The Study of Functional Glass Fiber Veils for Composites Protection: Flame Resistance and Mechanical Performance

: The flame-retardant performance of carbon fiber-reinforced composites is crucial for ensuring structural stability. Traditional additive flame-retardant methods often struggle to balance structural integrity with fire resistance. Herein, Ni(OH) 2 and 9,10-dihydro-9-oxa-10-phosphaphenanthrene-10-oxide (DOPO) were used as flame-retardant agents and mixed with glass fibers to construct the flame-retardant functional fiber veil which was used as the skin layer on the composite surface for fire protection. The structure performance and flame retardancy of composites were characterized via Fourier transform infrared spectroscopy (FTIR), scanning electron microscope (SEM), and a cone calorimeter test. The results confirmed that a flame-retardant glass fiber mat could effectively improve the flame-retardant and smoke-suppressive properties of the composite material. Due to the synergistic flame-retardant mechanism of Ni(OH) 2 and DOPO, the C-N3-D2 composite with the highest LOI value of 32.3% has shown significant reduction in peak heat release rate (PHRR) and total smoke production (TSP) by 31.3% and 19.5%, respectively. In addition, due to flame-retardant agents only being employed in the skin layer of the composite, the core layer of a carbon fiber-reinforced structure could be protected without structure disruption. This approach maintained consistent interlayer shear strength, highlighting the effectiveness of using a flame-retardant fiber veil as a protective skin layer. This strategy could offer a viable solution for safeguarding high-performance composite materials from fire hazards without compromising their structural integrity.


Introduction
Polymer composites reinforced by high-performance fibers have been widely adopted in several cutting-edge fields, including aviation, aerospace, and rail transportation as well as marine applications.This is attributed to their excellent specific strength and environmental resistance [1].However, due to the flammability of a polymer matrix such as epoxy, polymer composites could be burnt off with smoke emissions causing health risks when exposed to combustion environments, and their structural stability could be significantly affected leading to composite structure failure [2].With the increase in demand for highperformance composites, the improvement of fire protection for composites becomes the current research key point [3].
Enhancing the flame-retardant properties of carbon fiber-reinforced composites is crucial for increasing their commercial value in industrial applications.The prevailing commercial approach involves incorporating flame retardant additives into the polymer matrix [4].These additives, typically in particulate or liquid form, are physically dispersed within the resin matrix, which can significantly compromise the structural integrity of the composites [5].Inorganic flame retardants, despite their advantages of being non-toxic and smokeless, must be used in large volumes, which adversely affect the fiber-polymer matrix interface.Additionally, halogen-based flame retardants, while highly efficient in flame suppression, release toxic gases during combustion [6].
Currently, incorporating fire-retardant barriers within polymer composites as fiber protection has emerged as an effective alternative to mitigate heat transfer and enhance the structural stability of these composites [7].Kandola et al. proposed for the first time a heat-insulating functional fiber mat, which was used to wrap the surface of the composite to form a heat-insulating barrier showing a reduction in heat release rate of 50% [8,9].Li et al. developed a carbon fiber-reinforced composite covered with a flameretardant mat, resulting in a decrease in heat release with better structure stability [10].Subsequently, Cong et al. validated this flame-retardant design and enhanced surface fire protection by optimizing the composition of fire retardants [11].Nevertheless, concerns have arisen regarding the insulative thermal barrier provided by these composites, as they have been linked to the generation of unexpected toxic smoke and surface structural failures attributed to limited oxygen availability in the pyrolysis region, hindering complete combustion [12,13].To reduce the smoke emission from combustion, it is necessary to consider designing smoke-suppression functional units in the surface flame retardant structure, so that smoke can be captured instantly when it penetrates the surface flame retardant layer.
According to the literature, nickel hydroxide (Ni(OH) 2 ) employed as an inorganic flame retardant presents the function of decomposition, with heat absorption close to that of traditional inorganic flame retardants such as calcium hydroxide [14].In this case, the decomposed product NiO catalyzes with the carbon particles in the smoke and converts them to carbon dioxide [15].On the other hand, as a kind of phosphorus flame retardant, DOPO is not only compatible with the epoxy resin, but also has high efficiency and stable cohesive phase flame-retardant performance [16].Normally, the synergistic effect between flame retardants can be achieved via proper combinations of different mechanisms [17].Based on this theory, the mixture of DOPO and Ni(OH) 2 is expected to present multifunctional behavior for flame retardance and smoke suppression.However, the final performance of synergistic effect for flame-retardant agent combination depends on the material composition and chemical reaction mechanism during the combustion process [18].
Therefore, based on the high demand for composites with excellent flame retardance and structure performance, it is important to develop flame-retardant veils for composite protection, and investigate the synergistic mechanism of different flame-retardant agents for fire resistance and smoke suppression.Therefore, based on DOPO and Ni(OH) 2 , this study proposes a kind of synergistic flame-retardant fiber veil with glass fibers and flame-retardant agents as the surface protection layer "Skin", to protect the carbon fiber reinforced epoxy structure "Core", with a "sandwich" structure.This study will evaluate the flame-retardant veils first.The chemical structure and capability of impregnation with epoxy resin were characterized via Fourier transform infrared (FTIR) spectroscopy and contact angle analysis, respectively.Thus, the basic properties of functional veils could be determined before further evaluation of flame-retardant composites.To understand the effect of flame-retardant veils on the composite structure with the synergistic mechanism of flame-retardant agents, the thermal stability, flame retardancy, and mechanical performance were assessed through scanning electron microscopy (SEM), thermogravimetric analysis (TGA), cone calorimetry, and mechanical testing such as flexural, interlaminar shear, and impact tests.

