Optimization of process parameters and mechanism of strengthening and toughening of Nb-W alloy prepared by chemical vapor deposition based on orthogonal test

In this paper, the chemical vapor deposition (CVD) method was used to prepare Nb-W binary alloys on Mo substrates for the first time. The influences of the process parameters on the composition, average deposition rate, and average deposition efficiency of the Nb-W alloys were studied by orthogonal experiments. Nb-W alloys with W contents ranging from 0.33% to 50.48% were successfully prepared. The mechanical properties and fracture morphologies of the Nb-W alloys prepared by the CVD method (CVDNb-W alloys) were tested by metallographic microscopy, scanning electron microscopy, electron probe analysis, and in situ tensile tests. The results showed that the Nb-W alloy prepared by the CVD method had uneven macroscopic distributions of Nb and W.The substrate temperature had the greatest influence on the lateral composition gradient, and the H2 flow had the least influence. The influence of the Cl2 flow through the Nb and W on the average deposition rate was also examined. The chlorination temperature of W had the least effect on the average deposition rate, and the average deposition efficiency decreased with the increase in the gaseous chloride ratio of W. The effects of other factors on the average deposition efficiency showed different degrees of wave dynamics. The metallographic observation of 16 samples showed that except for samples 3# and 4# (in which the mass percentages of W were less than 1%), the microstructures of the other samples (in which the mass percentages of W were all more than 1%) showed similar microstructures with layered structural features. Comprehensive analysis of the sample and inverse pole figures revealed that the (101) 111¯ textural component mainly existed in the CVDNb-W alloys.With the increase in the mass percentage of W, the tensile strength of the CVDNb-W alloy increased correspondingly. The maximum strength of the alloy containing 9.68% W by mass reached 475 MPa, which exceeds the room-temperature tensile strength of Nb521, but the elongation at break was related to the alloy composition. Thus, the relationship did not show a discernible trend.

High-temperature of preparation process and low density of finished products Vacuum arc melting It is easy to add alloying elements, the alloy composition is controllable The microstructures and properties of different parts are quite different and the microstructure is uneven and uncontrollable Directional solidification It can better control the grain orientation of the solidified structure, eliminate most or even all of the transverse grain boundaries, and improve the longitudinal mechanical properties.
The equipment is complicated and the temperature gradient is difficult to accurately control.

Chemical vapor deposition
Good shape retention, high purity, uniform dense surface, good bonding, and controllable thickness Low preparation rate which is then poured into the mold for molding. The alloy composition of arc melting technology is controllable, and impurities are not easily introduced. The final material obtained has high purity, and the losses of the raw materials are low. However, the density of the material is poor, and the morphology and properties of different parts are quite different. Sikka and Rosa [10] prepared a Nb-10%W alloy by vacuum arc melting technology and studied its oxidation resistance. In directional solidification technology, a temperature gradient is developed in a specific direction between the solid metal and the unsolidified melt during the solidification of the liquid alloy so that the melt solidifies in the opposite direction to the heat flow, yielding columnar crystals with a specific orientation or single crystals [11]. Directional solidification technology can better control the grain orientation of the solidified structure, eliminate most or even all the transverse grain boundaries, and improve the longitudinal mechanical properties of the alloy.Liu Zhongyuan [12,13] of Northwestern Polytechnical University used a self-developed directional solidification device to prepare the superalloy DZ22. The dendrite segregation was suppressed and the dendrite arm spacing was refined by changing the cooling rate. The minimum arm spacing of the obtained dendrites could reach 8.4 μm, and the element segregation ratios were all close to 1. The high-temperature durability of DZ22 alloys can be greatly improved by directional solidification technology.
