Sintering mechanism of the Cu–Ag core–shell nanoparticle paste at low temperature in ambient air

Abstract

Copper (Cu) nanoparticle (NP) paste has emerged as a promising choice for future high power electronics. However, the application of CuNP paste is generally limited because of its easy oxidation and complex processing parameters such as high sintering temperature (∼400 °C) and necessary protection atmosphere. Although capping CuNPs with a stable layer to form core–shell NPs could improve its anti-oxidation ability, current coating methods usually required complex equipment and protection atmosphere, and there have been no reports of fabricating joints with core–shell NPs. In this paper, Cu–silver (Ag) core–shell NPs with a unique structure, i.e., tiny AgNPs with an average diameter of 6.9 nm which almost completely cover the surface of initial smooth CuNPs, after first developing them in ambient air and then using a mild two-step method, are reported. After coating the tiny AgNPs on the surface, not only was the anti-oxidation ability improved during keeping (∼2 months), but also the sintering temperature of NP paste was lowered ∼70 °C. Through in situ observation during sintering, it was found that the tiny AgNPs tended to pre-melt and pre-wet to form active quasi-liquid Ag films, after which the Cu–Ag core–shell NPs were joined by the first to be found pure Ag sintering necks. Furthermore, the application of the Cu–Ag core–shell NP paste was demonstrated by fabricating voidless joints with a shear strength of 26.5 MPa in ambient air at 250 °C, and this showed a good potential for the future applications in high power electronics.

---

1. Introduction

With rapid developments in the area of high power electronic devices, there has been a dramatic increase for high temperature interconnection materials for joining dies and substrates has been attracting attention from both industry and academia.¹ In the meantime, the ban of lead (Pb) across the world has aggravated the need for a suitable medium and high temperature interconnection materials.² Silver (Ag) and copper (Cu) are two of the most promising candidates for high temperature interconnection materials because of their high melting temperatures (963 °C and 1083 °C, respectively), as well as their excellent thermal and electrical properties.^3–5 But the application of Ag and Cu micro particles as interconnection materials has been limited by the high sintering temperature required for their use. As the diameter of particles reaches nanoscale, especially less than 20 nm, the initial sintering temperature of these nanoparticles (NPs) would be dramatically decreased.⁶ Furthermore, after low temperature sintering, the sintered NPs lose their small size effect and could endure a much higher functioning temperature comparable to that used with bulk materials. Thus, a metallic NP paste made up of Ag or Cu or both of them is one of the most promising choices for future high power electronic devices.

Over recent decades, AgNP paste has been exhaustively investigated, and applied in industrial manufacture, as die attached materials in power devices, wiring materials and electrodes in flexible integrated circuits.^7–9 The success of AgNP paste has demonstrated the practicality of the idea, i.e., in metallic NP paste. However, the wide application of AgNP paste has been seriously limited because of severe migration and high cost.¹⁰ However, although CuNP paste is much cheaper compared with AgNP paste, the sintering process for CuNP paste is more complicated because of the spontaneous oxidation of the Cu surfaces in air.^11–13 Therefore, the sintering process of CuNP paste required a higher temperature (∼400 °C) and special protection environments (e.g., a nitrogen (N₂) or hydrogen (H₂) atmosphere), which could limit suitable choices for substrates and increase the total costs.^14,15

To overcome the problems as mentioned previously, Cu–Ag core–shell NP paste was proposed as it combines the merits of CuNP paste and AgNP paste.¹⁶ As the Cu core was capped by an Ag shell, Cu–Ag core–shell NP paste had good anti-oxidation and anti-migration properties. Also, the cost was much lower than that of AgNPs. Lee et al. reported that Cu–Ag core–shell NP paste could be used as a wiring material in flexible electronics because of its improved anti-oxidation ability as well as its better electrical conductivity compared with CuNP paste, but the synthesis of Cu–Ag core–shell NPs needed the protection of an argon (Ar) atmosphere, which would increase the total costs.¹⁶ Kim et al. reported that Cu–Ag core–shell NPs with AgNPs partially covering the surfaces of the CuNPs could stay unoxidized for six months,¹⁷ and this structure is common in a large variety of core–shell systems, such as gold (Au)–Ag, Cu–Au, Ag–palladium (Pd).^18–20 However, the facility for preparing the Cu–Ag core–shell NPs was very complex and it also required the protection of an additional Ar atmosphere.¹⁷ However, Cu–Ag core–shell NPs with tiny AgNPs on the surface required a lower initial temperature for sintering, and the density of the sintered Cu–Ag core–shell NP structure was also increased. As well the joining of the Cu–Ag core–shell NPs and micro-sized substrates was enhanced by decreasing the porosity between them, which eventually enhanced the mechanical strength as well as the electrical and thermal conductivity of the joints.^10,21 There exist several other reports on the synthesis of Cu–Ag core–shell NPs, but the fabrication of joints using Cu–Ag core–shell NP paste has not been performed until now.^22–24

