2019 Volume 60 Issue 4 Pages 602-610
The cold spray method is a coating process that should improve the wettability of solder to aluminum surfaces. In this study, Cu powder particles were deposited on commercially pure Al via cold spraying, and the characteristics and microstructural developments of the deposited layer, interface, and substrate were investigated. Cu coatings, with an electronic resistivity similar to that of commercial solder, could be formed on the Al substrate. Solid-state bonding was partially achieved between the Cu particles. Interdiffusion occurred at the interface between the cold sprayed Cu particles and the Al substrate, forming the reaction phases, and the Cu coating was bonded to Al substrate along almost the entire bond interface. The impact of high-velocity particles induced dynamic recrystallization and grain refinement of Al, resulting in increased hardness near the surface of the Al substrate. The thickness of the hardened region was ∼10 µm. Tin-based solder paste exhibited good wettability to the cold-sprayed Cu coating.
The thermal spraying method is a surface coating process, in which powder particles in a molten state are sprayed onto a substrate to form a deposition layer. Cold spraying is a lower-temperature process than thermal spray coating. The metal powder particles are heated to a temperature below the melting point and collided against a substrate at high velocity of several hundred meters per second or more to form a deposition layer owing to the kinetic energy of the particles.1–4) Since this process is believed to be a solid-state process, various dissimilar combinations of powder particles and substrates can be selected, such as Ti particles/Al substrate,5) Cu particles/Al substrate,6) Al particles/Cu substrate, or Al particles/stainless steel substrate.7) In addition, recently, not only metallic powders but also ceramic powders have been deposited via cold spraying.8) In particular, many studies on Cu deposition have been reported because the cold spray method can easily form a Cu coating on various substrates in air. Li et al.9) reported that a fully dense Cu coating was obtained by using a mixed feedstock of dense and porous Cu particles. Kang et al.10) reported that a high-strength deposition layer was realized by using micro-alloyed Cu particles containing small amounts of Fe, Si, and Ag. Since a thick coating can be obtained via cold spraying, additive-manufactured bulk Cu was fabricated on an Al substrate, and the characterization and anisotropy of the layer were reported.11) A Cu deposition layer with electrical resistivity of ∼11 µΩ cm could be formed on a polyvinyl chloride polymer substrate.12) Ni–Cu hybrid electrodes, which can be used as substitutes for Ag electrodes, were formed on a Si wafer via cold spraying.13) In terms of the bonding mechanism between the deposition layer and substrates, interdiffusion on the order of ∼1 µm or less occurred across the Cu/Al interface.14)
Al is often used in electric devices because of its superior electric and thermal conductivities. Since it is difficult to solder to Al, Ni or Ni–P plating, which is a wet coating process, is usually applied to Al surfaces to improve its solderability in the field of electronics packaging. However, at the manufacturing site, a dry coating process, which is an alternative to Ni plating, is needed to realize a more streamlined manufacturing process of electronic devices. The cold spraying of copper is one candidate for a dry surface coating method because the commercial solder pastes exhibit good wettability to copper. Thus, the feasibility of cold spraying of Cu onto Al as a dry surface coating process for electronic devices must be investigated. In the cold spraying process, as the metal particles collide against the substrate at high velocity, the surface of the substrate is deformed with a high strain rate, and the temperature at the impact points may approach the melting point owing to the kinetic energy of the particles.15) Al parts used in electronic devices are small and often used as thin foils, films, and wires. Therefore, for cold spraying on Al used in electronic devices, it is important to evaluate the work-deformed region of the aluminum caused by the collision of Cu particles and the interfacial strength between the cold sprayed coating and the Al substrate and the physical properties of the coating.
The purpose of this study is to investigate the feasibility of the cold spraying of copper as a dry coating processes for aluminum used in electronic devices. The physical properties of the Cu coating, the microstructural development of the coating layer and substrate, and bonding states at the interfaces were investigated.