CVD is a technology that uses various energy sources, such as heating, plasma excitation, or light radiation, to make chemical substances in a gaseous or vapor state in a deposition chamber form solid deposits through chemical reactions in the gas phase or on the gas-solid interface. It is widely used for purifying substances, preparing thin films, growing crystals, and other applications [14]. The prepared product has the advantages of shape retention, high purity, a uniform and dense surface, and good bonding, and the thickness is controllable. For the deposition and preparation of refractory metals, the process temperature is much lower than the melting point of the material, and the finished product is pure and free of impurities. Compared with the traditional casting method and powder metallurgy, it has significant advantages. The preparation of single refractory metals by CVD has been extensively studied. Tungsten foils with thicknesses of only 10 μm were prepared under conditions with an evaporation temperature of 20°C and a deposition temperature of 300°C at the Moscow Institute of Refractory Metals and Cemented Carbides [15].Du Jihong [16] of the Northwest Institute of Nonferrous Metals successfully prepared a high-purity tungsten coating with an actual density of 99% and a purity of 99.9% and studied the effects of process parameters such as the deposition temperature and time on the deposition rate, coating thickness, and coating microstructure. Cai Hong-zhong [17][18][19] from the Kunming Institute of Precious Metals used pure tantalum as the initial metal source. The gaseous TaCl 5 generated after chlorination underwent a reduction reaction with H 2 in the deposition chamber to deposit a tantalum layer on the surface of a molybdenum substrate. The influences of the deposition parameters, such as the deposition temperature and reactive gas flow rate, on the microstructure of the tantalum layer were investigated. It was found that when the deposition temperature was in the range of 900°C-1300°C, with the increase in temperature, the columnar crystal structure increased, but the hardness and density decreased. Through the CVD of a single refractory metal, by selecting the type and proportion of the metal source precursor and the regulation of the deposition process, the composite deposition of a variety of refractory metal elements can be achieved to prepare alloys.
Continuous W-Mo, W-Ta, and W-Re alloy coatings have been obtained by CVD around the world [20,21]. However, the reports on the preparation of Nb-W alloys by CVD are very limited. As far as the composition of the binary system is concerned, Nb has a very low plastic-brittle transition temperature and good roomtemperature processing properties, and tungsten has a very high melting point and shear modulus. Both niobium and tungsten have body-centered cubic structures, and a continuous solid solution can be formed between them, that is, there is no intermediate phase over the entire composition range. The addition of tungsten to niobium can improve the high-temperature and room-temperature strength of niobium, and it has a significant strengthening effect on niobium [22].
In this study, the design idea of the Nb521 alloy was used as a reference. Tungsten was selected as the solidsolution strengthening element of the niobium alloy, and an orthogonal experiment table L 16 (4 5 ) was designed (that is, five factors, four levels, a total of 16 groups of experiments were performed). CVD technology was applied to attempt to prepare Nb-W alloys with similar or even better performances than those of the alloys obtained by traditional preparation technology. The effects of the CVD process parameters on the alloy composition, average deposition rate (deposition amount per unit time), and average deposition efficiency (deposition amount/chlorination amount) were determined, and the strengthening mechanism was explored.
1. Orthogonal experimental design 1.1. Calculation parameters There are many influencing factors in the preparation process of Nb-W alloys by CVD. The orthogonal method was used to design the experiments, and the experimental analysis was based on a statistical method (that is, when the relationship between each parameter and the index was studied, the result was based on the overall experiment macro-statistics, not on a single sample). Five factors were investigated: the chlorination temperature of W (A), substrate temperature (B), Cl 2 flow rate through Nb (C), Cl 2 flow rate through W (D), and H2 flow rate (E). Based on experience, four levels were selected for each factor. These representative horizontal combinations were 'evenly dispersed, neat, and comparable' (table 2). The specific experimental scheme was designed according to the L 16 (4 5 ) orthogonal table, as shown in table 2. The numbers in table 3 are the sample numbers, and a total of 16 groups of experiments were set up. To eliminate the human selection factor in the experiment process as much as possible, the order of experiments in the orthogonal experiment was generally selected by drawing lots (each experiment in the orthogonal experiment had the same status).

