions 1 to 6, it is evident that the compositions did not correspond to the specific IMCs expected from Cu–Sn and Fe–Sn systems based on the phase equilibria in a Cu–Sn–Fe ternary system (such as Cu6Sn5, Cu3Sn, FeSn and FeSn2). In addition, a very fine morphology in the area between the substrates can be observed in Fig. 9(a). Therefore, the joint between the two substrates is believed to have been composed of a combination of a liquid Sn alloy and various IMCs. Moreover, the very fine morphology in the joint area demonstrates that the IMCs generated in the Cu–Sn and Fe–Sn systems did not agglomerate with same IMC groups but rather were evenly distributed throughout the central region of the joint. A comparison of the EDX analysis results with the contrast differences in the high-magnification SEM images in Figs. 9(b) and (c) confirms that the IMCs Cu3Sn and FeSn2 were generated respectively in the vicinities of the Cu side (position 7) and Fe side (position 8). It should be noted that the thickness of the Cu3Sn layer was about 4 µm, so this layer was significantly reduced (by approximately 92%) compared with the layer obtained using the laser. This large decrease in the thickness of the Cu3Sn layer is attributed to the presence of Fe near the Cu substrate, as discussed below.
Photographic images showing (a) the sample after joining Cu and Fe substrates having surface fine crevice structures produced by reduction sintering of a CuO–Fe2O3 powder mixture and (b) the same sample inverted.
(a) A cross-section of the joint between the lower Cu and upper Fe substrates having fine crevice structures produced by reduction sintering of a CuO–Fe2O3 mixture. Magnified images of (b) the Cu substrate side and (c) the Fe substrate side at the joining interface.
A cross-sectional SEM image of the joint between Fe and Cu substrates having surface fine crevice structures produced by reduction sintering of a CuO–Fe2O3 mixture and EDX elemental maps for Sn, Fe and Cu.
First, it is helpful to consider the formation of IMCs near the Cu substrate in the case that Fe is not present in this region. That is, in the case that the Cu–Fe joint is formed via super-spread wetting through a surface fine crevice structure produced by laser irradiation. As the liquid Sn spreads into the surface fine crevice structure and the space between the two substrates during the joining process, the Cu in the surface fine crevice structure and in the substrate dissolves in the liquid Sn. The Cu–Sn binary phase diagram10) indicates that this dissolution increases the Cu concentration in the liquid Sn alloy until approximately 75 at% Cu is obtained at 773 K, where liquid Sn–Cu alloy is equilibrated with Cu3Sn, representing the formation of Cu3Sn. In contrast, if Fe exists near the Cu substrate (meaning that the Cu–Fe joining using the surface fine crevice structures were produced by the reduction-sintering of a CuO–Fe2O3 powder mixture), the Cu–Sn binary phase diagram,10) Fe–Sn binary phase diagram11) and Cu–Sn–Fe ternary phase diagram16) must all be considered. Based on these phase diagrams, it is predicted that Fe would rarely dissolve in the liquid Sn at 773 K to form FeSn2 and FeSn owing to low solubility of Fe in the liquid Sn alloy. At a high Sn concentration, a liquid Sn–Cu alloy with a small amount of Fe would therefore coexist with FeSn2. Then, a region containing three phases, in which the liquid Sn–Cu alloy, FeSn2 and FeSn are all equilibrated, appears when decreasing the amount of Sn. Additional reductions in the Sn concentration result in the appearance of the liquid Sn–Cu alloy, FeSn and Cu3Sn. Based on the above, it is thought that the formation of FeSn and FeSn2 suppresses the formation and growth of Cu3Sn as the liquid Sn spreads through the surface fine crevice structure produced by the reduction-sintering process. Hence, the differences in volume shrinkage and thermal expansion between the IMCs become negligible in conjunction with the reduced thickness of the Cu3Sn layer, which inhibits void formation at the joint.
The dissimilar metals Cu and Fe were joined via the super-spread wetting of liquid Sn through surface fine crevice structures. Two methods were applied to produce the surface fine crevice structures: laser irradiation and the reduction-sintering of a CuO–Fe2O3 mixture on the Cu and Fe surfaces. The main conclusions are as follows.
The above conclusions show that the reduction-sintering of a metal oxide mixture is superior to the laser irradiation method with regard to producing reliable joins between dissimilar metals.
