The relation between residual stress, interfacial structure and the joint property in the SiO2f/SiO2-Nb joints

In order to achieve a high-quality joint between SiO2f/SiO2 and metals, it is necessary to address the poor wettability of SiO2f/SiO2 and the high residual stress in SiO2f/SiO2-Nb joint. Here, we simultaneously realize good wettability and low residual stress in SiO2f/SiO2-Nb joint by combined method of HF etching treatment and Finite Element Analysis (FEA). After etching treatment, the wettability of E-SiO2f/SiO2 was improved, and the residual stress in the joint was decreased. In order to better control the quality of joints, efforts were made to understand the relationship between surface structure of E-SiO2f/SiO2 and residual stress in joint using FEA. Based on the direction of FEA results, a relationship between residual stress, surface structure and joint property in the brazed joints were investigated by experiments. As well the FEA and the brazing test results both realized the high-quality joint of E-SiO2f/SiO2-Nb and the shear strength of the joint reached 61.9 MPa.

Our latest work showed that a 3D-pinning structure was beneficial to the joint strength. However, the structure introduced the residual stress among the braided quartz fibers, which complicated the distribution of the residual stress in the joint. Furthermore, the complex structure of the 3D SiO 2f /SiO 2 -metal gradient transition zone had a decisive influence on the mechanical property of joint. Thus, the relationship between the residual stress, the surface structure and joint property in the brazed joints need be explored in depth and in detail.
In this paper, the etching treatment with HF acid solution was designed to regulate the surface structure of SiO 2f /SiO 2 . Finite element analysis (FEA) models were developed to study the relationship between the residual stress and the surface structure in the brazed joints, which can guide the subsequent experiment and provide the theoretical basis. Furthermore, the relationship between the residual stress, the surface structure and the joint property in the brazed joints was discussed in detail.

Experimental procedures
Materials. The 3D four-directional braided SiO 2f /SiO 2 29 and commercially available Nb were used as the parent materials. The dimension of SiO 2f /SiO 2 brazing specimen was 5 mm × 5 mm × 3 mm. Nb was cut into 10 mm × 10 mm × 3 mm slices for the microstructure observation and 10 mm × 15 mm × 3 mm for shear tests, respectively. The active brazing alloy foil Ag-21Cu-4.5Ti (wt.%) with a thickness of 200 μm was used to braze SiO 2f /SiO 2 and Nb. The bonding surfaces of samples were ground up to 400 by SiC sandpaper. All materials were ultrasonically cleaned in acetone for 15 min.
The etching treatment on the surface of SiO 2f /SiO 2 was performed with 20 wt.% HF acid solution. The schematic of the etching process is shown in Fig. 1. Firstly, the HF acid solution was directly placed on the surface of SiO 2f /SiO 2 for a few seconds. Then, the surface was washed with deionized water carefully. Finally, SiO 2f /SiO 2 with etching treatment (E-SiO 2f /SiO 2 ) was obtained. In addition, by controlling etching process, the surface structure dimension can be tunable.
Wetting and brazing processes. Wetting experiments were performed using the sessile drop technique in which the alloys are placed on the substrate and the system is heated to 840 °C. For comparison, during the sessile drop experiments, AgCuTi brazing alloy foils were placed on SiO 2f /SiO 2 with different infiltration depth, respectively. And the brazing experiments were performed with AgCuTi active brazing alloy foils between the parent materials. The structure of the assembly was SiO 2f /SiO 2 /AgCuTi/Nb, and the assemblies were held by graphite jigs. In order to keep the specimens in close contact, a load of 0.01 MPa was applied. The assemblies were heated to 840 °C with a rate of 10 °C min −1 in a vacuum furnace, isothermally held for 10 min, and then cooled down to room temperature at a rate of 5 °C min −1 .
The drop images, which were produced by an optical system coupled with a zoom (magnification 30×), were recorded by a video camera connected to a computer, permitting automatic image analysis. This device enables the contact angles of the drop were measured with an accuracy of ±2°. The interfacial microstructures of the joints were analyzed by a scanning electron microscopy (SEM) fitted with an energy dispersive spectroscopy (EDS). To identify the phases formed in the reaction layers adjacent to SiO 2f /SiO 2 and fracture, a JDX-3530M X-ray diffraction (XRD) was used. To evaluate the mechanical properties of the joints, shear tests were carried out using an Instron-1186 universal testing machine at room temperature. The average stress strength was identified by five shear specimens brazed under the same condition.