Starting materials
The starting materials used to prepare Nb-W alloys via the CVD method were pure niobium sheets (purity > 99.95%) and tungsten sheets (purity > 99.95%) using Cl 2 (purity > 99.6%) and H 2 (purity > 99.95%) as the reactive gases. The base material was metal molybdenum in the powder state, which was suitable for metallurgical processing into substrates with dimensions of Ø35 mm × 30 mm, and the surface of each matrix was treated the same to ensure the surface state was the same.

Experimental equipment and procedure
The schematic diagram of the apparatus used to deposit the Nb-W alloy coating on the molybdenum substrate is presented in figure 1. The deposition apparatus consisted of a chlorination chamber, deposition chamber, heating system, gas transport, and vacuum system. The chlorination chamber and the substrate after surface treatment were heated by an electric furnace and an intermediate-frequency induction furnace, respectively. The pressure of the deposition chamber was controlled by vacuum pumps, and the flow rates of Cl 2 and H 2 were controlled independently by mass flow controllers. The whole CVD system was evacuated, and the vacuum state was maintained for 24 h to ensure good sealing of the vacuum system. The substrate and the raw metal materials, niobium and tungsten, were heated to the required temperature under hydrogen gas during the heating process. When the substrate and the chlorination  chamber reached the set temperature, purified and dried chlorine gas was introduced into the chlorination chamber(figure 1), and the chlorine gas reacted with the niobium and tungsten in the chlorination chamber to generate niobium pentachloride (NbCl 5 ) and tungsten hexachloride (WCl 6 )(Formula1,2). The gaseous NbCl 5 and WCl6 were transported to the surface of the molybdenum substrate in deposition chamber, which was inductively heated to a certain temperature, and then they underwent reduction reactions with hydrogen to deposit niobium and tungsten(Formula3,4). The substrate was rotated to ensure a uniform thickness of the coating during the deposition process. The main chemical reactions were: The exhaust gas generated by the reaction was cooled and captured by liquid nitrogen in the filter bottle and then entered the mechanical pump discharge system.
The deposition layer samples of the Nb-W alloys were polished and eroded for microstructure observation using a 4XC metallographic microscope. A PHLIPS XL30 scanning electron microscope and SPM-9700 electron probe were used to observe and analyze the surface morphology of the deposited layer. The polished metallographic samples were immersed in an electrolyte with a composition of H 2 SO 4 :HNO 3 :HF = 5:2:2 for electrolysis, and electron backscatter diffraction (EBSD) characterization was carried out. Polished and etched samples were prepared with the dimensions shown in figure 2, and in situ tensile experiments were performed on a FEI-Tecnai G2 F30 field-emission transmission electron microscope using a MINI-MTS2000 single-tilt tensile stage.

Results and analysis
3.1. Analysis of deposition of CVDNb-W alloys 3.1.1. Macroscopic distribution characteristics of Nb-W alloy composition Figure 3 shows a schematic diagram of the sample after the deposition was complete. The composition distribution of the upper end face of the deposited matrix was relatively uniform, while there was a macroscopic composition gradient on the cylindrical surface (side) of the deposition matrix. This is mainly due to the limitations of the self-assembled CVD device in our group.The chlorine gas first passed through the chlorination chamber to form chloride and then reached the deposition substrate surface in the deposition chamber from top to bottom.The upper surface of the deposition substrate was the first to contact the reactive gas, and the gas concentration involved in the reaction will be greater than that of the side of the substrate, and the reaction would start first. Therefore, the composition of sediments at different positions of the substrate was different.To quantitatively analyze the distribution of components (mass percentage of W) in different parts of the sample surface, three regions labelled a, b, and c, were selected from top to bottom on the sides of the 16 samples for component determination. For the upper surface, region d around the center of the circle was selected for composition measurement. Compositional differences of varying degrees existed in different regions on the sides of different samples. To facilitate the analysis, the relative range was used to measure the size of the composition gradient on the side of each sample. The relative range of the composition is defined as follows: here X max , X min , and X _ represent the maximum., minimum, and average values of the components in areas a, b, and c on the side of the sample, respectively. Table 4 shows the composition measurement results and the relative ranges of the side surfaces in areas a, b, c, and d. The relative ranges of the side components are shown in table 5 for intuitive analysis, and the influence degree of each deposition parameter on the side composition is shown. The larger the relative