Preparation of Ni(OH) 2
The aqueous solution of nickel sulfate was prepared by dissolving NiSO 4 •6H 2 O (10 g, 38.04 mmol) in deionized water (50 mL).Meanwhile, the aqueous solution of NaOH (0.76 g, 19.00 mmol) was prepared in deionized water (3.8 mL) under continuous stirring [19].Then, both of them were mixed and stirred with a magnetic stirrer for 2 h before being transferred to a 100 mL autoclave.
The hydrothermal reaction was carried out at 120 • C for 24 h.The produced green precipitate was filtered using filtration paper and then washed with deionized water, which was finally moved into a freeze dryer for drying, to isolate the final Ni(OH) 2 .

Manufacture of Glass Fiber Veil with Flame-Retardant Agents
Based on the paper-making process (shown in Figure 1), 1 g of PEO was mixed with 500 mL of deionized water to form a dissociative solution, and then 2 g of glass fibers were dispersed and dissociated with a fiber standard dissociator for 15 min to obtain a glass fiber slurry, which was then put into a copying machine for extraction and filtration.

Preparation of Ni(OH)2
The aqueous solution of nickel sulfate was prepared by dissolving NiSO4•6H2O (10 g, 38.04 mmol) in deionized water (50 mL).Meanwhile, the aqueous solution of NaOH (0.76 g, 19.00 mmol) was prepared in deionized water (3.8 mL) under continuous stirring [19].Then, both of them were mixed and stirred with a magnetic stirrer for 2 h before being transferred to a 100 mL autoclave.
The hydrothermal reaction was carried out at 120 °C for 24 h.The produced green precipitate was filtered using filtration paper and then washed with deionized water, which was finally moved into a freeze dryer for drying, to isolate the final Ni(OH)2.

Manufacture of Glass Fiber Veil with Flame-Retardant Agents
Based on the paper-making process (shown in Figure 1), 1 g of PEO was mixed with 500 mL of deionized water to form a dissociative solution, and then 2 g of glass fibers were dispersed and dissociated with a fiber standard dissociator for 15 min to obtain a glass fiber slurry, which was then put into a copying machine for extraction and filtration.
Thus, the preformed body of glass fiber mats was prepared with a thickness of 0.1 mm.Furthermore, Ni(OH)2 was evenly sprayed on the surface of the glass fiber mats with DOPO particles, and then the formed undried fiber mat was dried in a plate dryer at 60 °C to obtain a functional fiber mat enriched with flame retardant.