FEA calculations. The FEA method was employed to investigate the distribution of the residual stress along the SiO 2f /SiO 2 -Nb brazed joint in our research, because the method was proved to be a useful tool for predicting residual stress in the joint 30,31 . Thus, in this paper, the distribution of the residual stress, which yielded in the

Results and Discussion
Microstructure of the joint brazed with and without etching treatment. The interface analysis was first performed on the brazing joints before and after etching treatment, evidencing the establishment and the application of model for FEA. The typical microstructure of SiO 2f /SiO 2 -Nb joint is shown in Fig. 2a. It can be found cracks formed at SiO 2f /SiO 2 side, which the typical interface structure of the joint was SiO 2f / SiO 2 /Ag(s,s) + Cu(s,s)/Cu 3 Ti 3 O/TiSi 2 , based on our latest research. And the mismatch of CTE between SiO 2f /SiO 2 (CTE SiO2f/SiO2 = ~2.0 × 10 −6 /K) and Nb (CTE Nb = ~7 × 10 −6 /K) or AgCuTi active brazing alloy (CTE AgCuTi = ~15.4 × 10 −6 /K) is high, which results in forming cracks 32 . Figure 2b shows the typical microstructure of E-SiO 2f /SiO 2 -Nb joint. Compared with SiO 2f /SiO 2 -Nb joint, the primary compositions of beam were nearly same. The brazing alloy infiltrated into E-SiO 2f /SiO 2 and formed a "3D-pinning structure", which contributed to form a good CTE gradient transition and to reduce its mismatch between different materials 27,33 . Consequently, the joint exhibited sound bonding without any defect and crack.
In order to further investigate the interfacial microstructure of the E-SiO 2f /SiO 2 -Nb joint, the main elements distribution of the joint produced at 840 °C for 10 min are analyzed, as shown in Fig. 3. The Fig. 3a clearly presents that a sound joint has been obtained. Figure 3b-f shows the distribution of Ag, Cu, Ti, Si and Nb, respectively. It can be seen that the Si had a strong tendency to extremely react with Ti, as shown in Fig. 3d and e. In addition, notice that Ti-rich granular were formed adjacent to SiO 2f /SiO 2 composite, revealed that Ti segregated in 3D SiO 2f /SiO 2 -metal gradient transition zone. Furthermore, the distribution of Ti in that zone was not even because the brazing alloy gradually infiltrated into E-SiO 2f /SiO 2 , and Ti reacted with the contacted quartz fibers. Thus, the remaining Ti became less and less as the infiltration depth increasing. Moreover, it was important to note that SiO 2f /SiO 2 and Nb did not spread or dissolve during brazing, as shown in Fig. 3b and f, respectively. Therefore, based on the above results, the model for FEA can be developed as three parts: SiO 2f /SiO 2 (or E-SiO 2f /SiO 2 ), AgCuTi brazing alloy and Nb (see Fig. 4). It was worth noting that the special structure of the E-SiO 2f /SiO 2 by the etching treatment needed a completely new design system (Details on the model were shown in supplementary material).
Estimation of residual stress in the brazed joint using FEA. Recently, many researches have focused on reducing the residual stress in the composite/ceramic and metal brazed joints [34][35][36][37] . In fact, it is very difficult to measure the residual stress of the brazed joints directly through the experimental measurement. Thus, it is general to analyze the residual stress of the brazed joints from the joint fracture path. However, this analysis only horizontal contrast (trend), cannot be quantified (experimental value) contrast. In addition, when the fracture path exists in the same area, it is difficult to analyze the residual stress of composite/ceramic and metal brazed joints. In order to analyze the residual stress variation better, some researchers always investigated the distribution of residual stress by Finite Element Analysis (FEA) 38,39 . In our case, the fracture path is different before and after etching treatment 40 . Especially, after etching treatment, fractures all exist close to the etching area, then it is not accurate to analyze the residual stress through the fracture path. Therefore, based on the typical experimental results, the values of residual stress of the samples with varied etching depth can be examined by FEA method.