range was, the  greater the lateral composition gradient of the prepared sample was. The maximum relative range corresponding to factor B (matrix temperature) was 0.36, and the minimum relative range corresponding to factor E (flow rate of H 2 ) was 0.065. The substrate temperature had the greatest influence on the lateral composition gradient, followed by the Cl 2 flow rate through W and the chlorination temperature of W. The Cl 2 flow rate through Nb and the H 2 flow rate had the least influence on the lateral composition gradient. According to the above results, three factors, B (substrate temperature), D (Cl 2 flow rate through W), and A (chlorination temperature of W), had the most significant influences on the composition gradient on the sample side. These three factors had a direct impact on the absolute content of W chloride, indicating that the lateral composition gradient of the matrix had a significant relationship with the absolute content of W chloride and was closely related to the absolute content of Nb chloride. However, it had little relationship with the absolute content of Nb chloride. This was mainly because under the selected chlorination conditions, the variation range of the chlorination rate of Nb (12-26 g h −1 ) was smaller than that of W (1.5-25 g h −1 ).
Since the composition of the upper surface was relatively uniform, the composition of the upper surface d region was used as the analysis index. According to the results in table 4, the visual analysis results of the calculated upper face components are listed in table 6. The order of influence of each factor on the Nb-W alloy composition was B,A > D,C,E, as shown in table 6 That is, the deposition temperature and the chlorination temperature of W had the greatest influence on the compositions of the Nb-W alloys, the other three factors had less influence. The same results were found for the composition gradient of the sample side, which confirmed the universality of this chemical vapor deposition characteristic for Nb-W alloys.

Average deposition rate range analysis
The average deposition rate is defined as the total mass of Nb-W alloy deposited on the Mo substrate per unit time, which is defined as:  Table 5. Intuitive analysis of relative ranges on sides of samples. The average deposition rate results of the 16 samples are listed in table 7, and the intuitive analysis results are listed in table 8. As shown in table 8, the Cl 2 flow rate through Nb (C) had the greatest effect on the average deposition rate, and the chlorination temperature of W (A) had the least effect on the average deposition rate. The other parameters had intermediate effects, and there was little difference between them. When the Cl 2 flow rate through Nb (C) was at the C1 level (100 ml min −1 ), the average deposition rate was the lowest, the average deposition rates of other level factors were higher. When the chlorination temperature of W(A) was at the A1 level (550°C), the average deposition rate was the highest, and the average deposition rates of other level factors were slightly lower. For the Cl 2 flow through W (D), the top three levels factors showed that the average deposition rate decreased with increasing Cl 2 flow.
Increasing the proportion of gaseous chloride in the Nb chamber was beneficial for the increase in the average deposition rate. In contrast, the higher the proportion of W gaseous chloride, the lower the average deposition rate. Under the conditions of this experiment, to maximize the average deposition rate, the factor level combination that should be selected is A1B3C3D1E4.

Average deposition efficiency range analysis
The average deposition efficiency is defined as the total mass of Nb-W alloy deposited on the Mo substrate per unit time, which is defined as:

average deposition efficiency
Nb W alloy quality chloride content of Nb and W 100% 8 -( ) =T he average deposition efficiency results of the 16 samples are listed in table 9, and table 10 shows the intuitive analysis result of the average deposition efficiency. As shown in table 10, the chlorination temperature of W (A) had the greatest influence on the average deposition efficiency, followed by the Cl 2 flow rate through W (D), the Cl 2 flow rate through Nb (C), the H2 flow rate (E), and the substrate temperature (B). Adjusting three factors, the chlorination temperature of W (A), the Cl 2 flow rate through W (D), and the Cl 2 flow rate through Nb (C), mainly caused changes in the composition ratio of Nb and W gas-phase chlorides. Therefore, in the deposition reaction, the proportions of Nb and W gas-phase chlorides in the reaction chamber were a direct factor that affected the average deposition efficiency. The average deposition efficiency decreased with the increase in the W chlorination temperature (A) and the Cl 2 flow rate through W (D), indicating that the average deposition efficiency decreased with the increase in the gaseous chloride ratio of W. The effects of other factors on the average deposition efficiency showed different degrees of wave dynamics. Under the conditions of this experiment, to maximize the average deposition efficiency, the level of each factor should be selected as A1B3C2D1E4.

