Experimental Study on the Influence Mechanism of Carbon Fiber/Epoxy
Composite Reinforcement and Matrix on Its Fire Performance

The effects of the number of layers, the arrangement of carbon fiber (CF) tow and the epoxy resin (ER) matrix on the fire performance of carbon fiber/epoxy composites (CFEC) were studied by a variety of experimental methods. The results show that the number of layers of CF tow has influence on the combustion characteristics and fire propagation of the composites. The arrangement of CF tow has influence on flame propagation rate and high temperature mechanicalproperties. The mechanism of the influence of the number of layers of CF tow on the composite is mainly due to the different thermal capacity of ER matrix. The effect of the arrangement of CF tow on the fire performance of the composite is mainly due to the inhibition and obstruction of the tow on the combustion of ER matrix. The influence on the high temperature mechanicalproperties is mainly due to the different arrangement direction of CF tow. The fitting equation of the mechanicalproperties of the samples was obtained. This equation could be used to predict the samples’ tensile strength from 25°C to 150°C by comparing with the experimental results. Taking the carbon fiber woven cloth (C) applied in the fuselage material as an example, combining the influencing factors of various parameters in the fire field, some suggestions are put forward combined with the research conclusion.


Introduction
CFEC is an ideal load-bearing component. It is lightweight and strong [1][2][3]. It has the advantages of excellent thermostability and service durability [4,5]. It is widely used in the fields of civil engineering, automobiles, and sporting goods [6][7][8]. It has been widely used as a structural material for aviation aircraft in the aeronautical field [9][10][11][12][13][14]. The properties of CFEC depend on the crosslinking properties of reinforcement and matrix to a great extent [15,16]. The crosslinking properties of CF and ER have a great influence on the overall properties of CFEC [17][18][19][20]. However, the ER matrix is flammable [21]. In case of fire, it will not only cause fire accidents, but also destroy the bonding between carbon fiber and epoxy resin. The flammability of epoxy matrix has great influence on the strength, rigidity and stability of CFEC in fire [22]. The number of layers and arrangement of CF tow are important contents in the development of structure and property design of composite [23].
The number of layers and arrangement of CF tow determine the properties of ER matrix. It is of great significance to study the layers and arrangement of CF tow. This will be helpful to study the mechanism of fire resistance of CFEC.
Scientific research personnel in various countries have carried out research on fiber-based composites around their own properties [24,25]. The research on epoxy resin mainly focuses on its flame retardant treatment. It is applied to the preparation of flame retardant composite. Gu et al. prepared a kind of high efficient flame-retardant phenyl phosphonate epoxy resin (FREP) nanocomposites, which has low flammability and excellent conductivity [26]. A new type of high performance resin was prepared for MFP/ed composite [27]. Li et al. provides a simple and effective method for manufacturing polymer matrix composites with excellent heat conduction but insulation [28]. Song et al. immerse the obtained rGH framework into epoxy resin to prepare the corresponding rGH/epoxy composite [29]. Zhao et al. introduced the synthesis of silicon-containing polyborozines (spbz) ceramic precursor into the preparation of modified phenolic resin (spbz-pr), which was used as the resin matrix to obtain carbon fiber reinforced spbz-pr (CF/spbz-pr) laminated composite by hot compression [30].
The research on CFEC by scientific research personnel in various countries focuses on the performance change of composite and the improvement of its mechanicalproperties base on the above review. The research of epoxy resin mainly focuses on its flame-retardant treatment. However, there are few studies on the influence mechanism of the number of layers and different arrangement of CF tow and the ER matrix on the thermal properties of the composite in the research work of CFEC.
In this paper, the influence mechanism of the comprehensive properties of CFEC in the fire was studied by various experimental methods. All kinds of working conditions are set in the experiment. The layers and arrangement of CF tow were changed, and the experimental results were compared in the experiment. The thermal properties of carbon fiber/epoxy woven fabric, carbon fiber/epoxy prepreg and CFEC with different laying design were evaluated in fire.

