Investigation on Interface of CuW/Al Composite Using Ni Interlayer by Vacuum Hot-Pressing Diffusion Bonding

Abstract

In this study, the bonding of a CuW/Al composite with a Ni interlayer was designed and established by vacuum hot-pressing diffusion bonding. The interfacial microstructure was systematically discussed based on experimental characterization and first-principles calculations. The result indicated that the interface consisted of intermetallic compounds (IMCs) of Al₃Ni₂ and a few of Al₃Ni. The interfacial microstructure significantly differed from the interface without the Ni interlayer. The growth kinetics of the Al₃Ni₂ layer followed a parabolic behavior, which was mainly affected by the volume diffusion mechanism. The interfacial thickness decreased significantly, and the average thickness was ~35 μm. The microstructural evolution revealed that Al₃Ni₂ was the phase that was formed first. By introducing a Ni interlayer, the interfacial strength was significantly enhanced due to the IMCs that were changed from Al–Cu and Al–W IMCs to Al–Ni IMCs. The maximum shear strength reached 90.9 MPa, which was increased by 76% compared to that of the sample without the Ni interlayer. An analysis of the fracture morphology analysis showed that the crack was prone to exist at the Al₃Ni₂/Ni interface and presented a cleavage fracture characteristic.

1. Introduction

Dissimilar metal composites have attracted much attention because these composites offer excellent comprehensive properties due to the combined advantages of each metal. These composites have been extensively utilized as the key components in electronics, aerospace, and defense fields [1,2,3]. For example, the CuW/CrCu composite has been successfully applied as the dominated contact materials in high voltage breakers, as it possesses the combined advantages of high thermal and electrical conductivity of CrCu, as well as the high hardness and excellent arc erosion resistance of CuW [4,5,6]. Nevertheless, the bottleneck, due to the scarcity of Cu resources and demand for cost reduction, restricts the practical applications of the CuW/CrCu composite in a high-voltage switch. To solve this problem, a systematic exploration of an alternative material to CrCu alloy is urgently required to reduce the cost of contact materials and enhance the market competitiveness. Notably, Al and Al alloys have been widely used as structural material in automation, aerospace, and the electronics industry, owing to their high specific strength, excellent electrical conductivity, and corrosion resistance [7,8]. Moreover, Al and Cu share the same crystal structure, i.e., face-centered cubic (fcc); have a similar hardness (Al: 35 HBS, Cu: 37 HBS); and share the same electrical conductivity (Al, 65% IACS; Cu, 103% IACS) [9]. Consequently, it may be considered an effective substitute material for CrCu alloy in the CuW/CrCu composite to obtain the CuW/Al composite, which can be used as part of the contact materials, thus reducing the cost and promoting the sustainable development of the electric power industry. Therefore, the development of a CuW/Al composite is highly desirable and urgently required.

Dissimilar metal composites have been successfully prepared by many technologies, in particular, the binary system such as Al–Fe [10], Al–Cu [11], and Al–Mg [12]. These methods can be divided into two categories. The first category includes solid–liquid bonding methods, such as compound casting [13] and transient liquid phase bonding [14]. The second one includes solid–solid bonding methods, such as diffusion bonding [15], friction stir welding [16], and vacuum hot-pressing diffusion [17]. However, for the CuW/Al composite, one base is CuW with a refractory nature, and the other is Al with a low melting point. The CuW alloy consists of two incompatible phases of Cu and W, in which W is distributed as a skeleton structure [18]. Owing to these characteristics, the bonding of CuW and Al is significantly different from the bonding in the abovementioned simple binary systems. Al and Cu exhibit a strong chemical affinity due to their similar physical and chemical properties [19], while Al and W have a great difference in their crystal structure, melting point, and thermal expansion coefficient [20,21]. Intermetallic compounds (IMCs) such as AlCu, Al₄Cu₉, and Al₂Cu were prone to generate at the interface during solid–liquid bonding. These IMCs are hard and brittle and, thus, not conducive to achieve a reliable interfacial bonding. The W is too hard to be wetted, so a metallurgical bonding cannot be established during solid–solid bonding. Based on these reasons, it was difficult to obtain a CuW/Al composite with sound interface by traditional solid–liquid or solid–solid method. In our previous studies, the CuW/Al composite was creatively fabricated by a multiple processing technique. That is, the CuW alloy was pretreated with the surface of the W skeleton structure, infiltrated with liquid Al, and then combined by vacuum hot-pressing diffusion bonding [22]. The results showed that three transition zones involving five types of Al-based IMCs formed at the CuW/Al interface. An effective metallurgical bonding was achieved by the multiple processing method, but the interfacial strength was low (59.8 MPa). This was mainly due to the following reasons. IMCs are usually brittle and hard due to their complex crystal structure, so the formation of various IMCs would weaken the interfacial bonding; also, the thickness of the interface was so large (1048 μm) that it resulted in a low interfacial strength. In fact, due to the high activity of Al, it can react with almost all metal elements and form Al-base IMCs. Therefore, it becomes difficult to completely suppress the formation of IMCs when Al is bonded with other metals.