Micro-structures of Nb/W alloys produced by CVD
Based on the metallographic analysis of all 16 samples, the microstructures of the CVD-prepared metal materials can be divided into two regions: an initial fine grain region and a columnar grain region. Depending on the deposition conditions, the degree of development of the fine-grained regions varied. Based on the morphological distribution of the grains, the CVD-prepared Nb-W alloy also had a similar microstructure, but the initial fine-grained region was not evident, and the columnar crystals were the main ones. All 16 samples had columnar crystal features similar to that shown in figure 4. Figure 5 shows the interface characteristics of the columnar grains. The columnar grain interface of the CVD-prepared Nb-W alloy had a serrated occlusion structure, indicating that the interfaces between the columnar grains were firmly bonded. It is worth noting that in figure 4, the layered structure is distributed perpendicular to the direction of the columnar crystals. The metallographic observation of the 16 samples showed that except for samples 3# and 4# (in which the mass percentage of W was less than 1%), the microstructures of the other samples (in which the mass percentages of W were all greater than 1%) showed similar layered structural features to that shown in figure 4. This indicated that the layered structure was mainly related to the alloy composition of the deposited matrix. When the mass percentage of W was less than 1%, the composition was macroscopically similar to that of pure Nb, and the deposited microstructure was correspondingly close to that of pure Nb, without a layered structure. When the mass percentage of W in the Nb-W alloy was greater than 1%, the presence of W influenced the microstructure of the Nb-W alloy, which exhibited a special layered structure. The layered structure was arranged layer by layer along the crystal growth direction and could still be connected at the grain boundary, but there was bending. According to the analysis of the metallographic photographs and scanning electron microscopy images, the average layer thickness of all the samples with layered structures was between 10 and 15 μm. The two samples (3# and 4#) without layered structures were not studied further, and the focus was on the group of samples with layered structures. To make the experimental analysis more accurate, it is necessary to  reduce the surface composition gradient of the sample as much as possible to prepare a more uniform and denser Nb-W alloy. Combined with the data analysis in section 3.1, the combination of factor levels to be selected was A3B2C3D2E3 to prepare the 17# sample, and the CVD Nb-W alloy's layered structure was analyzed in depth. Figure 6 shows the scan results of the electron probe surface of the 17# sample (the right side was close to the substrate), and the green and red areas represent the distributions of W and Nb in the alloy sample, respectively. Overall, with the progress of deposition, the Nb showed an upward trend, and the W showed a downward trend, In the transition area between the different color layers, indicating that the relative contents of Nb and W in each layer structure were significantly different. The surface scan results showed that the deposition of W was mainly in the initial stage of deposition, and W atoms were the main component of the CVDNb-W alloy in the initial stage of growth (the right side was close to the substrate side). As the deposition progressed, the W content decreased slowly, then transitioned to a W-dominated (Nb-W) solid solution, and finally evolved into an Nbdominated (Nb-W) solid solution.