In this paper, an exciting new method for developing Cu–Ag core–shell NPs with tiny AgNPs covering the surfaces of medium-sized CuNPs using a mild two-step method, is reported. This new method combines the merits (high thermal and electrical conductivity as well as low cost), overcomes their weaknesses (migration and oxidation), and also enhances the density of the sintered NP paste. Cu–Ag core–shell NPs with tiny AgNPs fully covering the surfaces of the CuNPs were firstly developed using a simple two-step method. The unique structure and anti-oxidation ability during the conservation of the Cu–Ag core–shell NPs were investigated using transmission electron microscope (TEM), X-ray diffraction (XRD), and X-ray photoelectron spectroscopy (XPS) analysis. In situ sintering of Cu–Ag core–shell NPs was then conducted in a TEM to investigate their low temperature joining mechanism, which was further proved using energy dispersive X-ray (EDX) analysis. Furthermore, the microstructure of the joints of the Cu/Cu–Ag core–shell NP paste/Cu sandwich structure, fabricated at low temperature, was studied, and the corresponding shear strength was measured.

2. Experimental

Materials

Analytical grade copper sulfate pentahydrate (CuSO₄·5H₂O), silver nitrate (AgNO₃·H₂O), poly(vinyl pyrrolidone) (PVP-K30), sodium hypophosphite (NaH₂PO₂·2H₂O), ammonium hydroxide (NH₄OH), diethylene glycol (DEG), hydrochloric acid (HCl) and sulfuric acid (H₂SO₄) were used as received without further purification.

Synthesis of CuNPs

CuNPs were synthesized using a modified method.²⁵ The synthesis process is illustrated in Fig. S1a (ESI†). Specifically, PVP (15 M), acting as a capping agent, was dissolved in DEG (70 mL), and CuSO₄·5H₂O (10 M), acting as precursor, was dissolved in water (20 mL). NaH₂PO₂·H₂O (20 M), used as reducing agent, was added to the DEG solution and the solution was heated to the reaction temperature (140 °C). The aqueous solution of CuSO₄·5H₂O (10 M) was then injected into the hot reaction medium with a syringe pump within 2 minutes. The injection rate of the Cu salt solution was 5 mL min⁻¹. After reacting for one hour, the solution was cooled to room temperature and the particles were separated using centrifugation and then washed with ethanol. Then the product was washed with diluted H₂SO₄ and water ready for further use.

Synthesis of Cu–Ag core–shell NPs

The synthesis process is illustrated in Fig. S1a (ESI†). Firstly, NH₄OH was added dropwise into silver nitrate (10 M) solution until a colorless transparent solution was obtained, which is a silver–ammino complex. Then the obtained solution was added dropwise into CuNP (10 M) solution with vigorous stirring. The solution quickly changed colour to black from deep purple, indicating the formation of the AgNPs. The reaction was continued for 5 minutes. After the reaction, the as-synthesized Cu–AgNPs were obtained using centrifugation, and then washed with ethanol and water. The ultraproduct was conserved in ethanol.

Preparation of NP paste

Cu or Cu–Ag core–shell NPs were firstly cleaned using HCl to remove the undesired PVP layer as reported before.²⁶ Then they were mixed with DEG with a mass ratio of 6 : 1, and the mixture was ultrasonically stirred for 30 min to obtain the NP paste with the expected viscosity and liquidity. After mixing, the mixture was heated at 60 °C to evaporate the undesired solvents as shown in Fig. S1b (ESI†).

Fabrication of the CuNPs paste/Cu sandwich structure

The fabrication process is demonstrated in Fig. S1c (ESI†). Firstly, a thin layer of NP paste was placed on a Cu substrate using a simple screen printing method, and then a smaller Cu substrate was put onto the nanopaste to form a sandwich structure. This CuNP paste/Cu sandwich structure was then sintered at 200 °C or higher for 20 min under a pressure of 5 MPa to form effective joints.