Gas-atomized spherical copper powder particles, 10 µm and 20 µm in size, were deposited on a 1-mm-thick commercially pure aluminum AA1050 substrate via the cold spray process. The aluminum substrate was annealed at 623 K for 3.6 ks, immersed in 5% H2SO4 aq. for 30 s, and subsequently cleaned using acetone before being subjected to the cold spray process. The cold spraying was carried out using a Dymet 413 gun-head (OCPS) and an ACGS controller (Medicoat) with air as the propellant gas heated to 723 K. The cold spraying process and the stencil masks are schematically shown in Fig. 1. The rectangular- and dumbbell-shaped stencil masks shown in Fig. 1(b) and (c) made of SUS304 (0.1 mm-thickness) were used to make specimens for the measurement of electric resistivity and tensile strength of the Cu coatings, respectively. The divergent section length of the nozzle was 120 mm, the inlet diameter was 4 mm, and the outlet diameter was 4.8 mm. The gas pressure was varied from 0.45 to 0.73 MPa. The standoff distance between the nozzle and the Al substrate was 12 mm. The other cold spraying conditions are listed in Table 1. These conditions were determined through preliminary experiments. To increase the thickness of the coating layer, six-pass depositions were carried out using Cu powder with a particle size of 20 µm. The temperature was measured during the spraying process using a thermocouple on the back surface of the Al substrate directly under the sprayed line.
Schematic illustration of cold spray process of copper particles on aluminum substrate, (a), and SUS304 stencil masks for measurement of electric resistivity, (b) and tensile strength, (c). (unit: mm)
The Al substrates with Cu coatings were immersed in NaOH aq. to dissolve the Al substrate, leaving behind only the Cu coatings, so their physical properties could be evaluated. The density of the Cu coating was evaluated by the Archimedes method. The ratio of the density of the Cu coating to that of bulk Cu (8.94 g/cm3) is defined as the packing density in this study. The electric resistance of the Cu coatings was measured by the four-terminal method using a Micro-ohm meter (Agilent 34420A) as shown in Fig. 2. The measurement current was 1.0 mA and the length of electric potential measurement (L) was varied from 10 to 30 mm for each specimen. The electrical resistivity was calculated in terms of the measured electric resistance, measurement length (L), and cross-sectional area. The tensile strength of Cu coatings extracted from Al substrates was evaluated by tensile tests. Tensile tests of five specimens were carried out for each set of cold spraying conditions at a crosshead speed of 0.01 mm/s. When the fracture occurred outside the evaluation area in the test pieces, those results were eliminated. The cross sections of the Cu coatings showed a domed shape that was not ideal for evaluating tensile strength. However, since the extracted Cu coatings were very thin and brittle, as described later, it was difficult to apply further machining. Therefore, the tensile strength of the Cu coatings in their as-sprayed shape was measured. The tensile strength and electric resistivity of the Cu coatings were calculated by using the cross-sectional area observed by scanning electron microscopy.
Set-up of measurement of electric resistance of Cu coating by four terminal method. The length of electric potential measurement (L) was varied from 10 to 30 mm.
Heat treatment was carried out at 673 K for 2.5 ks to estimate the initial bonding state at the interface between the Cu coatings and Al substrate.
The cross-section was polished using emery paper, and some of them were finished using a cross-section ion polisher (JEOL, IB-19510CP). The microstructures of the cross-sections were observed using an optical microscope, a field-emission scanning electron microscope (FE-SEM, JSM-7100F, JEOL), and a scanning transmission electron microscope (STEM) combined with energy-dispersive X-ray spectroscopy (EDS). The microstructures were also evaluated using electron back scatter diffraction (EBSD) patterns.
The hardness of the Al substrate near the interface between the Cu coating and Al substrate was measured using an ultra-micro hardness tester (DUH-211S, Shimazu Co., Ltd.) with a Berkovich-type indenter under a load of 4.9 × 10−3 N. The hardness was evaluated using the indentation size after unloading and this hardness is denoted as HT115 in the present study.
To perform reflow soldering, commercial Sn–5 mass%Sb solder paste was applied on the Cu layer and heated at 553 K for 300 s under reduced pressure. The microstructure of the solder/Cu coating interface was observed by FE-SEM.
In order to observe the adhesion state of Cu in the first layer formed via single-pass deposition, 10-µm Cu particles were sprayed onto the Al substrate under various gas pressure conditions. Figure 3 shows the cross-sections of Cu coatings deposited on the Al substrate at different gas pressures. When the gas pressure was 0.45 MPa, a coating of thickness 5–10 µm was formed. With increasing gas pressure, the coating became thicker, and when the gas pressure was 0.73 MPa, the layer thickness reached ∼25 µm. In the cold spray method, particles are ejected from a nozzle at a velocity higher than the velocity of sound and they directly impact the substrate. Particles impacting the substrate at high velocity are deformed from their original spherical shape into a flat shape, and the substrate is also plastically deformed.4) Since the initial particle diameter was 10 µm, several copper particles could be piled up in the longitudinal direction even in the first Cu layer formed by single-pass deposition with a gas pressure of 0.6 or 0.73 MPa. The detailed microstructures of the Cu coating layer will be shown later. On the surface of the Al substrate, wavy irregularities corresponding approximately to the diameter of the Cu particles were formed owing to the impact of the particles.