According to the energy spectrum analysis of the 17# sample, the W content of the area highlighted in figure 7 was 9.70%, and the structure had a periodic layered arrangement of bright and dark areas. Points in the bright and dark areas were scanned (figure 8). In the bright area, the composition of Nb was 83.74% and the content of W was 16.26%. In the dark area, the composition of Nb was 93.74% and the content of W was 6.26%. Based on the compositions of the two regions, the relative content of W in the bright region was higher than that in the dark region, and the relative content of Nb was lower than that in the dark region. This further showed that the alternating light and dark layered structure of the Nb-W alloy had a direct relationship with the alloy composition. The content of W in the bright area was greater, mainly because this part of the alloy was dominated by the Nb-W solid solution, and the surrounding areas were dominated by Nb. Compared with the bright area, the relative content of Nb in the dark area was greater, and the relative content of W was reduced, because the alloy composition in this area was mainly Nb, with the Nb-W solid solution distributed around. The compositional fluctuation differences in different regions of the CVDNb-W alloy resulted in the formation of periodic layered arrangement of bright and dark regions in the microstructure. Due to this undulating feature of  the composition, there was a difference in corrosion resistance within and between layers, which resulted in a layered structure on the metallographic corrosion photographs ( figure 4). In summary, the distribution characteristics of the constituent atoms of the CVDNb-W alloy were as follows. Macroscopically, W atoms mainly formed W layers along the crystal growth direction, which then transitioned to Nb-W solid solution layers. Microscopically, the distribution of Nb and W atoms showed the characteristics of alternating contents. Since Nb and W formed a continuous solid solution, the change of the macroscopic content was mainly realized by the change of the relative contents of the two atoms in the solid solution.
During the whole deposition process, W chloride was preferentially adsorbed on the surface of the molybdenum substrate, and W chloride was also preferentially reduced by H 2 , thus forming the initial W layer. After the W layer formed, the adsorption effect of W chloride was stronger, and further reduction reactions occurred. After a period of reaction consumption, the content of W chloride in the deposition chamber decreased, and the reduction of the W chloride was weakened. At this time, the Nb chloride was enriched in the bottom layer of the boundary layer. Under the action of the concentration driving force for diffusion, the active center began to transform into the adsorption of Nb chloride. H 2 reacted with Nb chloride to generate Nb atoms, and the existence of Nb atoms provided a basis for the further adsorption of Nb chloride. This promoted the reduction reaction of Nb chloride. During the process of Nb chloride being depleted by the reduction reaction in the deposition chamber, W chloride was gradually enriched, and the adsorption of the active sites was dominated by W chloride again. As a result of the alternating reaction cycles described above, the different components in the Nb-W alloy had relatively significant fluctuations, which caused the CVDNb-W alloy to microscopically exhibit layered structure characteristics. The formation of the layered structure of the Nb-W alloy was mainly caused by the composition fluctuations caused by the combined action of the adsorption characteristics and the competitive reaction of the surface processes.
The samples were characterized by EBSD, and the (100), (110), and (111) pole figures of the CVDNb-W alloy are shown in figure 9.By looking up the included angle table [23][24][25][26] of the crystal planes of the cubic crystal system, we determined that the angle between the {100} and 〈111〉 crystal planes was 54.7°, the angles between the {110} and 〈111〉 crystal planes were 35.3°and 90°, and the angles between the {111} and 〈111〉 crystal plane were 0°, 70.5°, and 109.5°. The pole figures (figure 9) and the center of Wu's network were overlaid and rotated together, and the angle between the pole and RD (rolling direction) on each pole figure was read, which showed a high degree of agreement with the data in the table. The prepared CVDNb-W alloys had a preferred orientation of 〈111〉. figure 10 depicts the inverse pole figures of the CVDNb-W alloy sample in three directions: X (rolling direction), Y (transverse direction), and Z (direction perpendicular to the sample surface). The X (rolling direction) inverse pole figure showed that the (111) axis has the highest density, and the 〈111〉 orientation was parallel to the rolling direction. The Y (transverse) inverse pole figure showed that the {101} crystal plane was

Mechanical properties of Nb/W alloys produced by CVD