Preparation of CFEC
The composite used in the experiment is applied to the fuselage of a small aircraft. The reinforcement of the CFEC is carbon fiber. Its model is T300_3K. The main component is polyacrylonitrile. The matrix material of CFEC is epoxy resin [31]. The manufacturing process of CFEC is wet process. The wet forming process: 1. Preparation of ER matrix. The ER matrix is composed of the liquid epoxy AralditeLY1564SP and curing agent XB3487. 2. Brush the mixture of epoxy resin and curing agent on T300_3K CF tow with roller as required. Make sure that the mixture of epoxy resin and curing agent is evenly distributed on T300_3K CF tow. The T300_3K CF tow is laid layer by layer according to the process requirements. At this stage, the number of layers and arrangement of CF tow can be changed. Finally, the residual epoxy resin liquid was dried by the absorbent felt to ensure the uniformity and integrity of the CFEC.
3. Put it into a vacuum bag for vacuumizing. The CFEC was prepared after waiting for 24 h.
The CFEC with different arrangement of CF tow and layers can be prepared by changing the process.

Experimental Sample
Tab. 1 shows the layers and arrangement of sample CF tow used in each test.
The material of the sample used in the experiment is carbon fiber woven cloth. A1-A4 is a CFEC with the same arrangement of CF tow and different layers of CF tow. B1-B4 is an experimental sample with the same number of layers and different arrangement of CF tow. Fig. 1 shows the different arrangement of CF tow.

Experimental Method 2.3.1 Cone Calorimeter Method
The test shall be carried out according to the standard of ISO5660-1: the fire reaction test heat release rate, smoke production and mass loss rate, Part 1 [32]. The combustion characteristics of the samples were measured by cone calorimeter under the conditions of 25 kw·m −2 , 30 kw·m −2 , 35 kw·m −2 , 40 kw·m −2 and 50 kw·m −2 . The model of cone calorimeter is FTT-CONE-228. The size of the experiment sample is made to be 100 mm × 100 mm according to the standard. Each experiment was repeated three times. Take the average value of each experiment for the test results.

Experiment of Fire Spreading Characteristics
(1) High temperature oxygen index (OI) experiment  The test is carried out according to the standard of ISO 4589-3: Determination of oxygen index of plastics combustion performance Part 3: high temperature test [33]. The OI of the samples at different temperatures was measured by FTT-OL-1402072 OI tester. The test sample used in the experiment is made into a size of 150 mm × 10 mm according to the standard. Each experiment was repeated three times. The final result is determined after the results of three experiments are consistent.
(2) Vertical/horizontal combustion experiment The vertical/horizontal burning rate values of all test samples were tested. The test instrument used in the experiment is CZF-5 V/H combustion instrument. The experimental samples were prepared into rectangular samples according to IEC60695-11-10: 2003. The size is 150 mm × 10 mm. Three experiments were carried out under each working condition in the process of carrying out the experiment in order to ensure the repeatability of the experiment [34]. The experimental results are averaged to ensure the accuracy of the experiment.

Mechanical Property Experiment
The mechanicalproperties of four kinds of experimental samples in high temperature environment were tested by the self-developed mechanical instrument. The codes of the four samples are B1, B3, C and D. Tab. 2 shows the details of the experimental samples. Six temperature points (25°C, 50°C, 80°C, 100°C, 120°C and 150°C) were set in the experiment. In order to prevent large errors in the experiment, three experiments were carried out at each selected temperature point (25°C, 50°C, 80°C, 100°C, 120°C and 150°C) during the high temperature mechanical property test. Take the average value of the experimental data as the final result of the experiment.

Fire Spread Characteristics
The high temperature OI and vertical/horizontal burning rate of materials are important indexes to measure the spreading characteristics of combustible materials in the fire. The flame propagation speed of different combustible materials is different. The flame propagation rate of combustible materials is an important index to evaluate the fire risk of combustible materials and measure the development and propagetion of flame.

