3.1 Characterization of Bi-B-Zn glass
To sinter the silver paste at low temperatures, the glass used to prepare the silver paste should preferably be amorphous. The X-ray diffraction pattern of the Bi-B-Zn glass powder is shown in Fig. 2a. There were only two broad peaks and no crystal phase peaks detected, which indicated the Bi-B-Zn glass powder was amorphous. The glass transition temperature (Tg = 353 ℃) and the crystallization temperature (Tc = 448 ℃) of Bi-B-Zn glass were measured by a DSC analyzer, and are shown in Fig. 2b.
3.2 Preparation and characterization of silver paste
The silver paste was prepared using a planetary gravity mixer by mixing Bi-B-Zn glass powder, silver particles, and an organic carrier with a ratio of 17:69:14 (wt. %). Different organic additives were used to disperse the glass powder and the silver particles and to modify the rheological properties. Terpineol (C10H18O) was used as a solvent, while ethyl cellulose ([C6H7O2(OC2H5)]n) was used as a thickener to change the viscosity of the organic carrier for adjusting the rheology and formability of the silver paste; ethyl acetate (C4H8O2) was used as a dispersant to ensure that the components in the silver paste were dispersed and mixed evenly, while 1,2-propylene glycol (C3H8O2) served as an organic solvent and stabilizer. C3H8O3 (GP-330) was used as a defoamer to prevent the generation of bubbles in the silver paste after printing. The specific ratios of the different types of organic additives are given in Table 1.
Table 1 Composition and content of the organic vehicle
Component
|
C10H18O
|
[C6H7O2(OC2H5)]n
|
C4H8O2
|
C3H8O2
|
C3H8O3
|
Content (wt.%)
|
65.5
|
3
|
14.8
|
9.3
|
7.4
|
The DSC and thermogravimetric (TG) analysis of the low-temperature sintered silver paste is shown in Fig. 3. On one hand, the addition of organic carriers with different decomposition and volatilization temperatures played a role in adjusting the viscosity, on the other hand, it could prevent the printed pattern from collapsing due to a large amount of volatilization at a given temperature. The TG curve indicated that the weight of the silver paste had significant decreases at 100 ℃, 200 ℃, and 300 ℃, and remained almost unchanged above 300 ℃. Therefore, the debinding process should be carried out in three temperature sections: at 100 ℃, 200 ℃, and 300 ℃ held for 60 minutes with a rate of 2.5 ℃/min, respectively.
3.3 The microstructure of bondlines
A cross-section image of a typical bondline formed after sintering at 450 ℃ for 10 min is shown in Fig. 4. The silver film acted as a metallized layer on the tempered glass surface which was tightly bound to the tempered glass, and the tin alloy solder layer (SAC305) was between the two silver film layers. The following work will focus on the interface and sintering behavior of the thick silver film.
The cross-section images of the bondlines with different glass content (wt. %) sintered at 450 ℃ for 10 min are shown in Fig. 5. Two clear interfaces were observed between the silver film and the SAC305 solder Fig. 5(a), and between the silver film and the tempered glass substrate Fig. 5(d). The results indicated that the glass was crucial for the interfacial bonding between the silver film and the tempered glass substrate, and between the silver film and the SAC305 solder. The silver film was tightly bound to the tempered glass substrate and there was no separation at the interface with either 40 wt.% or 30 wt.% glass content. More glass exposure on the silver film surface caused a non-wetting of the SAC305 solder and the silver film, which led to the separation of the silver film and the SAC305 solder interface Fig. 5(a). Compared with the silver films with either 20 wt.% or 10 wt.% glass content, the transition layer was not noticeable, and the interdiffusion of silver and tin was hindered by a large amount of glass. Although the wettability of the SAC305 solder on the silver film was good and a transition layer was formed, there were some non-connected holes in the silver film, because it was difficult for the glass to fill these holes. Due to the smaller amount of glass flowing to tempered glass, there was an obvious interface separation between the silver film and the tempered glass with only 10 wt.% glass content Fig. 5(d). This seriously degraded the interface bonding between the silver film and the tempered glass substrate. In summary, the glass content greatly affected the density of the sintered silver film and the bonding strength with the tempered glass substrates. Due to the presence of the glass, silver microparticles were rearranged during the glass softening process. The silver microparticles were not only used in solid-phase sintering, but also diffused through grain boundaries to form a sintered neck. The glass also promoted the sintering of the silver microparticles and molten glass penetrated the gap between the sintering necks through capillary action.