Tensile tests were carried out on the prepared samples. The mechanical properties of the Nb-W alloys prepared by deposition under different conditions (1#-16# orthogonal test samples) were analyzed. Furthermore, the 17# sample deposited with a uniform composition on the side was subjected to heat treatment, and the typical influence of the heat treatment on the structure of the material was analyzed. Due to the brittleness of most of the samples, as some of the samples cracked even after the deposition was complete, they could not be processed into tensile specimens. Thus, tensile strength tests could only be carried out on the samples that could be processed into tensile specimens. The samples that could be tested for tensile strength were 3#, 4#, 8#, 7#, 1#, and 17#. The tensile samples were are all taken from the upper surface of the substrate, and their mass percentages of W were 0.68%, 0.79%, 1.50%, 3.87%, 9.68%, and 9.70% (interlayer 6%-16%, surface scan component 9.70%). Three tensile specimens were selected for each sample, and the average tensile strength was taken as the tensile strength of the sample. Table 11 lists the average tensile strength values and the average elongation of each sample. For the general trend, with the increase in the W mass percentage, the tensile strength of the CVDNb-W alloys increased accordingly. However, the relationship between the elongation at break and alloy composition did not show a discernible trend. It is worth noting that although the compositions of the 3# and 4# samples were similar, the room-temperature tensile strength and elongation at break still showed significant differences, indicating that the combination of different deposition parameters could deposit Nb-W alloys with similar compositions but different mechanical properties. The mechanical properties of the CVDNb-W alloys were not only dependent on the composition of the alloy but also on the deposition parameters. Figure 11 shows a line graph of the change of the strength of the Nb-W alloys(3#, 4#, 8#, 7# and 1#) with the composition. The strengthening effect of W on Nb could be divided into five intervals based on the W content: Below 1%, 1%-2%, 4%-10%, and above 10%. When the atomic mass percentage of W was below 1%, the performance of the alloy was not much different from that of pure metal Nb, and the strengthening effect was the strongest between mass percentages of 4% and 10%. In addition, the room-temperature tensile strength of the 1# sample reached that of Nb521, and that of the 17# sample even exceeded that of Nb521.
The average composition of 17# was close to that of the NbW10 alloy, with a room-temperature strength of 475 MPa and an elongation of 2%. The changes in strength and elongation after heat treatment are shown in figure 12. The black triangles and circles correspond to the strength change curves after heat treatment at 1400°C and 1600°C for 2, 4, and 6 h, respectively.However, the blue color represents the elongation curve under the same conditions.After 2, 4, and 6 h of heat treatment at 1400°C, the strength increased significantly to 569, 570, and 670 MPa, respectively. Compared with the as-deposited elongation, the elongation first increased to 4.6%,  then decreased to 4.5%, and then decreased to 1.3%. After 2, 4, and 6 h of heat treatment at 1600°C, the strength dropped to 469 MPa and then dropped sharply to 123 and 33 MPa. The elongations were 3%, 0.1%, and 0.1%, respectively. This was mainly because in the process of heat treatment at 1400°C, with the prolongation of the deposition time, the content of W in the Nb-W alloy gradually tended to become uniform, which improved the solid solution strengthening effect in the alloy and increased the strength [27]. Fracture observations (figure 13) revealed that the as-deposited fracture exhibited brittle layered dissociation fracture ( figure 13(a)). After heat treatment at 1400°C, it still exhibited brittle layered dissociation fracture. After 4 h heat treatment at 1400°C, tear edges were more evident and the connections were dense. The strength and elongation increased to a certain extent ( figure 13(b)). The layered structure was still maintained, the mass percentage of W in the bright area was 16.88 wt%, and the mass percentage of W in the dark area was 12.80 wt%. Compared to the samples without heat treatment, the W content in the interlayer increased, and the overall W content tended to be uniform ( figure 14). After 6 h of heat treatment at 1400°C, the thickness of the lamella increased, the tearing edges of the fracture were denser, the strength continued to increase, and the elongation decreased, but the whole process was still brittle dissociation fracture ( figure 13(c)). After heat treatment at 1600°C, the grains of the sample grew and became coarser, and the number of grain boundaries decreased. At the same time, the layered structure of the CVDNb-W alloy basically disappeared, the composition was homogenized, and there was significant recrystallization growth ( figure 15). The sample showed a distinct rock-sugar-shaped fracture, with brittle fracture characteristics and a significant decrease in strength ( figure 13(d)).