The Flame Propagation Rate of Composite
It is found that the flame burning rate in horizontal direction of all experimental samples (A1-A4, B1-B4) is significantly lower than that in the rear direction by testing the flame velocity of each experimental sample. This rule can be seen clearly in Fig. 2. The cause of this phenomenon is determined by the characteristics of the flame itself. When heated, the air changes into an updraft, making the flame face upwards. Ignite from the lower end of the material when burning vertically. Ignite from the lower end of the material when burning vertically. The unburned material is partially preheated, and the flame propagation speed is increased. Fig. 3 shows the flame propagation in different directions. Fig. 2a shows that the vertical/horizontal combustion rate increases slightly with the decrease of the number of layers. Four kinds of epoxy matrix materials (10 g, 16 g, 20 g, 38 g) with different quality were prepared in order to explain this phenomenon. The ignition time was measured by cone calorimeter (heat radiation flux 25 kW·m −2 ). Fig. 4 shows the curves of heat release rate and time of four epoxy matrix materials with different mass, and gives the ignition time and the appearance of the experimental samples. Fig. 4 shows that the larger the mass of the epoxy matrix, the longer the ignition time. The time to reach the peak of heat release rate is delayed. The more layers and thicknesses of the CFEC, the larger the thermal capacity of the epoxy matrix in the tested sample. The thermal capacity can prolong the time to reach the pyrolysis temperature of combustibles and reduce the vertical/horizontal combustion rate [35]. Fig. 2b shows the vertical/horizontal burning rate with different arrangement of CF tow. The highest rate is B3, and the lowest rate is B4. This is because the arrangement of B3 CF tow is (0°/90°) with the most layers, while B4 is the least. (0°/90°) CF tow are arranged in the same direction as flame vertical/ horizontal propagation. It is more conducive to the spread of flame. The propagation direction of vertical/ horizontal flame is 45°with that of CF tow when the arrangement of carbon fiber bundles is (± 45°). There are obvious obstacles to combustion.  The residual samples were observed by electron microscope. Select B4 to enlarge the image 500 times after the experiment. The arrangement of CF tow can be clearly seen (because of the upper CF tow block the lower vertical CF tow, only one direction of CF tow can be observed). (± 45°) CF tow can block the combustion spread in combination with Fig. 5.

Oxygen Index (OI)
OI is an important index to evaluate the flame retardancy of composites. The OI at room temperature can not really describe the state of composite in high temperature environment. The high temperature OI of the experimental sample was measured at 50°C, 90°C, 120°C, 150°C, 180°C and 220°C.
Figs. 6 and 7 show that the OI at high temperature is lower than that at low temperature. This is because of the process of temperature rising in the experiment, which makes the material of the experiment sample preheat itself. As a result, its ignition point is reduced and it is easy to be ignited. Fig. 6 shows that the OI (ordinary temperature OI and high temperature OI) of the experimental sample increases with the increase of the number of CF tow's layers. The results show that different layers of CF tow will lead to different OI. The number of layers and OI of CF tow show certain regularity. The microstructure of A1 and A4 were observed after the high temperature OI experiment by SEM to explain this phenomenon.
The residue of the sample after the high temperature OI test at 220°C was selected. The two samples with the thickness of 1.84 mm (A1) and 0.60 mm (A2) after the high temperature OI test were observed by SEM [36]. The layers of A1 and A4 are 8 and 2, respectively. Fig. 6 shows the microscopic image of two test samples (A1, A4) with different layers of CF tow after combustion. On the one hand, the larger the  number of layers, the thicker the experimental sample (A1) will have more epoxy resin. In the process of high temperature decomposition of epoxy resin, porous carbon layer will be formed, which will hinder combustion [37]. The low thermal conductivity of carbon layer hinders combustion. The more layers and thicknesses of CF tow, the more epoxy resin they have. Its polymer composite material has stronger ability of carbon formation. The higher the OI.
The arrangement of CF tow has no effect on OI. This can be seen clearly in Fig. 7.  samples with the condition of constant thermal radiation intensity. This is because A1 has the largest number of layers and thickness, and its epoxy resin has the largest heat capacity. A4 has the smallest number of layers and thickness, and its epoxy resin has the smallest heat capacity. The increase of thermal capacity will prolong the time when the epoxy matrix reaches pyrolysis temperature. The number and thickness of carbon fiber have a great influence on the ignition time of CFEC. Fig. 8b shows that the different arrangement of CF tow in each sample has no significant effect on ignition time with the same thermal radiation intensity.