Instruments and measurements

In situ TEM heating was conducted in a JEM-2010 TEM (Jeol), and high resolution transmission electron microscopy (HRTEM) and EDX analysis were conducted using a JEM-2100 TEM (Jeol) operated at 300 kV. Scanning electron microscopy (SEM) images were obtained using a Quanta 200F field-emission SEM (FEI). Differential scanning calorimetry (DSC) and thermogravimetric analysis were performed synchronously using a DSC 200 F3 thermogravimetric analyzer (Netzsch) in a N₂ atmosphere, and the preset temperature interval was from room temperature to 500 °C with a heating rate of 10 °C min⁻¹. The shear strength of the bonded Cu/Cu–Ag core–shell NPs/Cu sandwich structure was measured at room temperature with a Dage 4000 bond tester (Nordson DAGE), and the shear height of the blade tip above the Cu substrate was 50 μm, and the shear speed was 100 μm s⁻¹.

3. Results and discussion

Employing a mild two-step method, Cu–Ag core–shell NPs were successfully synthesized and their morphology, phase composition, and structure are shown in Fig. 1. Compared with the initial smooth surface of pure CuNPs (see Fig. S2 ESI†), the surfaces of the Cu–Ag core–shell NPs were extremely coarse because of the coating of tiny AgNPs on the surface as shown in Fig. 1a. The average diameters of the initial CuNPs and AgNPs were 57.5 nm (Fig. 1b) and 6.9 nm (Fig. 1c), respectively. Phase composition of Cu–Ag core–shell NPs was investigated using a combination of XRD analysis and the selected area electron diffraction (SAED) results. As shown in Fig. 1d, there were three strong peaks at 2θ = 43.3°, 50.4°, and 74.1°, corresponding to diffractions in the {111}, {200}, and {220} crystal planes of the face centered cubic (fcc) Cu (JCPDS 04-0836), respectively, and there were three other relatively weaker peaks at 2θ = 38.1°, 64.4°, and 77.4°, corresponding to the {111}, {220}, and {311} crystal planes of the fcc Ag (PDF 65-2871), respectively. It is worth noting that one of the main peaks of Ag at 2θ = 44.3° could not be identified because of the overlapping with the main peak of Cu, and the existence of Ag will be further discussed in the XPS results. Also, the SAED result showed the same result in Fig. 1e. The weak diffraction rings, marked as green arrows, correspond to the (111), (200), (220) and (311) crystal planes of AgNPs, indicating that there were a great number of tiny AgNPs, and other diffraction rings composed of bright diffraction points belonging to the CuNPs as marked by purple arrows. The HRTEM image displaying the unique structure of the as-synthesized Cu–Ag core–shell NPs is given in Fig. 1f. The (200) crystal planes of CuNPs and AgNPs are marked by arrows, and the interplaner distances were 0.186 nm and 0.209 nm, respectively. The CuNPs were tightly coated by AgNPs as indicated by the tight interface between them without any visual void, and they should be connected by Cu–Ag metallic bonds. Also, a moire fringe appeared at both the areas of the CuNPs and the AgNPs because of the overlapping of AgNPs with CuNPs, and with other AgNPs, respectively. These results indicated that the AgNPs fully covered the surfaces of the CuNPs, which was the first time they had been created using a mild two-step method.

Fig. 1 Characterization of Cu–Ag core–shell NPs: (a) TEM image of Cu–Ag core–shell NPs (b) size distribution of initial CuNPs, (c) size distribution of AgNPs on the surfaces of Cu–Ag core–shell NPs, (d) XRD pattern of Cu–Ag core–shell NPs, (e) SAED result of Cu–Ag core–shell NPs, (f) HRTEM image of Cu–Ag core–shell NPs.