Although the glass content was relatively small, the dissolution in glass and the precipitation process of the silver particles could promote sintering and accelerate the densification process [25,26]. To confirm that silver could dissolve in the Bi-B-Zn glass and precipitate out during the cooling process, the microstructure of the sample was further analyzed by SEM and TEM. There were nanoscale microparticles on the silver film surface. The high-angle annular dark field (HAADF) image near the interface between the glass and silver microparticles, and the elemental distributions of Ag and Bi are shown in Fig. 6. The silver particles gradually sintered together and grew larger during the sintering process. There were no nano-sized silver particles observed in the original material. However, some nanoparticles were detected on the silver film, so this result indicated that the silver dissolved into the glass and nanocrystals precipitated. It could be inferred that silver dissolved into the Bi-B-Zn glass during sintering and penetrated into the glass grid microstructure. Then, the silver dissolved in the glass and precipitated in a supersaturated state and finally formed silver nanoparticles on the surface of the glass during the cooling process. The silver diffused into the liquid glass and the precipitation process could accelerate the diffusion rate of the silver, which was equivalent to accelerating the sintering growth process of the silver microparticles. The mixed solid and liquid phase sintering process could be divided into three steps: (1) Formation of the liquid phase and rearrangement of particles: the glass softened and the liquid phase formed when the glass powder was heated to the softening temperature. The silver particles were mostly suspended within the liquid phase and moved collectively with the help of the surface tension of the liquid phase; (2) Solid-phase sintering and liquid-phase filling: the contact of the silver particles with each other through diffusion sintering, gradually formed a sintered neck and grew, and finally sintered into a massive silver layer. The molten glass filled some voids because of capillary action and its fluidity, thereby the microstructure of the thick silver film rapidly densified; (3) Dissolution and reprecipitation of the solid phase: the angular and convex portions on the silver particles’ surfaces preferentially dissolved into the molten glass. Then the supersaturated silver ions in the molten glass precipitated during the cooling process, and the nanocrystalline silver particles reached the silver particles and accelerated the growth of the silver microparticles.
The interface between the glass and the tempered glass, along with the interface between the silver and the glass were characterized by TEM, as shown in Fig. 7. The high angle annular dark-field (HAADF) TEM image of the silver film sintered at 450 °C for 10 min is shown in Fig. 7(a). The EDS element mappings of the Bi, Zn, Ag, O, and Si elements are shown in Fig. 7(b-f), and Fig. 7(g-h) show the diffraction patterns of regions A and B shown in Fig. 7(a), respectively. The amorphous ring of region A indicated that the glass had an amorphous structure with a low degree of a local order near the substrate surface. The bright Bi-rich region indicated that the Bi4B2O9 phase (PDF # 70-1458) was crystal structure, which agreed well with the XRD results shown in Fig. 8. According to the crystallization temperature of 448 °C from the DSC analysis, Bi4B2O9 crystals precipitated during sintering process when the temperature was above 450 °C. The nucleation and growth of Bi4B2O9 were mainly by diffusion and aggregation of the corresponding elements during the cooling process. Thermal convection with the external environment and the transition of the glass phase to crystalline could provide enough energy for the crystallization process. The precipitation of an appropriate amount of crystalline phase contributed to the strengthening of the overall structure of the thick silver film, which also explained why the sample had the highest shear strength under the process of heating at 450 °C for 10 min.
3.4 The bonding mechanism
The bright-field image and the high-resolution image of region A are shown in Fig. 9(a) and (c). The diffusion of Si, Zn, Bi, and O at the glass and tempered glass substrate interfaces were measured using scanning line B, as shown in Fig. 9(b). It could be seen that the diffusion layer from the glass into the tempered substrate was about 40 nm. The element diffusion curves clearly showed that the content of Bi, Zn, and O changed slowly at the interface. However, the Si content reduced sharply, which could be concluded that the Si remained stable in the glass matrix, but Bi diffused into the tempered glass substrate, indicating an excellent bond was formed at the interface. A large amount of fine Bi4B2O9 grains precipitated at the interface and inhibited grain growth, because of the large nucleation energy exchange and consumption in the supercooled glass and the tempered glass substrate. The supercooled melt and a large number of nucleation sites were provided with the help of the surface. More fine nano-grains were formed because of the high cooling rate. Thermal convection and heat exchange provided the energy for the transition process from the amorphous at the interface to crystalline, thus promoting the continuous nucleation of nanocrystals. The X-ray diffraction pattern of the interface was composed of a diffractive halo and a polycrystalline ring, indicating that the interface was mainly composed of a glass phase and a large number of Bi4B2O9 grains. After the Bi diffused, which aggregated at the interface and broke through the nucleation barrier, a large number of Bi4B2O9 nanocrystals were generated and distributed evenly throughout the interface. A diffusion bond was formed between the glass substrate and the silver film. The precipitation of small grains could hinder the crack propagation in the glass and played a vital role in dispersion strengthening, thereby inhibiting the occurrence of fracture modes at the interface during shear fracture processes, which will greatly increase the adhesion and shear strength of the sintered silver film.