Many investigations proved that the addition of a suitable interlayer can effectively enhance the interfacial strength of dissimilar metal composites [23,24]. The main objectives are to transform the reaction path, improve the wettability, relieve stress concentration, and change the types and morphology of IMCs in order to strengthen the interfacial bonding. The following factors should be considered when selecting a suitable interlayer. First, the interlayer should exhibit plasticity, and it must be close to the metal base to facilitate the diffusion of elements. Second, the interlayer should have a certain solid solubility in the metal base, and the diffusion of the element should occur relatively fast. Finally, the interlayer should not generate IMCs with the base, or form less brittle IMCs. Many metals, such as nickel (Ni) [25], titanium (Ti) [26], zinc (Zn) [27], copper (Cu) [28], and silver (Ag) [29], have been considered as the constructive interlayers for the bonding in dissimilar metal composites. For example, Yang et al. [30] investigated the influence of Ni interlayer on the interfacial reaction of Al/Mg joint. The results showed that the interfacial strength was improved with a Ni interlayer. Wen et al. [31] successfully prepared a good bonding of TC4/AZ91D bimetallic composites with the application of the Ni interlayer, and the interface was free of defects such as gaps; thus, the interfacial strength was significantly raised from 25.37 MPa to 97.35 MPa.

In this study, the idea of adding Ni interlayer was put forward to improve the strength of the CuW/Al composite, which was attributed to the following reasons. First, Cu Ni elements show good solid solubility, forms an fcc (Cu, Ni) solid solution, and it is free of IMCs; second, Ni-W IMCs do not exist at a temperature below 970 °C according to the Ni-W binary phase diagram; third, the hardness of Al-Ni IMCs was less than Al-Cu IMCs, and this was conducive to interfacial strength. Therefore, the adding of the Ni interface can avoid the excessive formation of IMCs as much as possible and reduce the hardness of IMCs formed at the interface. For those purposes, the CuW/Al composite with a Ni interlayer was designed and prepared by vacuum hot-pressing diffusion bonding. The interface morphology and composition were analyzed, the formation sequence and growth kinetics of IMCs were discussed, and the microstructural evolution was revealed based on characterization and simulating computation. The shear strength of the CuW/Al composite was evaluated to determine the interfacial mechanical property; the facture morphology and characteristics were also investigated. These ideas and achieved results can provide possible new methods to fabricate other bimetallic composites.

2. Experimental and Simulation

2.1. Materials and Characterization

First, a CuW pseudo-alloy was fabricated by using a sintering and infiltration method, and the detailed process is as follows. Cu powders (purity ≥ 99.9 wt.%, 45 μm, purchased from Avimetal Powder Metallurgy Technology Beijing Co., Ltd., Beijing, China), and W powders (purity ≥ 99.9 wt.%, 8 μm, purchased from Xiamen Golden Egret Special Alloy Co., Ltd., Xiamen, China) were selected as the initial powders. The nominal composition of the CuW pseudo-alloy was Cu30W70, indicating that the mass ratio of Cu to W is 3:7 in mixed powders. The powders were then blended in a V-type mixer machine (Wuxi Honghai Equipment Technology Co., Ltd., Wuxi, China) for 4 h with a rotational speed of 340 rpm. The well-mixed binder was pressed into a cylindrical green billet with a size of 21 mm in diameter and 15 mm in height by applying an axial pressure (250 MPa). Next, the green compact was placed in the vacuum furnace (Shanghai Haoyue Vacuum Equipment Co., Ltd., Shanghai, China) to prevent oxidation. Then, the CuW alloy was obtained by sintering and infiltration at 1350 °C for 2 h under a hydrogen (H₂, purity ≥ 99.9 wt.%) atmosphere condition. A Cu block (purity ≥ 99.9 wt.%) was used for infiltrating the W compact. The size was 21 mm in diameter and 10 mm in height.

Second, the CuW/Al composite was fabricated by vacuum hot-pressing diffusion bonding. This process consists of the following steps. First, the Al sheet (purity ≥ 99.9 wt.%, purchased from Xiamen Golden Egret Special Alloy Co., Ltd.) and the as-prepared CuW (30 wt.% Cu) pseudo-alloy were used as the raw materials. They were machined into a cylindrical sample with a size of 20 mm in diameter and 10 mm in height. To remove the oxide film and adhered contaminants, the surface of the CuW and Al cylinder was ground with abrasive paper, and then the surfaces of the samples were etched with 10 wt.% NaOH and 10 wt.% HCl solution (purchased from Xi’an Chemical Glass instrument, Shaanxi, China). Subsequently, the samples were ultrasonically cleaned in an acetone bath, followed by being dried in air. Next, Ni foil (30 μm thick, purity ≥ 99.9 wt.%, purchased by Beijing Guantai Metal Materials Co., Ltd., Beijing, China) was selected as the interlayer. Then, CuW, Ni interlayer, and Al samples were piled into a sandwich structure and placed in a crucible. The CuW and Al were bonded with a Ni interlayer in a vacuum hot-pressing sintering furnace (HVRY-2 type, Shanghai Haoyue Vacuum Equipment Co., Ltd.). The schematic illustration of the bonding of samples is presented in Figure 1a. The preparation parameters are as follows: the bonding was processed at 710 °C under various holding times of 10 min, 20 min, 40 min, and 60 min for observing the microstructural evolution at the interface. The bonding pressure was 20 MPa, and the vacuum degree was 4.0 × 10⁻³ Pa. Figure 1b shows the fabrication process. The sample with the sandwich structure was heated to 600 °C at a heating rate of 20 °C/min. Then, the temperature was increased to 710 °C at a heating rate of 7 °C/min. After the completion of the heat preservation at 710 °C with various times, the temperature was reduced to 620 °C at a cooling rate of 10 °C/min; at this moment, the pressure was applied during the cooling. Next, we turned off the heating system, and the sample was cooled with the furnace to room temperature. Finally, the CuW/Al composite with a Ni interlayer was obtained after furnace cooling.