Cross sections of single-pass deposited Cu on Al substrate using Cu particles of 10 µm in particle size with different gas pressures. (a) 0.45 MPa, (b) 0.6 MPa, (c) 0.73 MPa.
The Cu coatings should be thick in order to evaluate their electrical resistivity, strength, and packing density. Therefore, multi-pass depositions were performed using Cu powder with a particle size of 20 µm. Figure 4 shows the cross-sections of the six-pass depositions performed under different gas pressures. With the increase in the gas pressure from 0.45 to 0.6 MPa, the thickness of the Cu coating increased from 150 to 600 µm. Only the Cu coating was extracted from the Al substrate via the method described in the experimental section, and the density, electric resistance, and tensile strength of the coating were measured. The density, electric resistivity and thickness of the Cu coatings at the center are shown in Table 2. The Cu coatings sprayed under all of the gas pressure conditions showed electric conductivity. Regardless of the gas pressure, Cu coatings with electric resistivity in the range of 1.2 × 10−7 to 3.3 × 10−7 Ωm were obtained. Although the electric resistivity of the Cu coatings was ∼10 times larger than that of bulk Cu, it is almost equal to that of commercial tin-based lead-free solders. Many voids remained in the Cu coating sprayed at a gas pressure of 0.45 MPa (Fig. 4(c)) and the packing density was 72%. The packing density drastically increased as the gas pressure was increased from 0.45 to 0.73 MPa and reached approximately 90% at higher gas pressure. In the multi-pass depositions, the lower region of the deposition was dense, and the upper region had a porous structure because the cumulative collision energy of the particles was larger at the lower region than at the upper region of the deposition. The lower the gas pressure, the more remarkable was the distribution tendency of the voids. In addition, since the gas flow stagnates in the vicinity of the stencil mask edge, the deposition efficiency probably decreases there, which causes a decrease in the interfacial strength between the Cu particles. Therefore, the tensile strength of a Cu coating could have a distribution in the thickness and width directions of the coating. However, as described in the experimental section, the strength of Cu coatings in their as-sprayed shapes was evaluated in this study. The tensile strength of the Cu coatings and a typical tensile stress–displacement curve are shown in Fig. 5. The strength of the Cu coatings increased with increasing gas pressure, and it reached ∼160 MPa when the gas pressure was 0.6 MPa. The change in fracture strength as a function of gas pressure exhibited almost the same trend as the change in packing density. The tensile strength of 160 MPa corresponded to 72% of the tensile strength of annealed copper (220 MPa).16) However, the Cu coatings were fractured with almost no plastic deformation as shown in Fig. 5(b). A typical fracture surface of a Cu coating is shown in Fig. 6. The tensile specimen exhibited a brittle fracture surface and the fracture passed through the interface between the Cu particles. The Cu particles deformed by their own collision were laminated, and many detachments between the Cu particles and voids were observed on the fracture surface.
Cross sections of six-pass deposited Cu coating using ϕ20-Cu particles under different gas pressures. (a) 0.45 MPa, (b) 0.6 MPa. Images (c), (d) and (e) show highlighted regions shown in images of (a) and (b).
Results of tensile test of Cu coating. (a) tensile strength against gas pressure, (b) typical tensile strength-displacement curve.
Typical fracture surface of Cu coating sprayed under gas pressure of 0.6 MPa.
The tensile strength and electrical resistivity of the Cu coatings were inferior to those of bulk Cu, which is probably due to the unsound bonding state of the interfaces between Cu particles. Figure 7 shows the microstructure of a Cu coating that was deposited under a gas pressure of 0.73 MPa. The deformed and flattened Cu particles were laminated, and many gaps and voids remained between Cu particles. The ends of the deformed particles were finely folded and overlapped. At the interfaces between the Cu particles, inclusions with dark contrast, which might be oxides on the surface of the Cu particles, were observed (Fig. 7(b) and 7(c)). Those unsound bond interfaces decreased the electric resistivity and tensile strength of the Cu coating. On the other hand, there were some regions where the boundary lines between Cu particles were ambiguous. As shown in Fig. 7(c), the boundary between Cu particles was not visible on the extended line of the interface where an inclusion was present. Solid-state bonding might be achieved there. The collision velocity required for the deposition of Cu particles on the substrate depends on the particle size and gas pressure, and it has been reported that it can reach approximately several hundred meters per second.4) The Cu particles are plastically deformed by high-velocity collisions, thereby creating a newly formed surface. The newly formed surfaces are bonded in the solid state under high pressure. Sound interfaces were partially formed where Cu particles were probably bonded in the solid state. Therefore, the strength and electrical resistivity of the Cu coatings were dependent on not only mechanical bonding by mechanisms such as interlocking, but also solid-state bonding between copper particles.