Heat Release Rate (HRR)
Tab. 3 contains the peak heat release rate (P-HRR) and the time from the beginning of the experiment to the peak heat release rate (P-HRR) of each experimental sample under different thermal radiation intensity.
Tab. 3 shows that the P-HRR of each experimental sample (A1-A4, B1-B4) increases with the increase of thermal radiation intensity from 25 kW·m −2 to 50 kW·m −2 . The time from the beginning of the experiment  to the P-HRR of each sample decreased. This is because with the increase of the thermal radiation intensity from 25 kW·m −2 to 50 kW·m −2 , the thermal radiation energy of the ER matrix increases gradually, which leads to the easier and faster thermal decomposition of the matrix.
The P-HHR and the time to reach the P-HHR increase with the increase of the number of layers of CF tow. Take the experiment under the condition of thermal radiation intensity of 40 kW·m −2 as an example. The experiment sample A1 has the most layers. Its P-HHR is the largest, reaching 1138.26 MJ·m −2 , and the time to reach the P-HHR is the longest, reaching 131 s. A4 has the least number of layers. Its P-HHR is the smallest, which is 602.17 MJ·m −2 , and the time to reach the P-HHR is the shortest, which is 72 s. The arrangement of CF tow has little effect on the P-HHR and the time to reach the P-HHR. The P-HHR and the time to reach the P-HHR of B1-B4 are basically the same.

Mass Loss Rate (MLR)
Mass loss is a common parameter to characterize the properties of materials in fire. The amount of decomposed materials can be quantitatively characterized by measuring the mass loss of materials in fire. The decomposition rate of the material can be characterized by the test of mass loss rate at the same time. The mass loss and MLR of carbon/fiber epoxy resin composite can be measured by cone calorimeter. The change of the mass loss of the experimental sample was observed by the balance load-bearing device on the cone calorimeter [38][39][40]. Fig. 9 shows the relationship between the average mass loss rate Av-MLR of each experimental sample and the different thermal radiation intensity. It can be clearly seen that the number of CF tow has influence on the Av-MLR of CFEC. The Av-MLR increases with the number of layers. This is mainly due to the larger thickness of the experimental samples with higher quality of epoxy resin. Its greater thermal capacity. The arrangement of CF tow (B1-B4) has no effect on the MLR of CFEC. Fig. 10 shows the image of B1-B4 after the cone calorimeter experiment (40 kW·m −2 ). The ER matrix of four samples (B1-B4) has been burnt out. In addition, the CF tow are scattered and obvious delamination phenomenon appears. The layers and thickness of the four samples (B1-B4) are basically the same. The mass loss is basically the same. At this temperature, the oxidation of carbon fiber has not yet occurred. As a reinforced material, the mass loss of carbon fiber is small. The arrangement of CF tow has no influence on the MLR.
The thermal radiation intensity of the material has a linear relationship with the Peak-MLR (P-MLR) and Av-MLR. This relationship can be expressed by Eq. (1) [41].
m' represents the MLR, including P-MLR and Av-MLR. L represents the vaporizing heat. q net represents thethermal radiation intensity. C represents a constant.
The vaporizing heat L is the reciprocal of the slope of Eq. (1). Fig. 11 shows the curves of P-MLR and Av-MLR of CFEC with different layers of CF tow changing with the thermal radiation intensity. The P-MLR and Av-MLR were fitted to function. The fitting functions of P-MLR and Av-MLR of four experimental