In order to observe the interface, the samples were machined into a cube with dimensions of 6 mm × 6 mm × 15 mm cubic via a wire-cutting process. Furthermore, to measure the strength of samples, the test samples were cut into cylinders with a diameter of 10 mm and height of 15 mm. The interface specimens were mechanically ground using an abrasive paper with a grade of 1500 and polished using a diamond-polishing agent with a particle size of 1 μm. Then, the specimens were etched using an aqueous solution of hydrofluoric acid (HF, 5 vol.%, purchased from Xi’an Chemical Glass instrument) for 30 s to clearly observe the phases present at the interface.

The microstructure of samples was characterized and by scanning electron microscope (SEM, JEOL JSM-6700F, Tokyo, Japan). The composition analysis of the reaction interface was conducted by energy-dispersive spectroscopy (EDS). The mechanical properties of samples were examined using an automatic testing machine (HT-2402, Shandong Liantai Co., Ltd., Shandong, China), in which the rate of the applied loading was 1.0 mm/min. The shear strength was obtained by using the computer-controlled output. The value of shear strength under each condition was obtained by the average value of three times tests. The fracture location, surface morphology, and microstructure were further investigated by the secondary electron images obtained by SEM.

2.2. First-Principle Calculation

To further investigate the structural stability and formation sequence of IMCs present at the interface, first-principles calculations were conducted by using Cambridge Serial Total Energy Package (CASTEP) software 8.0 version based on the density functional theory [32,33]. The generalized gradient approximation of functional with Perdew–Burke–Ernzerhof (GGA-PBE) [34,35] was used to calculate the exchange correlation between electrons. The crystal structure was optimized by the Broyden–Fletcher–Goldfarb–Shanno (BFGS) method [36,37]. The calculation precision was set to “superfine” grade, and the value of the self-consistent field (SCF) convergence threshold was 5.0 × 10⁻⁷ eV/atom [38], accordingly. First, the type of crystal structure, space group, and coordinate parameters of IMCs were found in the Inorganic crystal structure database (ICSD) database. Then, the crystal structure model of IMCs was established. Next, the convergence test was performed. The results of the test showed that the cutoff energy (Ecut) was set as 400 eV, and the Brillouin zone K points were 7 × 7 × 6 for Al₃Ni₂ and 7 × 6 × 8 for Al₃Ni. The valence electron states were Al 3s²3p¹ and Ni 3d⁸4s² in pseudopotential. The values of convergence limit were set as follows: stress of 0.02 GPa, force of 0.01 eV/Å, energy change of 5.0 × 10⁻⁶ eV/atom, and displacement of 5.0 × 10⁻⁴ Å. Finally, the energy and density of states were calculated separately.

3. Results and Discussion

3.1. Morphology of CuW/Al Composite without and with Ni Interlayer

Figure 2 shows low-magnification SEM micrographs of the CuW/Al composite without and with the Ni interlayer which were prepared at a bonding temperature of 710 °C and holding time of 40 min. The figure illustrates that the CuW base with light gray color is on the left, and the Al base with a dark gray color is on the right. A wide transition zone formed between CuW and Al bases at the interface, marked by a yellow dotted line, accordingly, as shown in Figure 2a. Furthermore, the interface zone was observed to be irregularly connected with CuW and Al, forming a wavy interface.

It indicates the possibility of the existence of disturbance at the front of the interface when the reaction occurred between solid CuW and Al liquid. The microscopic measurement shows that the average thickness of the interface was 340 μm. Figure 2b displays the results of the EDS line scan of the CuW/Al interface without the Ni interface. The red line is Al, blue is Cu, and green indicates W. The EDS composition line of the Al element showed an obvious step when it passed across the interface zone, indicating the occurrence of composition mutation and generation of Al-based IMCs at the interface. The microstructure was comprehensively analyzed in our previous study [22]. The interface consisted of three transition zones from the CuW base to the Al base, including the Al-Cu IMCs layer, (Al + Al₂Cu) eutectic, and Al-W IMCs zones. Five types of IMCs were confirmed to exist in these zones, namely Al₄Cu₉, Al₂Cu, AlCu, Al₄W, and Al₁₂W, accordingly. Figure 2c displays the interfacial morphology of the CuW/Al composite with the Ni interlayer, exhibiting that the interfacial thickness was obviously less than that of the interface without Ni (Figure 2a). The thickness decreased significantly to 38 μm. To some extent, the reduction of the interfacial thickness was helpful for improving the mechanical properties such as interfacial strength, which is interpreted as follows [39]. The interfacial product consisted of IMCs, and these IMCs presented a brittle nature due to the bonding character and complex crystal structure. Thus, the overgrowth of IMCs could lead to interfacial embrittlement, which was detrimental to interfacial strength. The connection between the CuW and interface was regular and straight, while that between the interface and Al base was irregular and presented a wavy morphology. It was considered that a regular and straight interfacial morphology made a positive contribution to promoting the interfacial property, in particular, for the interface with laminar structure [40]. Figure 2d displays the results of the EDS line scan of the CuW/Al interface with the Ni interface. The red line is Al, green corresponds to Ni, blue denotes Cu, and purple indicates W. The EDS line curve of the Al and Ni elements showed a clear step when it passed across the interface zone adjacent to the Al base, indicating the occurrence of composition mutation and formation of Al-Ni IMCs at the interface. This is significantly different from the variation of the interface composition without the Ni interlayer, thus indicating that the interfacial IMCs underwent a change by incorporating a Ni interlayer. The interfacial microstructure of the CuW/Al with the Ni interlayer is further discussed and analyzed in the subsequent section.