Microstructures of Cu coating sprayed under the gas pressure of 0.73 MPa. (a) Macroscopic cross section, (b) Highlighted area in (a), (c) Highlighted area in (b).
The microstructure of the interface between the Al substrate and the Cu coating deposited under a gas pressure of 0.73 MPa was observed using STEM to evaluate the interfacial bonding mechanism. The microstructure of the interface and the results of semiquantitative analyses using EDS are presented in Fig. 8 and Table 3, respectively. Points 1 and 2 shown in Fig. 8 are Al and Cu, respectively. Reaction phases several tens of nanometers in size were observed at the interface (see Point 3). The composition of Point 3 was 39.1 at%Al–60.9 at%Cu, which probably corresponds to Cu1.5Al, Cu9Al4,17,18) or a non-equilibrium phase. The change in temperature during the first pass under a gas pressure of 0.73 MPa was measured on the back surface of the Al substrate. Two measurement results are shown in Fig. 9. The temperature increased rapidly and reached ∼353 K immediately below the nozzle. The temperature gradually decreased after the nozzle passed. On the back surface of the 1-mm-thick Al substrate, the temperature increased only to 353 K, which might be insufficient to cause reaction diffusion. However, numerical simulations showed that the temperature at the interface reached 723 K and 793 K for the impact cases of Al particle/Al substrate at 780 m/s and Ti particle/Al substrate at 700 m/s, respectively.5) In addition, the interface exhibited a wave form as shown in Fig. 3; moreover, mechanically mixed and locally folded interfaces were observed as shown in Fig. 8. It has been reported that many intermetallic compounds and non-equilibrium phases can be formed via mechanical alloying in the Al–Cu binary system.19) In the cold spraying process, mechanical alloying could also form intermetallic compounds or non-equilibrium phases at the interface. Therefore, the Cu particles and Al substrate were bonded in the solid state via reaction diffusion, mechanical alloying, or both mechanisms.
Microstructures of the interface between Cu coating and Al substrate. The gas pressure of cold spray is 0.73 MPa. (a) TEM bright field image, (b) scattered electron image of highlighted area in (a). Chemical compositions at points 1 to 3 are shown in Table 3.
Typical change in temperature on the back surface of Al substrates during the first pass. Two measured data are shown when gas pressure is 0.73 MPa.
The Cu and Al were partially bonded due to reaction diffusion as shown in Fig. 8. However, it is difficult to evaluate the bond area fraction against the whole interface in the as-sprayed specimens since the initial reaction phases were too tiny to be observed by SEM. Then, heat treatment was performed for growing the reaction phases at the interface, and the bonding area in the as-sprayed specimen was estimated. The Al–Cu intermetallic compounds should grow as a result of the heat treatment at the sites where the oxide film of aluminum is broken due to the collision of copper particles and aluminum and copper are bonded to each other.
Figure 10 shows the growth behavior of the intermetallic compound layer at the Cu/Al interfaces before and after the heat treatment at 673 K for 2.5 ks. Apparent reaction phases were not observed during SEM investigations of the Cu/Al interface before the heat treatment. After the heat treatment, the reaction phases grew in most regions of the interface (Fig. 10(b)). The composition of the reaction layer was approximately 60 at%Al–40 at%Cu, which probably corresponded to Al2Cu and/or AlCu. Furthermore, there were some regions where intermetallic compounds did not grow (Y-Y′ line in Fig. 10(d)). The Cu particles and the Al substrate were not bonded sufficiently at these regions before the heat treatment. However, the intermetallic compound layer grew across ∼90% of the total length of the interface, which indicates that sound bonds were achieved across 90% of the total interface even in the as-sprayed specimen.
Growth of intermetallic compound layer between Cu coating and Al substrate by heat treatment at 673 K for 2.5 ks. Gas pressure: 0.6 MPa. (a) SEM image before heat treatment, (b) SEM images after heat treatment, (c) Detail of interface before heat treatment, (d) Detail of interface after heat treatment, and (e) Composition of Al and Cu on line X-X′ and Y-Y′ shown in (d).
