A novel and predictive process of manufacturing 3DN C/SiC torque tube

In this research, a novel method of CVI + RMI 3DN C/SiC torque tube preparation has been researched. CVI + RMI 3DN C/SiC flat panel was evaluated by Archimedes drainage method for density and open porosity, SEM for morphological characterization, XRD for phase composition characterization, and chemical method for composition mass and volume content. It has an average density of 2.19 g cm−3, and open porosity of 10%. The computed chemical composition method findings are undoubtedly in line with the fiber design volume percentage of 30%. Tensile and shear mechanical tests on 3DN C/SiC standard sample were investigated, with good performance. The average tensile and shear strength was 141.64 MPa and 86.24 MPa respectively. 3DN C/SiC torque tubes were prepared by the same process, and the torsional mechanical tests were carried out. According to the mechanical parameters of the flat panel 3DN C/SiC sample, the mechanical properties of stress distribution and shear stress-strain curves for the torque tube are predicted by using the finite element simulation method, which is in good agreement with the test results. The test and simulation error of the maximum shear strength is only about 2%.


Introduction
With the rapid aerospace propulsion development in recent years, carbon fiber reinforced silicon carbide (C/ SiC) composite becomes one of the most crucial materials in ceramic matrix composite (CMC) family, for the superior specific strength and modulus, good corrosion resistance and ultrahigh temperature performance [1][2][3][4][5][6]. More researchers are focusing into CMC structural components like CMC tubes since CMC is used in engineering applications [7][8][9]. CMC thick tubes are primary candidate as rudder shaft in reusable launch vehicles, and in fusion and fission reactors, they are excellent structures as control rod guiding tube and core barrel. Nevertheless, CMC materials have large dispersion of mechanical property performance and hard to predict which is more serious for thick structures. How to manufacture thick wall C/SiC structure and predict mechanical properties still remains an issue during engineering applications [10][11][12][13]. A significant reason is that there are so many combinations of fiber preform structures and processes, so it is challenging to identify an appropriate matching preform structure and process. Hence, we're attempting to find a suitable preform and process combination mode. Generally the C/SiC fiber preform has 2-dimension woven (2D), 2.5-dimension, 3-dimension needle (3DN), 3 dimensional woven, 3 dimensional stitching, and so on. At present, there are many researches on 3DN preform [14][15][16][17][18]. 3DN C/SiC structural components have more infiltration channels in chemical vapor infiltration (CVI) and reactive melt infiltration (RMI) process. The commonly used traditional CVI process and RMI process both have disadvantages, for CVI it is costly, corrosive, poisonous, explosive, and has low deposition rates and bottle neck effects, and for RMI it is chemical reactions and fiber harming at infiltration temperature. As a result, we believe the combination of CVI and RMI (CVI + RMI) procedure is Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. appropriate. Taking advantage of the advantages of the two processes, some shortcomings of CVI and RMI are avoided, the preparation cycle is shortened, the cost is lower than CVI, and CVI SiC matrix can protect the C fiber from RMI Liquid Si in high temperature.
In this research we employed CVI + RMI process to solve the dispersion of performance and predictability of CMC torque tubes. Since CVI + RMI 3DN C/SiC has not before been the subject of any comprehensive research. The density, open porosity, micromorphology, phase composition, and composition percentage of the CVI + RMI process flat panel 3DN C/SiC are the topics we will be focusing on in this research. The torsional mechanical properties were anticipated by simulation after the mechanical characteristics of the flat panel 3DN C/SiC and the matching torque tube were examined.

Materials preparation
The 3DN preform (figure 1) was prepared by Jiangsu Tianniao Company. It involves cutting C fiber into shortcut fiber cloths and nonwoven fiber cloth, making successive layers of 0°nonwoven fiber cloth and short-cut fiber cloths and 90°nonwoven fiber cloth and short-cut fiber cloths, and finally hooking the C fiber in the shortcut fiber cloths into the adjacent layer using relay needle punching technology in the direction perpendicular to the ply. 6 K T300 carbon fibers and 12 K T700 carbon fibers (Toray Corporation) were employed to produce nonwoven fiber cloth and short-cut fiber cloths.
PyC interphase is formed on the 3DN preform fiber surface using propylene (C 3 H 6 ) as the raw material, hydrogen as the carrier gas, and argon as the dilution gas. The deposition temperature is 880°C-1000°C, and the total gas pressure is 3-5 kPa. When the PyC interphase reaches 200 nm, the 3DN preform is heat treated for 1 to 2 h under vacuum at 1800°C. The deposition of SiC matrix was by the high-temperature CVI furnace filled with CH 3 SiCl 3 (MTS) using the bubbling technique with argon serving as the dilution gas and hydrogen serving as the carrier gas. The hydrogen to MTS flow ratio is 10:1. The total pressure of the gas in the furnace is 0.01 MPa, and the chemical reaction of CH 3 SiCl 3 to produce SiC occurs at 1000°C and 5kPa. SiC matrix density may be controlled by CVI time, followed by the RMI procedure.
We used silicon particles with an average particle size of 45 μm from Jinan Yinfeng Silicon Products Corporation. In a graphite crucible, the specimens were packed with silicon particles, and a self-assembly vacuum furnace was used. The densification process was completed by the silicon particles melting and spontaneously infiltrating into the preform after the temperature was raised quickly to about 1600°C at a heating rate of 40°C min −1 . The surface residual silicon was removed from the specimens after they had cooled to room temperature, and the samples with the required dimensions were then cut from them, including 3DN C/SiC panel specimen and torque tube.