3.2. Microstructure of CuW/Al Composite with Ni Interlayer at Different Holding Time

Figure 3 shows the interfacial morphology and composition distribution of the CuW/Al composite with the Ni interlayer formed at 710 °C for 10, 20, 40, and 60 min. Clearly, the two color-contrasting layers differed from the CuW and Al base that were formed at 10, 20, and 40 min and were labeled as A and B, respectively. When the time was prolonged to 60 min, a new layer was formed at the interface adjacent to the Al base, which was denoted as C (Figure 3g). The interfacial thickness increased with the prolongation of time, as shown in Figure 3a,c,e,g. Based on the results of the microscopic measurement, the thickness of the B layer was 4, 6, 9, and 11 μm at a holding time of 10, 20, 40, and 60 min, respectively. Thus, it can be concluded that the increase of interfacial thickness was mostly contributed by the increase in the thickness of the B layer, which was attributed to the intensified atomic diffusion and sufficient reaction caused by the prolonged holding time. The corresponding kinetic analysis is discussed in Section 3.3. The results of the EDS line scan are shown in Figure 3b,d,f,h. The intensity variation of elements was found to be similar at different holding times. First, a mutation of the Ni composition was observed adjacent to the CuW base, and an obvious composition platform existed in the interface zone. Next, the second composition platform began to emerge which was mainly from the variation of the increased Al and decreased Ni element adjacent to the Al base, namely the B layer, as indicated by the orange shaded zone in the figures. This results indicate that IMCs composed of Al and Ni elements were formed there.

Third, another mutation occurred in intensity again adjacent to the Al base at the contact between the interface zone and Al base. At 60 min, the lines of the Al and Ni elements appeared as two platforms, indicating that the B and C layers consisted of different IMCs.

Figure 4 displays the SEM micrographs and EDS results of CuW/Al composite with the Ni interlayer bonded at 710 °C for 10 min. The figure reveals that two layers with color-contrasting layers differed from the base, and they were marked by A and B, as shown in Figure 4a. To identify the constitution of the A and B layers formed between the CuW and Al, an EDS analysis was employed. The results show that the atomic percent of Al was 61.02% and that of Ni was 38.98% in the B layer. The range of the Al atomic percentage for Al₃Ni₂ was 59.5–63.2% in the Al-Ni binary diagram phase [41]. Therefore, by combining the results of the EDS spot san analysis and Al-Ni binary diagram phase, it was considered that the B layer was Al₃Ni₂. Moreover, there were five types of IMCs (AlNi₃, Al₃Ni₅, AlNi, Al₃Ni₂, and Al₃Ni) in the Al-Ni system under the equilibrium condition, according to the Al-Ni binary phase diagram. It was reported that the formation energy of Al₃Ni₂ was less than that of the other Al-Ni IMCs, indicating that the Al₃Ni₂ was easily formed based on the thermodynamics condition [42]. This conclusion is consistent with our experimental results. For the A layer, the atomic percent of Ni was 96.43%, and that of Al was 3.57%, indicating that the A layer with a larger thickness was the Ni(Al) solid solution. An EDS element mapping scan of the CuW/Al composite was also conducted to clearly observe the distribution of the Cu, W, Ni, and Al elements, and the corresponding results are presented in Figure 4c–f. There existed a layer of with a Ni-diffused region on the right of the Ni interlayer and an Al-diffused region on the left of the Al base. The thickness of the diffused region was nearly 4 μm, which corresponds to the thickness of the B layer shown in Figure 4a. It further indicates that inter-diffusion occurred between the Ni interlayer and Al base. Moreover, a wavy-like connection was formed at the Ni interlayer/Al₃Ni₂ and Al₃Ni₂/Al. This was possibly due to the discrepancy in the crystal structure. Notably, Ni and Al have an (fcc) lattice, while Al₃Ni₂ belongs to the trigonal system. There existed a large internal stress at the connection of Ni/Al₃Ni₂ and Al₃Ni₂/Al caused by the abovementioned difference of the crystal structure.