Tensile Failure Strength
Tensile tests were carried out at 25°C, 50°C, 80°C, 100°C, 120°C and 150°C. The details of the test samples (B1, B3, C and D) used for the mechanicalproperties tests are listed in Tab. 2. The tensile failure strength of the samples was tested by experiments. The tensile failure strength is the stress of the experimental sample when it is broken in the experiment. Its physical significance lies in that it can characterize the mechanicalproperties of materials. At the same time, the strength of the material can be quantified [42]. The tensile modulus can be used to describe the difficulty of elastic deformation of materials under the action of force. Its physical meaning indicates the magnitude of the external force required by the unit elastic deformation of the material [43]. The change rule of tensile failure strength and tensile modulus of samples by testing the mechanicalproperties with temperature is basically the same. Fig. 12 shows the bar chart of tensile failure strength and the trend of change of samples at six temperature points.
The influence of high temperature on the mechanicalproperties of experimental samples can be generally characterized by the retention rate of high temperature mechanicalproperties. The higher the retention rate of the calculated experimental sample, the smaller the influence of the temperature on the mechanicalproperties of the experimental sample. The more stable the mechanicalproperties are at high temperature. The calculation formula is shown in Eq. (10): q 0 represents the mechanical property index parameter (tensile failure strength and tensile modulus) at room temperature (25°C); ρ represents the mechanical property index parameter (tensile strength and tensile modulus) at high temperature; ρ f represents the retention rate of mechanicalproperties (tensile failure strength and tensile modulus) at a specific temperature.
The mechanicalproperties of the experimental samples at different temperatures are calculated. Tab. 4 shows the retention rates of tensile strength and tensile modulus of four samples at 50°C, 100°C and 150°C.
Tab. 4 shows that the retention rate of mechanicalproperties of four samples are less affected by temperature, and the mechanicalproperties remain stable from 25°C to 100°C. The retention rate of mechanicalproperties of carbon fiber/epoxy woven cloth (B1, B2, C) decreased rapidly from 100°C to 150°C. The retention rate of tensile modulus of B1 decreases the fastest, which indicates that the retention rate of tensile modulus of B1 is the most affected by temperature.
It can be seen that the mechanicalproperties of samples are not affected by the temperature in the range of 25°C to 100°C from the above analysis. The internal force properties remain basically the same in this temperature range. The mechanicalproperties of samples is no obvious change when the ambient temperature is 25°C and 50°C. This is because the ER matrix in CFEC has not been pyrolyzed when the temperature of epoxy matrix rises from 25°C to 50°C. The crosslinking property of CF tow and ER matrix is not affected at this time. The increase of temperature has no obvious effect on the mechanicalproperties of the samples at this stage. When the ambient temperature rises to 100°C, the mechanicalproperties of the four samples show a slight downward trend compared with 25°C. The epoxy matrix began to decompose due to the increase of temperature at this temperature. The mechanicalproperties (tensile failure strength and tensile modulus) of the four samples decreased significantly when the ambient temperature increased from 100°C to 150°C. The mechanicalproperties of samples decreased significantly compared with 25°C when the temperature increased to 150°C. The experimental results are consistent with the conclusions in the literature [44].
The thermogravimetric analysis of ER matrix was carried out under the condition of heating rate of 20°C min −1 to explain this phenomenon. The pyrolysis temperature of ER matrix was measured. The heating rate of TGA experiment is the same as that of high temperature mechanical property experiment.   . 13 shows the TG and DTG curves of the epoxy matrix at heating rate of 20°C ·min −1 . The initial pyrolysis temperature of epoxy matrix is 249°C under these conditions.
The CFEC was tested by thermogravimetric experiment. In the thermogravimetric experiment, the pyrolytic residue of CFEC is taken out when the temperature rises to 100°C and 150°C. Fig. 14 shows the microstructure of the CFEC remnant magnified 1500 times by Scanning electron microscopy at 100°C and 150°C respectively. As the ER matrix belongs to non conducting organic compounds, the accumulated charge is bright under the thermal field scanning electron microscope [45,46]. The parts are shown in Fig. 14. The preparation process of CFEC determines that the reinforced CF tow and the ER matrix are bonded together. It can be clearly seen from Fig. 14 that at 100°C, most of carbon fiber and epoxy resin are still bonded together. At 150°C, although the initial thermal decomposition temperature of epoxy resin (249°C) was not reached, some epoxy resin had separated from carbon fiber and existed independently. The blisters will appear on the surface of ER matrix with the increase of temperature. There are also cracks on the surface of ER matrix [47]. At last, the mechanicalproperties of ER matrix decreased obviously [48][49][50]. The decrease of mechanicalproperties between CF tow and ER matrix is due to the decrease of cross-linking properties between CF tow and ER matrix [44,51]. This is the reason why the strength of the composite decreases signficantly at 150°C. The tensile failure strength of the samples decreased with the increase of the experimental temperature. This law can be seen clearly in Fig. 12. B3 has the best mechanicalproperties, while D has the worst mechanicalproperties, and B1 and C have the same mechanicalproperties at 25°C, 50°C, 80°C, 100°C, 120°C and 150°C. This is because B3 has the largest number of layers (6 layers) with the arrangement of CF tow (0°/90°), and both B1 and C have 4 layers, and D only has 3 layers. CF tow can bear more tensile strength along the direction of stretching. Experimental sample B3 can carry more load. The arrangement of CF tow has a great effect on the mechanicalproperties of the experimental samples at high temperature. It is possible to increase the number of layers with the arrangement of (0°/90°) in practical application without affecting other properties.