Characterization
The Archimedes drainage method is used to determine the material's density and open porosity. Mettler Company AG204 electronic balance and vacuum dish, with an accuracy of 0.01 mg, is the apparatus utilized. Scanning electron microscopy was utilized to describe the microstructures of the samples (S-4700, Hitachi, Japan). The phase compositions were analyzed by Rigaku D/max-2400 XRD from Japan with test parameters: Cu-Kα x-ray with wavelength λ 0.15406 nm, 40 kV target voltage, 40 mA target current, and step size of 0.02°. In order to extract Si from 3DN C/SiC composite materials, the sample was first corroded with an 80% fuming nitric acid +20% hydrofluoric acid mixed solution with volume content for 48 h. Next, C was removed by oxidizing the sample in air at 700°C for 10 h. Then it is possible to quantify each component's mass fraction content.
Refer to ASTM C1275 and ASTM C1292 for tensile test and shear test, each test had 5 samples, the shape and size of the samples were shown in figures 2(a) and (b), in which all dimensions are in mm, the loading rate is 0.5 mm min −1 . The thickness is 6 mm, which is different from 3 mm in CMC ASTM standard, due to the special structure of 3DN preform. In 3DN preform, each layer is about 0.5 mm, 4 layers is one unit, so 3 mm specimen may only contain 1.5 units. The shape and size of 3DN C/SiC torque tube was shown in figure 2(c), the loading rate is 2 mm min −1 . All the mechanical tests were conducted utilizing the Instron 1196 universal material test equipment (Instron Corporation, USA). To analyze the stress distribution and non-linear behavior of the 3DN C/SiC torque tube, a finite element method (FEM) was employed.

Structure
Density and open porosity results of CVI + RMI 3DN C/SiC 6 mm panel specimen were listed in table 1. The microstructure is very important for the mechanical properties and stability of CMC torque tube. For thick-wall torque tube, the bottleneck effect of CVI process leads to more serious low density and uneven distribution of microstructure, CVI + RMI process greatly reduces this disadvantage. The average density is 2.19 g cm −3 , and the average open porosity is 10%, which is compact with regard to this thickness. Since 3DN C fiber preform has efficient and appropriate CVI channels, and the RMI liquid approach also produces densification. As demonstrated in figure 3, by SEM examination, figures 3(a) and (b) show that the surface of the panel CVI 3DN  C/SiC specimen is uneven, the short-cut fiber web cloth has a large number of holes due to the inevitable permeability of CVI. While from the results from figures 3(c) and (d), Si filled the short-cut fiber web cloth during RMI process, increased density, and confirmed that the bottleneck effect of CVI was reduced.
As the XRD results exhibited in figure 4, CVI + RMI 3DN C/SiC panel specimen was composed of C, SiC and Si. Chemical method can quantitatively calculate the volume fraction and mass fraction of various constituents to help us understand the properties of torque tube. The matrix SiC, Si and C fibers had densities of 3.20, 2.30 and 1.76 g cm −3 , respectively. After using a chemical procedure to estimate the mass fractions, in accordance with the mixture rule, equation (1) can be used to determine the volume fractions of the C fiber, SiC and Si matrix: SiC SiC Si Si =´+´+Í n this equation d 3DN , d c , d SiC , and d s were densities for 3DN C/SiC panel specimen, C fiber, SiC matrix and Si matrix, respectively. V C , V SiC and V Si were volume fractions of C fiber, SiC matrix and Si matrix, respectively. The specific percentage content was shown in table 2. The mass fraction of C fiber, SiC matrix and Si matrix were 26%, 41%, 33%, and the volume fraction were 32%, 28%, 32%, respectively. It is clearly that the calculated results consistent with the fiber design volume fraction of 30%.