Figure 5 shows the morphological characteristics and element distribution of the CuW/Al composite with the Ni interlayer bonded at 710 °C for 20 min. The interfacial microstructure was similar to that of the interface bonded for 10 min. The interface was composed of two color-contrasting layers that were labeled A and B. Compared to the interface shown in Figure 4a, the thickness of the B layer increased with the increasing time. The average thickness of the B layer was 6 μm. The EDS results showed that the atomic percent of Al was 60.72% and that of Ni was 39.28% in the B layer. Consequently, the B layer was Al₃Ni₂ based on the Al-Ni binary diagram phase and the results of EDS spot san analysis. The atomic content of Al was 4.01% in the A layer near the CuW base, showing that the A layer was the Ni (Al) solid solution. Figure 5c–f show the EDS mapping distribution of the CuW/Al composite for the Cu, W, Ni, and Al elements. The Ni-diffused region with a light green color was easily observed, marked by a white dotted line, which was more obvious and thicker than that of the interface formed at 10 min. Furthermore, there also existed an Al-diffused region with a light orange color, as shown in Figure 5f. An increase in the diffusion region indicates that more Al and Ni atoms were involved in diffusion because of the increased holding time. As a result, more Ni atoms dissolved into Al liquid, and the Ni (Al) supersaturated solid solution was easily formed, promoting the formation of more Al₃Ni₂. During the growth of the Al₃Ni₂ layer, it first grew along the transverse direction and then longitudinally to form a continuous Al₃Ni₂ layer. Consequently, the thickness increased with the increasing holding time after the formation of a continuous Al₃Ni₂ layer.

Figure 6 presents the SEM micrographs and EDS results of the CuW/Al composite with the Ni interlayer at a bonding temperature of 710 °C and holding time of 40 min. The interface still consisted of A and B layers, which is consistent with that of the interface formed in short times, such as 10 and 20 min.

The A and B layers were further confirmed by an EDS scan analysis, and the corresponding results are presented in Figure 6b. The results show that the atomic percents of Al to Ni were close to 3:2 for the B layer; it was considered that the B layer with a dark gray color was Al₃Ni₂ based on the Al-Ni binary phase diagram. The average thickness of the Al₃Ni₂ layer increased to 9 μm with the increasing holding time, which was obviously thicker than the interface shown in Figure 4a and Figure 5a. The atomic percent of Al was 3.23% in the A layer, indicating that the A layer with a light gray color was a Ni (Al) solid solution. Figure 6c–f display the EDS mapping scan of the CuW/Al composite for the Cu, W, Ni, and Al elements. The results indicate that the Ni-diffused region with the light green color and Al-diffused region with the light orange color were all thicker than the interface formed at 10 min. More importantly, the difference occurred in the interfacial morphology between Ni(Al) and Al₃Ni₂. At shorter times (10 and 20 min), the contact of Ni(Al) and Al₃Ni₂ presented a wave-like structure, while the contact of Ni(Al) and Al₃Ni₂ was prone to smoothening at 40 min. This may be explained as follows: the growth of Al₃Ni₂ was mainly dominated by the diffusion of the Al atom into the Al₃Ni₂ layer, because the contact between the Ni interlayer and Al base was prevented due to the formation of the Al₃Ni₂ layer. Therefore, it presented the solid–solid interface characteristic; that is, the interface was smooth.

Figure 7 illustrates the interfacial morphology and elemental distribution of the CuW/Al composite with a Ni interlayer bonded at 710 °C for 60 min. Three color-contrasting layers labeled as A, B, and C existed at the interface, as presented in Figure 7a. The thickness of the B layer was greater than that of the C layer, which was obviously different from the interfacial microstructure at holding times of 10, 20, and 40 min. A multilayer microstructure was observed to be formed between the CuW and Al base at a prolonged holding time of 60 min. The contact interface between the A and B layers was straight and smooth, while that between the B and C layers was irregular. Moreover, the C layer presented a columnar morphology. To identify the constituents of the multilayer microstructure, an EDS spot scan analysis was carried out, as shown in Figure 7b.

The results show that the atomic percent of Al was 1.63, 60.07, and 74.78% in the A, B, and C layers, accordingly. According to the Al-Ni binary phase diagram, the atomic percentage of Al in Al₃Ni₂ was 59.2–63.2% and that of Al in Al₃Ni was 75%, indicating that the A layer was the Ni (Al) solid solution, the B layer was Al₃Ni₂, and the C layer was Al₃Ni. Consequently, there were two IMC layers: not only a homogenous Al₃Ni₂ layer formed next to the Ni interlayer but also an irregular Al₃Ni layer generated next to the Al base. The same morphological characteristics were also observed by Konieczny et al. [43]. Moreover, it was also shown that Al₃Ni₂ formed first, followed by Al₃Ni, according to the quasi-dynamic microstructure observation at different times of 10, 20, 40, and 60 min. This is not coincident with the results reported in the literature study [44]. Sun et al. considered that Al₃Ni was formed first when the solid solubility of Ni in Al liquid exceeded its solid solution limit, and then the Ni atom diffused to Al₃Ni to undergo reaction and formed Al₃Ni₂. Based on this formation sequence, herein, a combined analysis of first-principles calculations is presented and explained in detail in the following sections. A multilayer microstructure with three sub-layers was observed to be formed at the interface with an increased holding time. The reason is that a longer holding time promoted the sufficient diffusion of Al and Ni atoms at the front of the interface. This was mainly caused by the gradient distribution of the concentration along the interface. A similar multilayer microstructure was also found in other systems, such as the Al-Fe reaction system [45].