The capping of CuNPs by AgNPs could solve the easy oxidation problem of CuNPs, which has been reported before,¹⁶ and also the anti-oxidation ability of Cu–Ag core–shell NPs could verify whether Cu core–shell NPs have been entirely covered by tiny AgNPs on the surface. Fig. S3 (ESI†) shows the XRD pattern comparison of Cu–Ag core–shell NPs before and after storage in ethanol for two months. Diffraction peaks of Cu–Ag core–shell NPs did not change and there were no diffraction peaks belonging to Cu₂O or CuO reported in the Cu nanomaterials,^11,12 which means the as-synthesized Cu–Ag core–shell NPs had a good anti-oxidation ability. Then XPS analysis was further conducted to characterize the element composition of the different valence states of Cu–Ag core–shell NPs after having been stored in ethanol for two months. Survey spectra in Fig. 2a demonstrate that Cu, oxygen (O), and Ag elements coexisted in the Cu–Ag core–shell NPs with atomic percentages of 71.34%, 6.22%, and 22.44%, respectively, (Table S1; ESI†), and then Ag 3d, O 1s as well as Cu 2p_3/2 peaks were further studied. Fig. 2b clearly illustrates that the binding energies of Ag 3d_5/2 and Ag 3d_3/2 electrons were 374.25 eV and 368.15 eV, respectively. Usually, Ag exists in three different states, i.e., the AgO state with an energy of 367.0 eV, the Ag₂O state with an energy of 367.7 eV, and the Ag⁰ state with an energy of 368.2 eV as reported previously.²⁷ Comparing these values, obviously Ag mainly existed in the Ag⁰ state, i.e., AgNPs. O 1s spectra in Fig. 2c show two peaks at binding energies equal to 529.8 eV and 531.0 eV, corresponding to O²⁻ 1s and O⁰ 1s, respectively. The atomic percentages were further calculated based on the fitted areas (3.19% and 3.03% for O²⁻ 1s and O⁰ 1s, respectively), where O²⁻ ions might also belong to the remaining organics. Furthermore, the Cu 2p_3/2 spectra was used to identify the anti-oxidation ability of the Cu–Ag core–shell NPs, and it also demonstrated two peaks at a binding energies equal to 932.3 eV and 933.9 eV in Fig. 2d, which were Cu⁰ 2p_3/2 and Cu²⁺ 2p_3/2, respectively. The calculated atomic percentages for Cu⁰ 2p_3/2 and Cu²⁺ 2p_3/2 were 68.84% and 2.50% (27.5 : 1), respectively, the ratio of which has been reported to be nearly 4.8 : 1 without further oxidization after storage.¹⁶ Comparing these values, the Cu–Ag core–shell NPs clearly showed a good anti-oxidation ability during storage for two months.

Fig. 2 XPS analysis results of Cu–Ag core–shell NPs after storage for two months: (a) survey spectra, (b) Ag 3d spectra, which indicates the existence of Ag, (c) O 1s spectra, (d) Cu 2p_3/2 spectra, which indicates that the Cu element mainly existed as Cu⁰.

The fully coverage of AgNPs on the surface could not only improve the anti-oxidation ability of CuNPs, but could also enhance the joining of the Cu–Ag core–shell NPs. The nano joining process of Cu–Ag core–shell NPs was observed using in situ TEM heating technology, during which the joining steps were heated at different temperatures for different times were designed to fit well with the practical parameters, and this is displayed in Fig. 3. Fig. 3a shows the TEM image of two adjacent Cu–Ag core–shell NPs with a tiny nano gap between them before heating. After being heated at 100 °C for only two minutes, tiny AgNPs started to coalesce and grow bigger as marked by the purple and blue arrows, and the two Cu–Ag core–shell NPs formed direct contacts at the original gap area because of the rotation of Cu–Ag core–shell NPs during heating as shown in Fig. 3b. When Cu–Ag core–shell NPs were heated at 150 °C for 5 min, the tiny AgNPs coalesced, a process called Oswald ripening, with the small sized NPs being ingested by bigger NPs,²⁸ was extremely obvious and the Cu–Ag core–shell NPs tightly contacted as shown in Fig. 3c. Cu–Ag core–shell NPs started to coalesce at 180 °C as marked by a green arrow in Fig. 3d, and the surfaces of CuNPs were still fully covered by AgNPs at this point. After having been heated at 220 °C for 5 min, most of the tiny AgNPs disappeared and the surfaces of the Cu–Ag core–shell NPs begun to become smooth, a phenomenon which is seen well in Fig. S4 (ESI†). Then the temperature was further increased to 300 °C, and the morphology of the Cu–Ag core–shell NPs showed the same trend to smooth the surfaces because AgNPs tended to wet the surfaces of CuNPs, which will be further explained in Fig. 4. The whole joining process between two individual Cu–Ag core–shell NPs through the tiny AgNPs on the surface was also recorded and is shown in Movie 1 (ESI†). It is worth mentioning that the existence of tiny AgNPs on the surface were essential to the low temperature joining of the Cu–Ag core–shell NPs. Without the coating of tiny AgNPs on the surface, the starting temperature for the sintering of the CuNPs was 250 °C and the coalescence rate was very low even when they were heated to as high as 408 °C, as shown in Fig. S5 (ESI†). Compared with the Cu–Ag core–shell NPs, the starting temperature for the sintering of the CuNPs was much higher at 250 °C and the coalescence rate was much slower even when they were heated to as high as 408 °C as shown in Fig. S5 (ESI†). Also, the whole joining process between several CuNPs was recorded as Movie 2 (ESI†). Clearly, the CuNPs encountered much less change in morphology compared with the Cu–Ag core–shell NPs, and also the joining seemed to be not as reliable as that of the Cu–Ag core–shell NPs. From the comparison of the in situ sintering results of Cu–Ag core–shell NPs and CuNPs, it was concluded that the low starting temperature of the sintering and the fast coalescence rate of the Cu–Ag core–shell NPs should be attributed to the tiny AgNPs on the surface because of their extremely large surface to volume ratio because of their reduced particle size. The following equation is one of the most commonly used equations for expressing the starting temperature of sintering based on thermodynamic theory:²⁹