Based on the analysis in 3.1 section, the FEA was applied to simulate the distribution of residual stress in the brazed joint, and the details on simulation process were shown in supplementary material. The etching depth was related to the dimension of "3D-pinning structure" which directly affected the residual stress in the joint. Therefore, it is important to investigate the relationship between the surface structure (that is the layer thickness of 3D-pinning structure) and the residual stress in a brazed joint, which can provide the theoretical basis for the following brazing experiments.  joints brazed by AgCuTi at 840 °C for 10 min. It is obvious that the residual stress has been gradually changed with the etching depth increasing from 0 μm to 150 μm. As for 0 μm@E-SiO 2f /SiO 2 -Nb joint (that is SiO 2f /SiO 2 -Nb joint), the residual stress is constrained around the 0 μm@E-SiO 2f /SiO 2 -AgCuTi interface, and the peak residual stress of 380 MPa (see Fig. 6) generates in the SiO 2f /SiO 2 side close to brazing alloy, and then gradually decreased along the vertical direction of 0 μm@E-SiO 2f /SiO 2 -AgCuTi interface. After etching treatment, the higher residual stress in E-SiO 2f /SiO 2 -Nb joints has transferred in the "3D-pinning structure", which suggested that the structure played a key role in the distribution of the residual stress in the joints (see Fig. 5b-f). Moreover, it is worth noting that the maximum residual stress of E-SiO 2f /SiO 2 -Nb joints has transferred on the braided quartz fibers in the "3D-pinning structure", as shown in Fig. 5b-f. Furthermore, it clearly presents that the residual stress in the joints and on the braided quartz fibers were both reduced with the etching depth increasing. However, when the etching depth was too thick, the residual stress increased significantly, especially on the braided quartz fibers, as shown  in Fig. 5d-f. From the above results, it can be inferred that the "3D-pinning structure" can effectively reduce the residual stress and change the distribution of residual stress in the joints.
As for the 0 μm@E-SiO 2f /SiO 2 -Nb joints, the high CTE mismatch may cause the residual stress constraining around the SiO 2f /SiO 2 -AgCuTi interface. After etching treatment, a "3D-pinning structure" formed in the joint, which was contributed the brazing alloy infiltrating into the E-SiO 2f /SiO 2 side, forming a 3D SiO 2f /SiO 2 -metal gradient transition zone. The zone was beneficial to reduce the residual stress induced by the high mismatch of the dissimilar substrates. However, the zone also made the distribution of the residual stress in "3D-pinning structure" complicated, as shown in Fig. 5. In particular, when the etching depth further increased, the residual stress rather than reduced. In order to analyses the reason, we explored the relationship between the residual stress of joint in different directions and surface structure of the E-SiO 2f /SiO 2 by FEA. The schematic diagram of the profile of the model used in simulation procedure is shown in Fig. 7a. According to the structure of the E-SiO 2f /SiO 2 -Nb joints and our calculation results, it can be inferred that principal stress σ z changed with the etching depth increasing, but σ x and σ y did not or very little change. Another significant stress was shear stress τ xy , which latter, in combination with σ z , can induce fracture of the quartz fibers 41 . In addition, shear stress τ xz and τ zy changed very little with the etching depth increasing. So, the following analysis only focused on the largest principal stress σ z and shear stress τ xy in E-SiO 2f /SiO 2 -Nb joints. Figure 7b and c show the maximal σ z and τ xy in zone A of E-SiO 2f /SiO 2 -Nb joints, respectively. It can be observed that after etching treatment, σ z reduced markedly (from 0 to 100 μm), but a little varied with the etching depth increasing to 150 μm, as shown in Fig. 7b. In contrast, it is noteworthy that τ xy significantly increased with the etching depth further increasing (from 100 to 150 μm), as shown in Fig. 7c. Thus, the maximal resultant force in the zone descended first (from 0 to 100 μm) and then ascended (from 100 to 150 μm), as shown in Fig. 6. From the above results, it can be concluded that with the  Scientific RepoRts | 7: 4187 | DOI:10.1038/s41598-017-04531-w etching depth between 0 to 100 μm, the residual stress of the joint can be reduced, because of the formed 3D SiO 2f / SiO 2 -metal gradient transition zone, which was contributed to decrease the CTE mismatch. Nevertheless, with the etching depth further increasing (from 100 to 150 μm), the residual stress was no lower, but higher, due to the τ xy which was introduced by the zone and posed serious problems for the joint. Therefore, it can be concluded that although the "3D-pinning structure" can reduce the residual stress, the dimension of the structure should be in proper domain.