3.3. Growth Kinetics of Al₃Ni₂ Layer

The reaction between the Ni interlayer and Al base promoted the consumption of Al, mainly by transforming it into Al₃Ni₂. The interface thickness increased with the increase of the holding time, as presented in Figure 3, which exhibits an important effect on the properties at the interface. Thus, it is necessary to study the growth kinetics of the Al₃Ni₂ layer formed at the CuW/Al interface. The growth of the Al₃Ni₂ layer can be generally expressed by Equation (1) [46], as follows:

$$\Delta y=k{t}^n$$

(1)

where Δy is the average thickness of Al₃Ni₂ (m), k is the rate constant, t is the holding time (s) at 710 °C, and n is the time exponent. The logarithm form of Equation (1) can be deduced as follows:

$$\ln \Delta y=\ln k+n\ln t$$

(2)

Then, the curve of lnΔy versus lnt can be obtained by using Equation (2), as shown in Figure 8.

According to the results of linear fitting, the slope is n, and the intercept is lnk. The value of the time exponent, n, indicates a different growth behavior and mechanism. In general, there are mainly the following cases [47]: if n = 1/3, the interfacial growth is attributed to grain boundary diffusion mechanism; if n = 1/2, the relationship between lnΔy and lnt follows a parabola law, implying the volume diffusion is the main mechanism; and if n = 1, the relationship between lnΔy and lnt satisfies linear law, demonstrating that chemical reaction plays a predominant role in growth mechanism. In this study, the time exponent, n, of the Al₃Ni₂ layer was 0.5692; the value is similar to the available values reported in the literature study [48]. Consequently, the formation of the Al₃Ni₂ layer was mainly affected by the volume diffusion mechanism. López et al. [49] held the same opinion that the growth of Al₃Ni₂ followed a parabolic law and was controlled by volume diffusion behavior.

3.4. First-Principles Calculation of Al₃Ni₂ and Al₃Ni

The analysis of the interfacial microstructure indicates that two types of IMCs were formed at the CuW/Al composite with the Ni interlayer with a thickness of 30 μm, corresponding to the Al₃Ni₂ adjacent to the CuW base and Al₃Ni adjacent to the Al base. To further analyze the microstructural evolution mechanism, the interface phase characteristics such as structural stability and formation sequence were discussed by first-principles calculations. The corresponding crystal structure model of Al₃Ni₂ and Al₃Ni was obtained after structural optimization, as presented in Figure 9.

The lattice parameters are consistent with the available calculated and experimental results [50,51,52]. The results indicate that the following calculation was reliable. The cohesive energy (ΔE) and formation enthalpy (ΔH) of Al₃Ni₂ and Al₃Ni can be calculated as follows [53]:

$$\Delta E\left({\mathrm{Al}}_x{\mathrm{Ni}}_y\right)=\frac{1}{x+y}\left[E_{tot}({\mathrm{Al}}_x{\mathrm{Ni}}_y)-x{E}_{atom}\left(\mathrm{Al}\right)-y{E}_{atom}\left(\mathrm{Ni}\right)\right]$$

(3)

$$\Delta H\left({\mathrm{Al}}_x{\mathrm{Ni}}_y\right)=\frac{1}{x+y}\left[E_{tot}({\mathrm{Al}}_x{\mathrm{Ni}}_y)-x{E}_{\mathrm{solid}}\left(\mathrm{Al}\right)-y{E}_{\mathrm{solid}}\left(\mathrm{Ni}\right)\right]$$

(4)

where E_tot is the total energy of the IMCs cell; E_atom is the energy of isolated Al or Ni atom in free state; x and y are the number of Al and Ni atoms in each IMCs cell, respectively; E_solid is the energy of a single Al or Ni atom in each Al or Ni cell; ΔE usually refers to the stability of IMCs; and ΔH is used to assess the alloying ability.

The obtained ΔE and ΔH of IMCs are listed in

Table 1. Structure parameters and calculated results of formation enthalpy (ΔH), cohesive energy (ΔE), and bonding electrons at the Fermi energy level N(E_f) of Al₃Ni₂ and Al₃Ni.

Phase	Space Group	Crystal System	a (Å)	b (Å)	c (Å)	Formation Enthalpy (eV/atom)	Cohesive Energy (eV/atom)	N (Ef)
Al3Ni2	P-3m1 (164)	Trigonal	4.052	4.052	4.906	−0.6704	−4.6746	5.740
			4.042 a	4.042	4.905
			4.219 b	4.219	5.161
Al3Ni	Pnma (62)	Orthorhombic	6.620	7.351	4.801	−0.4581	−4.3193	4.764
			6.565 c	7.257	4.750
			6.598 b	7.352	4.801

^a Cal. in Ref. [50]. ^b Exp. in Ref. [51]. ^c Cal. in Ref. [52].