where T_b is the starting temperature of the sintering, T_m is the melting point of the bulk material, S_m is the bulk melting entropy, r is particle radius, r₀ = 3h (where h is the atomic diameter), and k is the Boltzman constant. The curves, demonstrating the relationship between T_b and r, of Cu and Ag, are given in Fig. S6 (ESI†). Clearly, only when the radius of the NPs was less than 10 nm could the starting temperature of the sintering decrease dramatically, which fits well with the experimental results obtained in this work.

Fig. 3 Joining process of two adjacent Cu–Ag core–shell NPs during in situ TEM heating: (a) initial Cu–Ag core–shell NPs with a nano gap between them, and Cu–Ag core–shell NPs after heating at (b) 100 °C for 2 min, (c) 150 °C for 5 min, (d) 180 °C for 5 min, (e) 220 °C for 5 min and (f) 300 °C for 5 min. Clearly, when the Cu–Ag core–shell NPs were heated at 100 °C, tiny AgNPs had already started to coalesce, and then when temperature was raised to 180 °C, the Cu–Ag core–shell NPs were joined together by tiny AgNPs. After heating at 300 °C, the Cu–Ag core–shell NPs became smooth with several AgNPs on the surface.

Fig. 4 Characterization of the sintered Cu–Ag core–shell NPs: (a) TEM image, (b) HRTEM image of the rectangular area.

Based on the fact of the high starting temperature of sintering and slow coalescence rate of CuNPs, the joining of Cu–Ag core–shell NPs at the low temperature must mainly be the result of the AgNPs on the surface. Also, it is of equal significance to understand the crystal structure of the Cu–Ag core–shell NPs sintered at low temperature. Fig. 4 shows the final structure of the Cu–Ag core–shell NPs after sintering at 300 °C for 5 min. Clearly, the Cu–Ag core–shell NPs were joined together to form a net structure as shown in Fig. 4a, and tiny AgNPs were missing after high temperature sintering. The grain boundary area, i.e., the rectangular area, is shown in Fig. 4b. It was found that the area only consisted of Ag atoms because all of the interplanar spacing corresponded to the (200) crystal plane of Ag. Also, the element distribution of Ag and Cu of sintered Cu–Ag core–shell NPs in Fig. S7 (ESI†) were used to further characterize the sintered Cu–Ag core–shell NPs. Comparing the element distribution of Ag and Cu in Fig. S7b and c (ESI†), it is easy to notice that there was almost only Ag element with a little Cu element in the joined area of the Cu–Ag core–shell NPs, in other words, the Cu–Ag core–shell NPs were connected by Ag sintering necks that were the initial AgNPs when sintered at a particular temperature. This is the first time the pure Ag sintering neck phenomenon was precisely found, and it is very important to know how they evolved into current structures during the low temperature sintering process, and this will be discussed in detail later.

To demonstrate the application of the Cu–Ag core–shell NPs as interconnection materials, a Cu/Cu–Ag core–shell NP paste/Cu sandwich structure was sintered at 200 °C for 20 min with a pressure of 5 MPa. As mentioned previously, the coating of tiny AgNPs on the surface could not only avoid the problem of oxidation, but it could also greatly decrease the starting temperature of sintering because of the size effect. Thus, the Cu substrates could be joined together by the Cu–Ag core–shell NP paste in ambient air at low temperature. The morphology of the cross section of the joint is shown in Fig. 5. Fig. 5a shows a very robust connection between two Cu substrates and there was not any obvious void in the sintered area. Compared with the porous sintered AgNP paste,³⁰ this result is extremely exciting because presence of the voidless structure is highly desirable because it will give enhanced stability during high temperature functions. Fig. 5b demonstrates the interface between the sintered Cu–Ag core–shell NPs and Cu substrate with higher magnification. Obviously, the Cu substrate was tightly connected to the sintered Cu–Ag core–shell NPs and there was also an absence of visual voids. SEM images of the interior sintered Cu–Ag core–shell NPs at different magnifications demonstrated the same void free structure in Fig. 5c and d. A similar voidless structure has been reported before in a sintered Cu₆Sn₅ structure because of the low melting temperature induced by a thick layer of quasi-liquid film.² But the voidless structure here, i.e., sintered Cu–Ag core–shell NPs without any obvious void, should also be attributed to the unique structure of the Cu–Ag core–shell NPs themselves. As mentioned earlier, NPs with a large size distribution can decrease the porosity of the sintered NP structure to a great degree, and in this case the perfect blending of CuNPs and AgNPs (with an average diameter of 57.5 nm and 6.9 nm, respectively) is another reason for the formation of voidless connection.¹⁰ Furthermore, this is more exciting because the existence of tiny AgNPs could not only decrease the porosity of the sintered structure, but it could also promote the bonding of NPs and Cu substrates at their interfaces as shown in Fig. 5b. Theoretically the porosity of the only layer of nanomaterials covering the substrate is solely controlled by the size of the nanomaterial itself, and thus the porosity at the interface between nanomaterials and substrate could be reduced by the decrease of the size of nanomaterials.^10,24 Thus, the addition of tiny AgNPs played a major role in promoting the bonding between the Cu–Ag core–shell NPs and Cu substrates, which would influence the mechanical strength of the joints.