Effect of surface structure on the wettability of SiO 2f /SiO 2 composite. It is well known that the wettability of SiO 2f /SiO 2 plays an important role in obtaining a high-quality brazing joint 42,43 . Therefore, it is necessary to investigate the effect of etching treatment on the wettability of AgCuTi brazing alloy on SiO 2f /SiO 2 surface. Figure 8 shows the contact angles (CA) of AgCuTi brazing alloy on the surface of 0 μm@E-SiO 2f /SiO 2 , 50 μm@E-SiO 2f /SiO 2 , 75 μm@E-SiO 2f /SiO 2 , 100 μm@E-SiO 2f /SiO 2 , 125 μm@E-SiO 2f /SiO 2 and 150 μm@E-SiO 2f / SiO 2 , respectively. It can been observed that the brazing alloy on SiO 2f /SiO 2 shows unsymmetrical round shape, with left side showing smaller contact angles compare to the right one. And there are two main reasons for it. Firstly, the SiO 2f /SiO 2 brazing specimen was obtained through cutting into slices and grounding the bonding surface. So, after that treatment, the bonding surface of SiO 2f /SiO 2 was not flat, which led to the surface of E-SiO 2f / SiO 2 uneven, even the slope. Secondly, the surface stability of liquid brazing alloy on E-SiO 2f /SiO 2 was affected in vacuum furnace by mechanical pump and molecular pump during vacuumizing. Therefore, the brazing alloy on the surface of E-SiO 2f /SiO 2 showed unsymmetrical round shape in Fig. 8. However, the left contact angles (CA) of AgCuTi brazing alloy on the E-SiO 2f /SiO 2 decreased from 134° to 32° and the right CA decreased from 138° to 36°, as shown in Fig. 2. The CA decreased with the etching depth increasing, though the brazing alloy on SiO 2f / SiO 2 shows unsymmetrical round shape. Then, in our case, the right CA acted as the evaluation standard. Figure 8a shows the CA of AgCuTi brazing alloy on the surface of SiO 2f /SiO 2 was up to 138°, indicating poor wettability. After etching treatment, Fig. 8b shows the CA of AgCuTi brazing alloy on the 50 μm@E-SiO 2f / SiO 2 was significantly decreased to 53°, meaning favorable wettability. Furthermore, it can be seen that the CA decreased from 53° to 36° with the etching depth increasing. As a result, it suggested that after etching treatment, the wettability of all E-SiO 2f /SiO 2 was excellent. In order to further explore the reasons for the improving wettability, the microstructures of wetting region were analyzed in details, as shown in Fig. 9. From Fig. 9a, it can be observed the weak joining between SiO 2f /SiO 2 and brazing alloy, because of the poor wettability of SiO 2f /SiO 2 . In contrast, it can be observed that after etching treatment, the brazing alloy infiltrated into E-SiO 2f /SiO 2 and sound metallurgical bonding was formed between brazing alloy and braided quartz fibers (see Fig. 9b). Furthermore,  the infiltration depth was gradually increased with etching depth increasing, as shown in Fig. 9. After etching treatment, only the silica sol was consumed and numerous quartz fibers were left on the surface of SiO 2f /SiO 2 17 . Based on the above results, we believe that the poor wettability of SiO 2f /SiO 2 can be owing to the silica sol, and the wettability between brazing alloy and quartz fibers was extremely well. It suggested that the etching treatment was an easy and effective way to improve the wettability of SiO 2f /SiO 2 . Thus, it can be inferred that the space of the consumed silica sol was able to be filled up with sufficient brazing alloy.
Generally, the wettability of the materials with the same surface state is almost identical under the same condition. The surface of SiO 2f /SiO 2 , after etching treatment, is all the quartz fibers were left, which are the same regardless of the etching depth. However, it is worth noting that the CA of AgCuTi brazing alloy on the surface of E-SiO 2f /SiO 2 decreased with the etching depth increasing, as shown in Fig. 8. In order to illustrate the wetting process, a concept physical model was established, as shown in Fig. 10. After etching treatment, the fused silica was consumed and quartz fibers were left in the transition zone, which contributed to the brazing alloy infiltrating into E-SiO 2f /SiO 2 . Furthermore, the width of the transition zone increased with the etching depth increasing, and then less and less brazing alloy was left on the surface of E-SiO 2f /SiO 2 , as shown in Fig. 10. Thus, the CA of E-SiO 2f /SiO 2 decreased with etching depth increasing from the wetting experiments results. In fact, the CA of E-SiO 2f /SiO 2 was constant. According to the above results, it is reasonable to infer that the space of the consumed silica sol can be filled up with brazing alloy, as long as it was enough.   