In-situ observation is a new testing method from the end of the 20th century. Compared with traditional mechanical test microstructure observations, in situ tests can be used to observe the microstructures of materials during loading in real time, which not only simplifies research but also facilitates comparative analysis of the structure before and after deformation. It can also accurately and clearly reflect the typical characteristics of each stage of deformation, avoiding the wrong inference of material deformation and fracture caused by operation errors [28][29][30][31][32]. The mechanical properties of the Nb-W alloy prepared by the CVD method were significantly Figure 11. Relationship between tensile strength and composition of Nb-W alloys prepared by CVD. higher than those of the as-cast alloy with the same alloy composition. Observing the changes of the crosssectional structure during the tensile process through in situ tensile experiments is helpful for analyzing the reasons for the improvement of the material's strength at room-temperature and obtaining the hightemperature strength.
The in situ tensile test was carried out using a thermomechanical in situ observation system consisting of S-3400 scanning electron microscope (SEM), INCA energy dispersive spectrometer (EDS) and GATAN in situ dynamic tensile test bench with a maximum load of 5kN. The dimensions of in situ tensile test bench and tensile specimen are shown in figure 2. Before in situ stretching, metallographic etching solution (H 2 SO 4 :HNO 3 :HF = 5; 2; 2) The working section of the sample is corroded for 30s to highlight the grain boundary and the second phase under SEM. SEM secondary electronic imaging mode was used for in situ observation and EDS was used for composition analysis of the samples. The scanning voltage was 15 kV and the amplification was 20-1000 times. Pay close attention to the changes of the sample surface during the drawing process. When there is a slight change, stop the drawing immediately and maintain the stress, and then observe and take photos. The tensile rate was 0.1mm min −1 and the temperature was room temperature. The in situ tensile test was repeated for three times to ensure the repeatability of the results figure 16 shows the in situ tensile and fracture morphologies of the CVDNb-W alloy. The 17# sample exhibited a distinct necking phenomenon during the stretching process. The fracture surface morphology was mainly composed of cleavage steps. Cracks were generated at the grain boundaries during fracture and then expanded along the cleavage plane and formed a river-like pattern. The river-like pattern developed to another grain boundary and accumulated continuously, resulting in macroscopic fractures. In summary, when Nb-W alloys are stretched, cracks often occur at the grain boundaries, but there are often multiple paths in the process of crack propagation, that is, intergranular fracture, cleavage fracture, and microporous polymerization can occur on the microscopic level. Different proportions of the microscopic fracture modes result in macroscopically different fracture characteristics.

Conclusions
(1) Based on an orthogonal experimental design, by controlling the chlorination temperature of W, the temperature of the substrate, the Cl 2 flow rate through Nb, the Cl 2 flow through W, and the flow of H 2 Nb-W alloys with W contents of 0.33%-50.48% were prepared.
(2) The Nb-W alloy prepared by the CVD method had a non-uniform macroscopic distribution of components, the substrate temperature had the greatest influence on the lateral composition gradient, and the H 2 flow rate had the least influence.
(3) The layered structure of the CVDNb-W alloy was mainly related to the composition of the deposited body. When the mass percentage of W was less than 1%, the deposited microstructure was close to that of pure Nb and did not show a layered structure. When the mass percentage of W was greater than 1%, the presence of W affected the microstructure of the Nb-W alloy, thus showing a special layered structure.  (4) With the increase in the mass percentage of W, the tensile strength of the CVDNb-W alloy increased accordingly.The maximum strength of the alloy with a 9.68% W content reached 475 MPa, which exceeded the room-temperature tensile strength of Nb521.
(5) In the process of heat treatment at 1400°C, with the extension of time, the W content in the alloy gradually tended to become uniform, which improved the solid solution strengthening effect in the alloy, and the strength was significantly improved. After heat treatment at 1600°C, the grains grew and became coarser, the grain boundaries were reduced, the composition was homogenized, and there was significant recrystallization growth.