Fig. 5 (a and b) SEM images of Cu/Cu–Ag core–shell NP paste/Cu sandwich structure and (c and d) SEM images of sintered Cu–Ag core–shell NPs after sintering at 200 °C for 20 minutes. There was almost no visual void both at the interface between sintered the Cu–Ag core–shell NPs and the Cu substrates and in the interior of the sintered Cu–Ag core–shell NPs.

A series of shear strength values of bonded Cu/Cu–Ag core–shell NP paste/Cu sandwich structure were obtained and compared with that of bonded Cu/CuNP paste/Cu sandwich structure fabricated using the same processing parameters as well as the reported joints using CuNP paste.^31–33 For all the curves, the shear strength was improved to a certain degree at higher temperature because a higher temperature led to more acceptable joining between NPs. Generally, the shear strength of the Cu/Cu–Ag core–shell NP paste/Cu sandwich structure is much higher (3–6 times) than that of the Cu/CuNPs paste/Cu sandwich structure sintered under the same parameters, and is also much higher than that of the reported joints using CuNPs paste under the protection of N₂ or H₂ as shown in Fig. 6. For example, when the sintering temperature was 250 °C, the shear strength of the joints using CuNP paste in this work,^31–33 was 5.5, 5, 11 and 14 MPa, respectively, and the shear strength of joints using mixed Cu/AgNPs and Cu–Sn core–shell NPs in^30,34 was 16.5 and 18.8 MPa, respectively, whereas the shear strength was improved to 26.5 MPa for the Cu–Ag core–shell NPs in this work. Compared with previous reported complex sintering methods of CuNP paste, the sintering of the Cu–Ag core–shell NP paste in this work was operated in air at a low temperature, which was much easier than before because of the advantages of the Cu–Ag core–shell NP paste itself. The reason why the sintered Cu/Cu–Ag core–shell NPs/Cu joints had a much higher shear strength is because of the low starting temperature of the sintering of the Cu–Ag core–shell NPs as well as the decreased porosity at both the interior of the sintered Cu–Ag core–shell NPs and the interface between the Cu–Ag core–shell NPs and Cu substrates, referred to in Fig. 3 and 5, respectively.

Fig. 6 Comparison of the shear strength values between Cu/Cu–Ag core–shell NP paste/Cu sandwich structure and recently reported joints as well as joints in this work using CuNP paste. Briefly, shear strength values of joints using Cu–Ag core–shell NP paste were much higher than the others.

The joining mechanism of Cu/Cu–Ag core–shell NPs paste/Cu sandwich structure was concluded to be as shown Fig. 7. After low temperature sintering, Cu substrates could be bonded together using sintered Cu–Ag core–shell NP paste as shown in Fig. 7a and d. The joining here could be divided into two categories, i.e., joining at the interface between the Cu–Ag core–shell NPs and Cu substrate, and joining between the Cu–Ag core–shell NPs as shown in Fig. 7b and c. It is of great significance to know how the tiny AgNPs improved the whole joining process. Here the joining process between two adjacent Cu–Ag core–shell NPs was used to demonstrate the joining mechanism because the joining process at the interface between the Cu–Ag core–shell NPs and Cu substrate was similar except that it only involved the evolution of a single tiny AgNP instead of two. The initial state of the Cu–Ag core–shell NPs in Fig. 7e is shown in Fig. 3a. When Cu–Ag core–shell NPs were heated, tiny AgNPs could pre-melt and pre-wet the surfaces of Cu–Ag core–shell NPs as shown in Fig. 7f. As the name “pre-melting” implies, quasi-liquid films can be observed on the crystalline surfaces, which have been confirmed in ice, Pb, and other one-component materials.³⁵ Usually pre-melting occurs as the excess crystal-vapor interfacial energy is reduced when a “completely dry” crystalline surface is replaced with a crystal–liquid interface and a liquid surface,³⁵ and in this case pre-melting of AgNPs could enhance the wetting of AgNPs on the surfaces of CuNPs as shown in Fig. S8 (ESI†), which is the opposite of the reported dewetting of AgNPs on the surfaces of CuNPs.³⁶ Furthermore, atoms in this quasi-liquid film have a higher energy for diffusion because of the greater entropy induced by the reduced coordination number, which could contribute to the low temperature, rapid coalescence of Cu–Ag core–shell NPs.³⁵ As the result, the Cu–Ag core–shell NPs were connected by an Ag sintering neck as shown in Fig. 7g, which corresponded to the experimental results shown in Fig. 4, S7 and S8 (ESI†). Also, the thickness of the quasi-liquid film, d could be enlarged by increasing the heating temperature T as demonstrated by the following equation:³⁷