Fig. 11. It is obvious that the interfacial microstructure changed with the depth of brazing alloy infiltrating into SiO 2f /SiO 2 increasing. As for 0 μm@E-SiO 2f /SiO 2 -Nb joint, continuous cracks can be observed in SiO 2f /SiO 2 side near the brazing interface (see Fig. 11a). It may be because the poor wettability of SiO 2f /SiO 2 and the high residual stress induced by the CTE mismatch between SiO 2f /SiO 2 and Nb. So, the shear stress of the 0 μm@E-SiO 2f /SiO 2 -Nb joint was only 5 MPa (see Fig. 12). When SiO 2f /SiO 2 after etching treatment, the brazed joint exhibited a 3D SiO 2f /SiO 2 -metal gradient transition zone and the width of the zone increased with the etching depth increasing (see Fig. 11b-d). Further, the brazing alloy was able to fill up the space between the quartz fibers and formed a sound metallurgical bonding with them, due to the great wettability of E-SiO 2f /SiO 2 . Furthermore, the "3D-pinning structure" can effectively reduce the residual stress in the joint, which decrease the continuous cracks and strengthen the joint. Thus, the shear strength of the joints increases from 5 MPa to 61.9 MPa with the etching depth increasing from 0 to 100 μm. However, the shear strength of the joints decreased with the etching depth increasing over 100 μm, because of the residual stress in the joints was increasing by the τ xy of the "3D-pinning structure", according to the FEA results (see Fig. 12).    In order to further study the effect of the transition zone on the obtained joints, the fracture analysis was conducted. Figure 13 shows the fracture surface of joints after shear tests. It can be seen that only remained broken quartz fibers were left in the fracture of SiO 2f /SiO 2 -Nb joint, as shows in Fig. 13a, which may be due to the cracks at SiO 2f /SiO 2 side caused by the high CTE mismatch. By contrast, almost the same type of fracture morphology for the E-SiO 2f /SiO 2 -Nb joints can be observed in Fig. 13b-d. The brazing alloy infiltrated into E-SiO 2f /SiO 2 can be observed in fractures. In addition, XRD was used to confirm the reaction products at the fracture surface. As shown in Fig. 14a, it can be seen that the fracture of SiO 2f /SiO 2 -Nb joint was only composed of amorphous silicon dioxide. Correspondingly, amorphous silicon dioxide, TiSi 2 , Ag(s,s) and Cu(s,s) were the reaction phases on the SiO 2f /SiO 2 composite fracture side for the fractures of 50 μm@E-SiO 2f /SiO 2 -Nb, 100 μm@E-SiO 2f /SiO 2 -Nb and 150 μm@E-SiO 2f /SiO 2 -Nb joints (see Fig. 14b-d). According to the XRD patterns, it can be inferred that as for SiO 2f /SiO 2 -Nb joint, fracture occurred along the brazing interface on the SiO 2f /SiO 2 composite side. It may be due to the residual stress constraining around the SiO 2f /SiO 2 -AgCuTi interface. Moreover, for 50 μm@E-SiO 2f / SiO 2 -Nb, 100 μm@E-SiO 2f /SiO 2 -Nb and 150 μm@E-SiO 2f /SiO 2 -Nb joints, fracture occurred in the 3D SiO 2f / SiO 2 -metal gradient transition zone. Although the morphologies of the three kinds of fractures were almost the same, the mechanical properties of the joints were different, which may be owing to the residual stress concentrated in that zone.
Based on the above results, it is suggested that the FEA results can serve as a guide for the brazing test and a theoretical basis for the distribution of residual stress in the brazed joints. In addition, the results show clearly that the etching treatment plays a key role in two major aspects on active brazing SiO 2f /SiO 2 and Nb. On the one hand, it can effectively improve the wettability of SiO 2f /SiO 2 , which is a precondition of successfully brazing SiO 2f / SiO 2 and Nb. On the other hand, it can induce the brazing alloy infiltrating into E-SiO 2f /SiO 2 , which increases the bonded area, reduces the residual stress and enhances the mechanical properties. Therefore, regulating the surface structure is an easy and effective method to reduce the residual stress, which can strength the joints. As well the FEA and the brazing test results both reveal that the properties of 100 μm@SiO 2f /SiO 2 -Nb joint is the best among the obtained brazed joints with different surface structures.

Conclusions
In this paper, the optimized depth of brazing alloy infiltrating into SiO 2f /SiO 2 was achieved by combining FEA with experiments. After etching treatment, the fused silica with poor wettability has been consumed while the quartz fibers with good wettability were left, thus the wettability of E-SiO 2f /SiO 2 was improved. The good wettability of SiO 2f /SiO 2 played an important role in obtaining a high-quality joint. Moreover, "3D-pinning structure" formed in E-SiO 2f /SiO 2 -Nb joints, which can reduce the residual stress in the joint by form the sound gradient transition of CTE. However, the residual stress rather reduced with the etching depth increasing over appropriate size, due to the τ xy introduced by the "3D-pinning structure". A relationship between the residual stress, the surface structure and the joint property in the brazed joints was demonstrated. As well the FEA and the brazing test results both realized the high-quality joint of E-SiO 2f /SiO 2 -Nb joint and the shear strength of the joint reached 61.9 MPa, which was approximately 12 times than that of SiO 2f /SiO 2 -Nb joint.