. In general, the higher the absolute value of ΔE, the more stable the crystal structure. The higher the absolute value of ΔH, the stronger the formation ability of the IMCs. It can be deduced that Al₃Ni₂ exhibited better structural stability since the absolute value of ΔE for Al₃Ni₂ (−4.6746 eV/atom) was larger than that for Al₃Ni (−4.3193 eV/atom). Moreover, Al₃Ni₂ outperformed Al₃Ni in terms of formation ability because Al₃Ni₂ (−0.6704 eV/atom) showed a larger absolute value of ΔH than Al₃Ni (−0.4581 eV/atom). These results match well with the previously reported results by Zheng et al. [52], such as the ΔE and ΔH of Al₃Ni being −5.0446 eV/atom and −0.4262 eV/atom, respectively. It was therefore concluded that Al₃Ni₂ formed first, followed by Al₃Ni, which is consistent with the results provided by Ding et al. [54]. They considered that Al₃Ni₂ generated first, and then Al₃Ni was formed between Al₃Ni₂ and Al, dominated by a peritectic reaction.

The total and partial density of states (TDOS and PDOS, respectively) were analyzed to deeply estimate the bonding characteristics of IMCs (Al₃Ni₂ and Al₃Ni) for further revealing the structural stability mechanism. Figure 10 shows the DOS and PDOS of Al₃Ni₂ and Al₃Ni, where the black dotted line denotes the Fermi level, labeled as E_f, and its energy value corresponds to 0 eV. Notably, the value of TDOS for Al₃Ni₂ and Al₃Ni at the Fermi level did not equal zero; it can thus be deduced that they exhibited a metallic character. For Al₃Ni₂, the electrons’ energy contributing to bonding was concentrated in the range of −10 eV~E_f, which was derived from the valence electron contribution of Al-3s, 3p, and Ni-3d orbitals. The strongest peak was observed between −4 eV and E_f, and near the Fermi level, which originated from the interactions between the Al-3p and Ni-3d orbitals. For Al₃Ni, the electron energy contributing to bonding was mainly in the range of −10 eV~E_f. The strongest peak was located between −5 eV and −2 eV, which originated from the valence electron contribution of Ni-3d orbitals, as shown in Figure 10b. The electrons between −3 eV and E_f were mainly derived from the interactions of Ni-3d and a few of Al-3p, while the range of −10 eV~−5 eV originated from Al-3s and Al-3p. Clearly, the Fermi level (E_f) of Al₃Ni₂ was located at the valley of the pseudo-gap; that is, the bonding state was completely occupied, and the electron density at Fermi was very finite, indicating that Al₃Ni₂ exhibits better stability. However, the Fermi level (E_f) of Al₃Ni was located on the left side of the valley of the pseudo-gap, the bond state was not completely occupied, and the spacing of the peak at both sides of the Fermi level was large. By using the density integral method of electron states, the number of electrons could be obtained. The calculated results are listed in Table 1. The larger the number of bonding electrons, the more stable the structure [55]. Obviously, the N (E_f) of Al₃Ni₂ (5.740) was larger than Al₃Ni (4.764), implying that the stability of Al₃Ni₂ was superior to Al₃Ni, which is consistent with the previous stability conclusion achieved from the cohesive energy, ΔE.

3.5. Microstructural Evolution Mechanism

As analyzed above, the CuW/Al composite with the Ni interlayer was composed of two types of nickel aluminides (Al₃Ni₂ and Al₃Ni). The microstructural evolution mechanism is illustrated in Figure 11. With the increase of temperature, the atomic diffusion was activated, and the Al base began to melt. Ni atoms dissolved and diffused into Al liquid, and consequently, the front of the interface was constantly moving toward the Al side, driven by the concentration gradient (Figure 11a). The solubility of the Ni atom in Al liquid increased with the intensification of diffusion; the Ni atom first reached saturation in Al liquid due to its low solid solubility in Al. The solid solubility was 5 at% at 700 °C, according to the binary phase diagram. Then, Al₃Ni₂ was generated via a chemical reaction between Ni and Al (Figure 11b). Based on the formation theory of IMCs in dissimilar metal systems, when the Al₃Ni₂ nucleated at the interface and presented Al₃Ni₂ islands, it first grew perpendicular to the interface; it then grew parallel to the interface, and this was followed by the formation of the continuous Al₃Ni₂ layer with a certain thickness (Figure 11c). The average thickness increased with the prolonged holding time.

The ability of liquid Al to decompose Al atoms was greater than that of solid Ni to decompose Ni atoms [56]. Thus, the contact of the Ni interlayer and Al liquid was inhibited due to the Al₃Ni₂ layer; as a result, the growth of Al₃Ni₂ was mainly dominated by the diffusion of Al atoms into the Al₃Ni₂ layer. It was therefore observed that the sub-interface between Al₃Ni₂ and Ni was regular and smooth, showing solid–solid interface morphological characteristics (Figure 7a). With the dissolution of Al₃Ni₂, the Ni-rich area existed at the edge of Al₃Ni₂ and Al liquid, and then Al₃Ni was formed through a peritectic reaction (Al₃Ni₂→Al₃Ni + (Al)). A large amount of Al₃Ni precipitated and gathered at the front of the interface due to a higher concentration of the Ni atoms (Figure 11d). With the decrease in the temperature, the Ni atoms reached a supersaturated state in Al liquid, and Al₃Ni began to precipitate. The growth pattern was similar to that of Al₃Ni₂; that is, it grew laterally, then longitudinally, and finally formed Al₃Ni layer. It was considered that Al₃Ni₂ was the first phase formed at the interface during the holding time, followed by Al₃Ni formed during the cooling stage.