Design and Molding of Flame-Retardant Composites
As shown in Figure 1, the pre-cut carbon fiber fabrics were stacked layer by layer as the main fiber-reinforced structure (core layer), and then the carbon fiber stack was covered with glass fiber veils as a skin layer, then the whole stacked fiber structure was put into a mold and introduced into the epoxy resin, and then molded and cured for 1 h in a hot press at 80 °C with a pressure of 2 bar by the molding process.
The study investigated the behavioral changes of flame-retardant fiber mats in the combustion of flame-retardant carbon fiber reinforced composites with different ratios of DOPO and Ni(OH)2.As can be seen from Table 1, there were 7 groups designed for the flame-retardant glass fiber veil, named GF-N0-D5, GF-N1-D4, GF-N2-D3, GF-N3-D2, GF-N4-D1, GF-N5-D0, and GF-N0-D0, respectively, followed by composites named C-N0-D5, Thus, the preformed body of glass fiber mats was prepared with a thickness of 0.1 mm.Furthermore, Ni(OH) 2 was evenly sprayed on the surface of the glass fiber mats with DOPO particles, and then the formed undried fiber mat was dried in a plate dryer at 60 • C to obtain a functional fiber mat enriched with flame retardant.

Design and Molding of Flame-Retardant Composites
As shown in Figure 1, the pre-cut carbon fiber fabrics were stacked layer by layer as the main fiber-reinforced structure (core layer), and then the carbon fiber stack was covered with glass fiber veils as a skin layer, then the whole stacked fiber structure was put into a mold and introduced into the epoxy resin, and then molded and cured for 1 h in a hot press at 80 • C with a pressure of 2 bar by the molding process.

Characterization for Glass Fiber Veil
Contact angle for veil: the contact angle between the veils and epoxy was measured using a DSA100 contact angle tester.The final contact angle was determined by averaging at least five separate measurements taken across the surface, allowing for a comprehensive assessment of the fiber's hydrophobicity.
Chemical structure of veil: to comprehensively understand the chemical structure changes of the glass fiber veil before and after modification, we employed an FTIR spectrometer (S20, Thermo Scientific Nicoleti, Waltham, MA 02451, USA).The samples were prepared using the potassium bromide compression method, and the analysis was conducted over a wave number range of 400-4000 cm −1 .

Characterization of Composites
Thermal stability: the thermal stability of the composite was assessed using a thermogravimetric analyzer (Netzsch 209F1, Bavaria, Germany).The analysis was performed in a nitrogen atmosphere with a flow rate of 35 mL/min.The temperature ranged from 30 • C to 800 • C, with a constant heating rate of 20 • C/min.
Limiting oxygen index test: according to the ISO 4589-2 standard [20], the sample size was 100.0 mm × 10.0 mm × 2.0 mm, and the LOI value was calculated by observing whether it burned or not at the set oxygen concentration.
Vertical flammability test (UL-94): according to the UL94-2013 test standard, the sample size was 100.0 mm × 100.0 mm × 2.0 mm, and the vertical flammability test was carried out on a flame with a length of 40 mm to determine the fire resistance of the composite material.
Cone calorimetry test: according to the ISO 5660-1-2015 standard [21], the sample size was 100.0 mm × 100.0 mm × 2.0 mm, and tested in the heat flux of 50 kW/m 2 .The key data such as time to ignition (TTI), heat release rate (HRR), total heat release (THR), and smoke production rate (TSR) were recorded.
Mechanical test: the composite test was carried out using a universal testing machine (LD25.504;Lvliang, China) in accordance with the ISO 178-2010 standard [22].The test was performed with a span of 32 mm and a crosshead speed of 2 mm/min.Five specimens for each group were tested, and the results were calculated in average.The interlaminar shear strength (ILSS) was evaluated following the ISO 14130-1997 standard [23].The notch impact strength of the composite was characterized via a plastic pendulum impact testing machine (LZ21.251-B;Lvliang, China), according to the ISO 179-2010 standard [24], without notch spline.
Surface morphology analysis: the fracture surfaces of the composites, after the threepoint bending test, were examined using a scanning electron microscope (SEM).This analysis aimed to investigate the interface between the layers of the glass fiber mat and carbon fiber (CF).The SEM used was a Hitachi Regulus 8100, operating at a voltage of 10 kV.