where T_m is the melting temperature of the bulk materials. Thus, the coalescence of Cu–Ag core–shell NPs became more and more obvious as the temperature increased. However, this pre-melting and pre-wetting phenomenon of AgNPs on the surface could be the reason for the smooth surfaces of Cu–Ag core–shell NPs after sintering at a higher temperature, in which situation the Ag element still mainly existed on the surface as indicated by the Ag sintering necks in return.

Fig. 7 Schematic illustration of the joining process of Cu/Cu–Ag core–shell NP paste/Cu sandwich structure: (a) Cu/Cu–Ag core–shell NP paste/Cu sandwich structure before sintering, (b) joining at the interface between Cu–Ag core–shell NP paste and Cu substrate during sintering, (c) joining between Cu–Ag core–shell NP paste during sintering, (d) Cu/Cu–Ag core–shell NP paste/Cu sandwich structure after sintering, (e) the initial state of two adjacent Cu–Ag core–shell NPs before sintering, (f) tiny AgNPs pre-melt and pre-wetting on the surfaces of CuNPs, (g) Cu–Ag core–shell NPs are joined by an Ag sintering neck.

4. Conclusion

This paper reports a novel breakthrough approach by developing a simple method for the easy preparation of Cu–Ag core–shell NPs to combine the merits of Ag and CuNP paste, i.e., high thermal and electrical conductivity, and at the same time overcomes their individual weaknesses, such as high cost as well as electrical and iron migration of Ag, and easy oxidation and complex processing parameters of Cu. Specifically, the coating of AgNPs on the surfaces of CuNPs could contribute to the anti-oxidation ability of Cu–Ag core–shell NPs. Furthermore, the existence of tiny AgNPs could lower the starting temperature for sintering the Cu–Ag core–shell NPs to a great degree, compared with the much higher starting temperature for CuNPs, and could facilitate the joining of the Cu–Ag core–shell NPs at a low temperature by pre-melting and pre-wetting on the surface to form Ag quasi-liquid films. Furthermore, the perfect blending of Ag and CuNPs with two types of size distribution could promote the formation of high strength joints with voidless sintered Cu–Ag core–shell NPs paste structure by decreasing the porosity at both the interior of sintered Cu–Ag core–shell NPs and the interface between the Cu–Ag core–shell NPs and Cu substrates. Thus, Cu–Ag core–shell NP paste could be one of most promising interconnection materials for the future, low cost, high power electronics.

Acknowledgements

The authors are grateful for financial support from the National Natural Science Foundation of China (Grant No. 51522503) and support from Program for New Century Excellent Talents in University (NCET-13-0175).