Mechanical behaviors
The results for 3DN C/SiC panel specimen tensile and shear mechanical tests were shown in figure 5. The tensile strength formula is as follows: here σ t was tensile strength (MPa); P b was tensile ultimate load (N); b was width of gauge length section of tensile sample (mm); t was tensile sample thickness (mm).
The shear strength formula is as follows: here τ c was shear strength (MPa); P c was shear ultimate load (N); b was width of gauge length section of shear sample (mm); h was shear sample thickness (mm).
Tensile strength of the 5 specimens is 141.64 ± 31.73 MPa with tensile strain 4200.56 ± 1056.00 με. Shear strength of the 5 specimens is 86.24 ± 3.51 MPa with shear strain 16818.81 ± 728.83 με. Shear strength and strain has a substantially smaller standard deviation than tensile, which implies CVI + RMI 3DN C/SiC has stable shear properties and suitable for torque applications.
The average linear tensile and shear modulus were 92.37 GPa and 15.86 GPa respectively, also derived from the 3DN C/SiC panel specimen mechanical test. All the parameters will be feed into the finite element method numerical model to predict the torsional behavior of CVI + RMI 3DN C/SiC torque tube. According to the mechanical test results, the mechanical strength and strain stability of the torque tube are excellent, especially the shear properties. Therefore, the average strength can be directly used for finite element calculation, which is very convenient and can ensure the validity of calculation.

Comparison between the test and simulation results
3DN C/SiC torque tube test is complicated and expensive, so the development of numerical method to validate structural properties is critical. A finite element method with ABAQUS software by Dassault Corporation was utilized to carry out the numerical simulation.
In the FEM model, one end of the 3DN C/SiC torque tube was constrained and the torque was applied through a reference point to another end. The total mesh number was 66000, and C3D8 element type was used. ABAQUS UMAT was introduced to simulate the nonlinear torsion behavior of 3DN C/SiC material. The stiffness reduction is considered to calculate the stiffness matrix E d , as shown in formula (4).   Derived from 3DN C/SiC panel tensile and shear tests, the modulus and Poisson ratio parameters were listed in table 3. Cylindrical system was utilized in FEM model and 1, 2, 3 represent three directions, 1 for radial direction, 2 for transverse direction and 3 for axial direction.
At the stress of panel specimen failure, we got the shear stress S23 result contour of 3DN C/SiC torque tube, as shown in figure 6. Small to large shear stresses are distributed radially from the interior to the exterior. The minimum stress is 71.87 MPa and the maximum stress is 86.24 MPa. 3DN C/SiC torque tube failed when exterior in-plane shear reaches the panel specimen shear strength 86.24 MPa.
Stress-strain curve of test and simulation results for 3DN C/SiC torque tube were exhibited in figure 7, and it is clearly seen that they coincide well with each other. The slight difference between stress-strain non-linear behaviors may belong to various reasons. For instance, structural variations such as thickness and curvature, mechanical test device and fixture differences, or inevitable error of CVI + RMI manufacturing process. In general, it is a stable and predictable manufacturing process, employ which we can obtain torque tube properties conveniently in engineering applications. Large dispersion, complicated testing, and expensive testing costs have all been issues that have limited the adoption of CMC engineering structures. Realizing the prediction of the thick-wall torque tube from a mechanical test of a standard small sample will be of tremendous technical value. To accomplish this purpose, we employ the finite element approach in this work.

Conclusion
In this work, CVI + RMI flat panel and torque tube specimens were generated by combining the benefits of the CVI and RMI methods. The novel CVI + RMI process achieves the purpose of preparing thick-wall torque tube with small dispersion and predictable mechanical properties. 3DN C/SiC flat panel samples have a high density, a low porosity by Archimedes drainage method, and a dense micro morphology by SEM. Chemical method for composition contents revealed that fiber volume fraction measurements are consistent with the design volume fraction, demonstrating the process' high degree of controllability. The standard 3DN C/SiC panel specimens had good mechanical characteristics with average tensile and shear strengths of 141.64 MPa and 86.24 MPa, respectively. The FEM numerical simulation demonstrates that this method has a unique and significant benefit in that it can forecast the 3DN torque tube's torsional stress-strain curve using the mechanical test parameters from the 3DN C/SiC standard panel samples. The practical benefit of this study is that it enables the preparation of nuclear reactor control rod guiding tubes and rudder shafts with cheaper costs, less dispersion, and more stable mechanical properties.