Structural Analysis of Flame-Retardant Fiber Veil
The flame-retardant agent treatment can substantially affect the chemical structure and wettability of veils, thus it is essential to study the contact angle of epoxy resin to veils and gain insights into their wettability characteristics.As shown in Figure 2a, no significant difference could be observed for the contact angle of dropped epoxy to veils (GF-N0-D0, GF-N0-D5, GF-N2-D3, and GF-N5-D0).However, after resin soaking at GF veil surface for 10 s, a decrease of contract angle from 43.3 • to 30.1 • was observed for GF veils with the addition of DOPO, which was continuously replaced by Ni(OH) 2 when compared with the control GF-N0-D0 veil.Normally, a lower contact angle for epoxy resin indicates a beneficial impregnation of the glass fiber veil with better structure stability of composites.In this study, with the addition of a flame-retardant agent such as DOPO and Ni(OH) 2 , the contact angle for initial contact was consistent, whilst a decrease could be found after infiltration.This was due to the effect on the surface roughness of GF veils, which was attached with nano particles of DOPO and Ni(OH) 2 .
Surface morphology analysis: the fracture surfaces of the composites, after the threepoint bending test, were examined using a scanning electron microscope (SEM).This analysis aimed to investigate the interface between the layers of the glass fiber mat and carbon fiber (CF).The SEM used was a Hitachi Regulus 8100, operating at a voltage of 10 kV.

Structural Analysis of Flame-Retardant Fiber Veil
The flame-retardant agent treatment can substantially affect the chemical structure and wettability of veils, thus it is essential to study the contact angle of epoxy resin to veils and gain insights into their wettability characteristics.As shown in Figure 2a, no significant difference could be observed for the contact angle of dropped epoxy to veils (GF-N0-D0, GF-N0-D5, GF-N2-D3, and GF-N5-D0).However, after resin soaking at GF veil surface for 10 s, a decrease of contract angle from 43.3° to 30.1° was observed for GF veils with the addition of DOPO, which was continuously replaced by Ni(OH)2 when compared with the control GF-N0-D0 veil.Normally, a lower contact angle for epoxy resin indicates a beneficial impregnation of the glass fiber veil with better structure stability of composites.In this study, with the addition of a flame-retardant agent such as DOPO and Ni(OH)2, the contact angle for initial contact was consistent, whilst a decrease could be found after infiltration.This was due to the effect on the surface roughness of GF veils, which was attached with nano particles of DOPO and Ni(OH)2.To investigate the chemical functional groups of the glass fiber veil modified by the flame-retardant agent, Fourier transform infrared (FTIR) spectroscopy was employed.As can be seen from Figure 2b, the band at 894 cm −1 in the GF-N0-D5 and GF-N2-D3 veils can be attributed to the P-O-C asymmetric stretching vibration of DOPO [25,26], while the symmetric and antisymmetric stretching of carbonate anions present in the interlayer space of Ni(OH)2 is exhibited at 1630 cm −1 for the GF-N5-D0 and GF-N2-D3 veils [27,28].As such, Ni(OH)2 and DOPO were confirmed to be coated on the GF veil's surface.

Characterization of Flame-Retardant Properties of Composites
Thermogravimetric analysis (TGA) was utilized to assess the mass loss of flame-retardant composites containing different flame-retardant agents in a nitrogen atmosphere.The pyrolysis process, depicted in Figure 3a, indicates that the weight of the specimens began to decrease at 100 °C, likely due to moisture evaporation from the composites.The TGA curves reveal two primary thermal decomposition stages for composite specimens with the GF veils of GF-N0-D5 and GF-N2-D3.The first stage, occurring between 200 °C and 300 °C, could be related to the breakdown of epoxy impregnated within the GF veils.To investigate the chemical functional groups of the glass fiber veil modified by the flame-retardant agent, Fourier transform infrared (FTIR) spectroscopy was employed.As can be seen from Figure 2b, the band at 894 cm −1 in the GF-N0-D5 and GF-N2-D3 veils can be attributed to the P-O-C asymmetric stretching vibration of DOPO [25,26], while the symmetric and antisymmetric stretching of carbonate anions present in the interlayer space of Ni(OH) 2 is exhibited at 1630 cm −1 for the GF-N5-D0 and GF-N2-D3 veils [27,28].As such, Ni(OH) 2 and DOPO were confirmed to be coated on the GF veil's surface.