References

1. J. G. Bai, J. Yin, Z. Zhang, G. Q. Lu and J. D. van Wyk, IEEE Trans. Adv. Packag., 2007, 30, 506–510.
2. Y. Zhong, R. An, C. Wang, Z. Zheng, Z. Q. Liu, C. H. Liu, C. F. Li, T. K. Kim and S. Jin, Small, 2015, 11, 4097–4103.
3. S. Rtimi, R. Sanjines, C. Pulgarin and J. Kiwi, ACS Appl. Mater. Interfaces, 2016, 8, 56–63.
4. R. Z. Li, A. Hu, D. Bridges, T. Zhang, K. D. Oakes, R. Peng, U. Tumuluri, Z. Wu and Z. Feng, Nanoscale, 2015, 7, 7368–7377.
5. Y. Jo, S. J. Oh, S. S. Lee, Y. H. Seo, B. H. Ryu, J. Moon, Y. Choi and S. Jeong, J. Mater. Chem. C, 2014, 2, 9746–9753.
6. Q. Jiang and F. G. Shi, J. Mater. Sci. Technol., 1998, 14, 171–172.
7. S. Wang, M. Li, H. Ji and C. Wang, Scr. Mater., 2013, 69, 789–792.
8. A. Russo, B. Y. Ahn, J. J. Adams, E. B. Duoss, J. T. Bernhard and J. A. Lewis, Adv. Mater., 2011, 23, 3426–3430.
9. Y. Li, Y. Wu and B. S. Ong, J. Am. Chem. Soc., 2005, 127, 3266–3267.
10. P. Peng, A. Hu, A. P. Gerlich, G. Zou, L. Liu and Y. N. Zhou, ACS Appl. Mater. Interfaces, 2015, 7, 12597–12618.
11. Z. Jiang, Y. Tian and S. Ding, Mater. Lett., 2014, 136, 310–313.
12. S. Ding, J. Jiu, Y. Tian, T. Sugahara, S. Nagao and K. Suganuma, Phys. Chem. Chem. Phys., 2015, 17, 31110–31116.
13. S. Jeong, H. C. Song, W. W. Lee, S. S. Lee, Y. Choi, W. Son, E. D. Kim, C. H. Paik, S. H. Oh and B. H. Ryu, Langmuir, 2011, 27, 3144–3149.
14. R. Watanabe and T. Ishizaki, Part. Part. Syst. Charact., 2014, 31, 699–705.
15. Y. Kobayashi, T. Shirochi, Y. Yasuda and T. Morita, Solid State Sci., 2011, 13, 553–558.
16. C. Lee, N. R. Kim, J. Koo, Y. J. Lee and H. M. Lee, Nanotechnology, 2015, 26, 455601.
17. C. K. Kim, G. J. Lee, M. K. Lee and C. K. Rhee, Powder Technol., 2014, 263, 1–6.
18. Y. Liu and Y. Sun, Nanoscale, 2015, 7, 13687–13693.
19. J. P. Lee, D. Chen, X. Li, S. Yoo, L. A. Bottomley, M. A. EI-Sayed, S. Park and M. Liu, Nanoscale, 2013, 5, 11620–11624.
20. A. I. Mares and J. M. van Ruitenbeek, Phys. Rev. B: Condens. Matter Mater. Phys., 2005, 72, 205402.
21. M. Li, Y. Xiao, Z. Zhang and J. Yu, ACS Appl. Mater. Interfaces, 2015, 7, 9157–9168.
22. E. Butovsky, I. Perelshtein and A. Gedanken, J. Mater. Chem., 2012, 22, 15025–15030.
23. M. Tsuji, S. Hikino, Y. Sano and M. Horigome, Chem. Lett., 2009, 38, 518–519.
24. M. Tsuji, S. Hikino, R. Tanabe, M. Matsunaga and Y. Sano, CrystEngComm, 2010, 12, 3900–3908.
25. B. K. Park, S. Jeong, D. Kim, J. Moon, S. Lim and J. S. Kim, J. Colloid Interface Sci., 2007, 311, 417–424.
26. Y. Tang, W. He, S. Wang, Z. Tao and L. Cheng, CrystEngComm, 2014, 16, 4431–4440.
27. Z. Jiang, Y. Tian, S. Ding, J. Wen and C. Wang, CrystEngComm, 2016, 18, 1200–1206.
28. Y. Ji, S. Yang, S. Guo, X. Song, B. Ding and Z. Yang, Colloids Surf., A, 2010, 372, 204–209.
29. Z. Z. Fang and H. Wang, Int. Mater. Rev., 2013, 53, 326–352.
30. J. Yan, G. Zou, A. Wu, J. Ren, A. Hu and Y. N. Zhou, J. Electron. Mater., 2012, 41, 1924–1930.
31. T. Ishizaki and R. Watanabe, J. Mater. Chem., 2012, 22, 25198–25206.
32. J. Yan, G. Zou, A. Hu and Y. N. Zhou, J. Mater. Chem., 2011, 21, 15981–15986.
33. Y. Morisada, T. Nagaoka, M. Fukusumi, Y. Kashiwagi, M. Yamamoto and M. Nakamoto, J. Electron. Mater., 2010, 39, 1283–1288.
34. T. Hu, H. Chen and M. Li, Mater. Des., 2016, 108, 383–390.
35. J. Luo, Crit. Rev. Solid State Mater. Sci., 2007, 32, 67–109.
36. A. Muzikansky, P. Nanikashvili, J. Grinblat and D. Zitoun, J. Phys. Chem. C, 2013, 117, 3093–3100.
37. J. G. Dash, A. W. Rempel and J. S. Wettlaufer, Rev. Mod. Phys., 2006, 78, 695–741.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra16474a

---