3.6. Mechanical Properties of CuW/Al Composite without and with Ni Interlayer

Figure 12 shows the stress–strain curves of the CuW/Al composite with a Ni interlayer at 710 °C for different holding times, exhibiting that the largest shear strength was 90.9 MPa, which was obtained at the holding time of 40 min. The strain was close to 6.2%, indicating that little plastic deformation occurred and was mainly dominated by the brittle fracture during the action of stress. The shear strength increased by 76% compared to that of the composite without a Ni interlayer, as was reported in a previous study [22]. This indicates that the interfacial strength was effectively enhanced due to the introduction of the Ni interlayer.

This can be explained as follows: the Ni interlayer changed the reaction path. The interfacial IMCs were changed from a series of Al-Cu and Al-W IMCs to Al-Ni IMCs, among which Al-Ni IMCs exhibited a lower hardness than the former, contributing to the interfacial strength. Moreover, the Ni interlayer acted as a physical barrier, and the reaction between the CuW and Al base was weakened. Consequently, a reduction in the interfacial thickness and change in the type of IMCs were observed. These two facts are conducive to improving the interfacial strength. The shear strength of the CuW/Al composite with the Ni interlayer increased with the increase in the holding time in the range of 10–40 min. This may be attributed to the following reasons. Based on the analysis of growth kinetics of the Al₃Ni₂ layer presented in Section 3.2, the growth of Al₃Ni₂ was controlled by volume diffusion, i.e., the lattice diffusion mechanism. Therefore, although the interfacial microstructure was similar for holding times of 10, 20, and 40 min, the shear strength increased with the prolongation of the holding time. This may be related to grain refinement with the prolonged holding time. The same law was also found in another interface of an Al/metal composite, i.e., an Al-Fe system [57]. When the time was increased to 60 min, the shear strength was less than that of samples with a holding time of 40 min due to the formation of Al₃Ni with a hardness higher than that of Al₃Ni₂, according to the available experimental data [58]. The IMCs with higher hardness led to the increase in the stress concentration and embrittlement of the interface, resulting in a decrease in interfacial strength.

Figure 13 displays the morphology of the fracture surface of the CuW/Al composite without and with the Ni interlayer bonded at 710 °C for 40 min. For the composite without the Ni interlayer, typical cleavage fracture characteristics such as cleavage step and cleavage river pattern were observed on the fracture surface, as shown in Figure 13a. However, the surface morphology for the composite with a Ni interlayer was relatively flat, and it was mainly composed of the flocculent P₃ and the light gray layer-like P₂, as shown in Figure 13b. An EDS analysis was further carried out to determine the fracture phase composition. The results show that the atomic percent of Cu to Al in point P₁ was 32.2:67.8, indicating that the P₁ corresponded to Al₂Cu. The atomic percents of Al to Ni in points P₂ and P₃ were 4.94:95.26 and 67.14:32.86, respectively; thus, it could be speculated that P₂ was the remaining Ni interlayer and P₃ was Al₃Ni₂. These results clearly indicate the existence of a crack at the Al₂Cu/Al interface due to its weakest nature among these IMCs phases [59] of the composite without a Ni interlayer. Chen et al. [60] studied the micromechanical behavior of the Cu/Al₂Cu/Al system and observed that the dislocation first nucleated at the Cu/Al₂Cu interface. For the CuW/Al composite with Ni interlayer, the crack nucleated at the Al₃Ni₂/Ni interface. An almost impossible slip-bands slide was observed through the Al₃Ni₂/Ni interface due to the mismatch of the crystal structure between Ni and Al₃Ni₂. The failure characteristic of this composite was similar to that reported in the previous investigation of the Ti-Al₃Ti system [61]. In fact, the fracture behavior of the laminated metal composites was similar, as most IMCs were brittle owing to few slip systems; thus, the plastic deformation was limited through the metal/IMCs interface.

4. Conclusions

In this study, the CuW/Al composite with a Ni interlayer was fabricated by using the vacuum hot-pressing diffusion bonding technique. The interfacial microstructure and mechanical properties were investigated in detail. According to the results of the experimental characterization and first-principles calculation, the main conclusions can be drawn as follows:

• By adding a Ni interlayer with a thickness of 30 μm, the CuW/Al interfacial microstructure was effectively optimized. The interface phase consisted of Ni (Al) solid solution and Al₃Ni₂ IMCs. The average interfacial thickness reduced significantly from 340 to 35 μm. The growth kinetics of the Al₃Ni₂ layer followed a parabolic behavior, which was mainly affected by the volume diffusion mechanism.
• The analysis of the microstructural evolution revealed that Al₃Ni₂ was the first formed phase with a prolonged holding time of 60 min, which was attributed to a lower formation enthalpy (ΔH = −0.6704 eV/atom) and a larger number of bonding electrons at the Fermi energy level (N_Ef = 5.740) compared to Al₃Ni.
• The interfacial strength was significantly improved because of the variation in the type of interfacial IMCs. The maximum shear strength was 90.9 MPa, corresponding to an increase by 76% compared to the composite without the Ni interlayer. The fracture analysis indicated that the composite presented a brittle fracture characteristic, and a crack existed at the Al₃Ni₂/Ni interface.