As shown in Fig. 3 and Fig. 4, the surface of the Al substrate was deformed into a wavy shape owing to the collisions of Cu particles at extremely high velocity. In electronic devices, aluminum is often used in the form of foils, films, and wires. It is therefore important to evaluate the region of aluminum deformed due to the collisions of particles. It has been reported that even a stainless steel substrate is work-hardened at the region within 100 µm of the interface owing to the collisions of Cu particles in the cold spray method.20) The hardness (HT115) distribution in the Al substrate near the interface was measured and the results are shown in Fig. 11. The hardness increased within 10 µm of the interface. No significant difference in the hardened region was observed for different gas pressure conditions. Kim et al. reported that dynamic recrystallization and grain refinement were induced in titanium particles owing to the impacts at high velocity.21) Then, the crystal orientation of the Al substrate near the interface was investigated using EBSD to evaluate the microstructural development in the hardened region. Figure 12 shows the inverse pole figure (IPF) maps near the Cu/Al interface obtained under different gas pressures. The grain size of the original Al substrate was several tens of micrometers. Moreover, fine grains several micrometers in size were formed in the Al substrate near the interface. Plastic deformation and the increase in temperature owing to the collisions of Cu particles at high velocity caused dynamic recrystallization, resulting in grain refinement. Noisy signals were also detected in the recrystallization area, which indicated that a large amount of strain was introduced in this region. The recrystallization area with residual strain corresponded to the hardened region. Clear EBSD patterns were not detected in the Cu coating near the interface because the Cu particles were also plastically deformed and a large amount of strain was introduced in the particles by the cold spray process.
Hardness (HT115) of aluminum near the interface between Cu coating and Al substrate.
IPF maps showing crystal orientation near interface between Cu coating and Al substrate with gas pressure of (a) 0.45 MPa, (b) 0.6 MPa and (c) 0.73 MPa.
Subsequently, heat treatment was carried out to relieve the strain introduced in the aluminum substrate. Figure 13 shows the IPF maps near the interface between the deposited Cu and the Al substrate after heat treatments at 473 K and 673 K for 3.6 ks. Both Al and Cu grains grew only a little in the hardened region of the Al substrate and in the Cu coating, respectively, during the heat treatment at 473 K. Strain seemed to remain in the Cu coating and the Al substrate. After heat treatment at 673 K, both the Al and the Cu coating were almost fully recrystallized and coarse grains were observed.
SEM images and IPF maps near the interface between Cu coating and Al substrate after heat treatment at 473 K, (a) (b), and at 673 K, (c) (d) for 3.6 ks. The cold spraying was carried out under the gas pressure of 0.6 MPa. The arrows show Al–Cu intermetallic compounds.
To investigate the feasibility of soldering on the Cu coating, the reflow soldering of Sn–5 mass%Sb paste onto the Cu coatings was carried out. The cross-section of Sn–5 mass%Sb solder on the Cu coating is shown in Fig. 14. Although some pores remained in the solder near the interface between the solder and the Cu coating, the wetting angle of solder to the Cu coating was acute and the solder exhibited good wettability against the Cu coating, forming intermetallic compounds at the interface. However, the molten solder did not completely penetrate the voids or gaps between the Cu particles near the surface of the coating (Fig. 14(c)). The intermetallic compound layer was formed in contact with the gaps between the Cu particles. In general, the dissolution rate of Cu into lead-free Sn-based molten solder is higher than that in molten Pb–Sn eutectic solder.22) The dissolution of the Cu particles in contact with the molten Sn–5Sb proceeds, while the intermetallic compounds were formed at the interface immediately. The penetration of molten solder into the narrow gaps between the particles might be prevented by the formation of intermetallic compounds closing the gaps, even under reduced pressures. However, the formation of gaps between Cu particles was limited and the number of voids could be decreased by increasing the gas pressure of the cold spraying process, as shown in Fig. 4.
Microstructures of Sn–5 mass%Sb solder on Cu coating. (a) Cross section. Images of (b) and (c) show highlighted areas in (a).
As a dry coating method to replace Ni plating on aluminum used in electronic devices, the feasibility of cold spraying of Cu particles on aluminum was investigated in terms of the characteristics of the Cu coating, work hardened thickness of the Al substrate, and the interface between Cu and Al. The following results were obtained.
This work was partially supported by the Grant-in-Aid for Scientific Research (B), MEXT KAKENHI Grant Number 18H01723. The authors express their gratitude to Startack Co., Ltd. for their experimental support.