Tensile Strength at Different Temperatures
It takes a long time to test the mechanicalproperties of the composite at high temperature. The mechanicalproperties at high temperature are only measured at dispersed temperature. It is thought that the mechanical parameters of high temperature at different temperatures can only be obtained by experiments. The theoretical analysis of high temperature mechanicalproperties is of great significance to improve the test method, analyze the test results and obtain the mechanicalproperties at continuous temperature [52]. At present, the research on the mechanicalproperties of materials is mainly carried out at room temperature, and mainly focused on the high temperature mechanicalproperties of materials. The theoretical study on the mechanicalproperties of CFEC at high temperature is rare [53]. The experimental data are polynomial fitted after the mechanicalproperties of the material are tested in the high temperature environment. The relationship between the mechanicalproperties of the material and the corresponding temperature was obtained [54]: In the above expression, f refers to the mechanicalproperties (strength, rigidity or adhesive force) at a certain temperature T, f o refers to the room temperature value of the mechanicalproperties, a is the assumed constant describing the residual value of the mechanicalproperties, and b and c are the constants (using least square regression analysis) derived from experience to describe the center temperature (c) and the degradation severity (b) with the change of temperature.
However, the goodness of fit of the above formula is very low in the lower range (from 25°C to 150°C). The above formula is more suitable for high temperature environment [54]. Therefore, the relationship curve between the tensile strength of the experimental samples under each temperature and the corresponding temperature was drawn by regression analysis in the environment of 25°C-150°C. Regression analysis and test standards are in accordance with ISO-527-3 [55]. Each experimental sample chooses the best fitting equation as its own fitting function by curve fitting. Tab. 5 shows the best fitting function and corresponding goodness of fit R 2 of four experimental samples.   The carbon fiber/epoxy woven cloth (C) applied to the actual fuselage material is taken as an example for comprehensive evaluation based on the above research. Considering the mass limitation of aircraft, the number of layers of CF tow is determined to be 8. Tab. 7 shows the influence of different arrangement of CF tow on each parameter.
Therefore, when the carbon fiber/epoxy woven cloth (C) is applied to the fuselage, the number of layers (0°/90°) can be increased at the place with large load under the premise that the total number of layers remains unchanged. In places where high temperature is easy to cause fire, such as near the engine, it can be considered to increase the number of layers (± 45°).

Conclusion
1. The more layers of CF tow, the higher the quality of epoxy resin and the higher the heat capacity. As a result, the carbon fiber epoxy/composite is more difficult to burn, the burning rate is lower, and the Peak-HHR is increased. When the carbon fiber bundle is (± 45°), the flame spread is restrained because of the existence of CF tow. 2. When the arrangement of CF tow is (0°/90°), the composite has better mechanicalproperties compared with the arrangement of CF tow (± 45°). When the temperature range is from 100°C to 150°C, the mechanicalproperties of the composites decrease obviously due to the partial separation of the ER matrix and the reinforcement CF tow. 3. The mechanicalproperties of six temperature points were tested and the experimental results were fitted.
The fitting functions of four samples (B1, B3, C and D) were obtained. The theoretical tensile strength at 40°C and 130°C was calculated, and the error was less than 4%. This function can calculate the mechanicalproperties of four samples in the range of 25°C-150°C. The thermomechanical properties of the composite can be predicted. 4. Take carbon fiber cloth (C) used in fuselage material as an example. Considering the influence factors of various parameters of the fire site, it can be considered to increase the number of layers (0°/90°) where the load is large. Increase the number of floors (± 45°) in the place where high temperature is prone to fire.
Funding Statement: This work was sponsored by Project 51874313 supported by National Natural Science Foundation of China.

Conflicts of Interest:
The authors declare that they have no conflicts of interest to report regarding the present study.