The fracture mode was mainly brittle fracture when glass content was 40 wt.%, as shown in Fig. 10(e). The cracks mainly extended along the inside of the glass or the surface of the tempered glass substrate because the silver particles were separated by large glass resulted in relatively low sintering dense. More glass existed at silver film and tempered glass substrate interface weakened silver particles on inhibiting crack propagation, which resulted in a low bonding strength of 14.8 MPa. The silver film was well bonded to the tempered glass substrate with 30 wt.% glass content, as shown in Fig. 10(b). Silver particles were peeled off from the glass in the fracture interface and the large brittle fracture mode disappeared. The crack expanded from the glass inside when the tip of the crack met the silver particles, so that the crack penetrated the silver film along the weakest part, and the shear strength could reach 30.1 MPa. Since the glass content in the thick silver film was high, the overall thick silver film still exhibited a certain degree of brittleness, but the silver particles were not separated in the glass, and some silver particles had been sintered to form a whole. The maximum shear strength of 42.3 MPa was obtained with 20 wt.% glass content. The silver film and the tempered glass substrate were well bonded and most of the silver particles were peeled from the glass, as shown in Fig. 10(c). Without a large brittle fracture formed, and visible bowl-shaped structure on the tempered glass substrate surface. The silver particles were sintered well and the gap between them was filled with glass solder, which promoted the bonding force between the silver film and tempered glass substrate. However, the shear strength had the lowest value of 12.8 MPa with 10 wt.% glass content. There was not enough molten glass solder to fill the interface between the silver film and the tempered glass substrate, which resulted in a lot of holes and left an interface that was not fully bonded, as shown in Fig. 10(d). A small amount of glass which existed at the interface was not enough to ensure that the Bi-B-Zn glass presented evenly at all interfaces, which meant that not all fracture interfaces had a bowl-shaped structure, which was described by Fig. 10(h).
One of the most important performance metrics of vacuum tempered glass is airtightness. Good air tightness and internal getter could maintain a high vacuum of the vacuum layer for a long duration. This experiment used the GJB 548B-2005 method 1014.2 test condition A1 (fine leak detection) to test the gas leakage rate of the seal component, as shown in Fig. 11(a). The leak rate was 7.2 × 10-3 Pa.cm3/s with 80 wt.% Ag silver paste sealed at 450 ℃ for 10 min. The single-layer glass, double-layer glass, hollow glass component, and tempered vacuum glass component were placed on a 200 ℃ heating table for the experiment. The temperature of the upper surface was measured with time and plotted, as shown in Fig.11(b). The experimental results showed that the heat transfer of the single-layer glass was the fastest, and the thermal equilibrium could be reached in 5 minutes, at a temperature of 200 ℃. Due to the heat dissipation from the surroundings and the upper surface to the environment, the final upper surface temperature was maintained at 144.5 ℃. When the double-layer glass and the hollow glass component were selected for the experiment, the heat transfer rate decreased significantly. Thermal equilibrium was reached in 15 minutes and the temperatures were 115.1 ℃ and 108.5 ℃, respectively. However, the gas heat conduction and convection heat transfer were negligible since the middle layer was close to the vacuum in the vacuum glass component. The experimental results also showed that the tempered vacuum glass components reached a thermal equilibrium temperature of 82.5 ℃ after 24 minutes, which had good thermal insulation performance since the main heat conduction mode was radiant heat transfer. Therefore, these results indicated that the vacuum glass components exhibited excellent thermal insulation performance, compared to hollow glass and other glass structures.