Characterization of Flame-Retardant Properties of Composites
Thermogravimetric analysis (TGA) was utilized to assess the mass loss of flameretardant composites containing different flame-retardant agents in a nitrogen atmosphere.The pyrolysis process, depicted in Figure 3a, indicates that the weight of the specimens began to decrease at 100 • C, likely due to moisture evaporation from the composites.The TGA curves reveal two primary thermal decomposition stages for composite specimens with the GF veils of GF-N0-D5 and GF-N2-D3.The first stage, occurring between 200 • C and 300 • C, could be related to the breakdown of epoxy impregnated within the GF veils.The second stage, between 300 • C and 420 • C, corresponds to the decomposition of the epoxy's molecular chains in the carbon fiber-reinforced region, along with the breakdown of DOPO and Ni(OH) 2 , which contributes to combustion resistance.On the other hand, for comparison between C-N0-D0 (control) and C-N5-D0, the addition of Ni(OH) 2 for GF veils can affect the pyrolysis process of composites as well as DOPO, thus increasing the residue of composites by the end of testing from 53.2% (C) to 55.4%.
The second stage, between 300 °C and 420 °C, corresponds to the decomposition of the epoxy's molecular chains in the carbon fiber-reinforced region, along with the breakdown of DOPO and Ni(OH)2, which contributes to combustion resistance.On the other hand, for comparison between C-N0-D0 (control) and C-N5-D0, the addition of Ni(OH)2 for GF veils can affect the pyrolysis process of composites as well as DOPO, thus increasing the residue of composites by the end of testing from 53.2% (C) to 55.4%.To assess the fire safety performance of composite materials under realistic fire conditions, several tests were conducted via the limiting oxygen index (LOI) test, the UL-94 vertical burning test, and the cone calorimeter test.These evaluations provided comprehensive insights into the combustion behavior of the materials.As shown in Table 2, all the specimens showed continuous burning without dripping (V-1) for the UL-94 vertical burning test, and the limiting oxygen index (LOI) (shown in Figure 3b) also changed significantly with the addition of GF veils.
In addition, the C-N0-D0 composite covered by glass fiber veil without flame-retardant agents presented a limiting oxygen index of 22.6%, close to the control (LOI: 22.4%), which was related to the heat insulation behavior of the glass fiber on the surface of the composite and also confirmed by the research of Kandola et al. [29,30].With the addition of flame-retardant agents for C-N-D composites, the LOI values were observed to increase significantly when compared to the control.With further comparison, the LOI values of C-N-D composites were observed to increase from 27.6% to a peak value of 31.6%, and were then followed by a decrease to 25.1% with the substitution of DOPO by Ni(OH)2.This change can be attributed to the synergistic flame-retardant effect between DOPO and Ni(OH)2 for composite fire safety improvement.To assess the fire safety performance of composite materials under realistic fire conditions, several tests were conducted via the limiting oxygen index (LOI) test, the UL-94 vertical burning test, and the cone calorimeter test.These evaluations provided comprehensive insights into the combustion behavior of the materials.As shown in Table 2, all the specimens showed continuous burning without dripping (V-1) for the UL-94 vertical burning test, and the limiting oxygen index (LOI) (shown in Figure 3b) also changed significantly with the addition of GF veils.In addition, the C-N0-D0 composite covered by glass fiber veil without flame-retardant agents presented a limiting oxygen index of 22.6%, close to the control (LOI: 22.4%), which was related to the heat insulation behavior of the glass fiber on the surface of the composite and also confirmed by the research of Kandola et al. [29,30].With the addition of flame-retardant agents for C-N-D composites, the LOI values were observed to increase significantly when compared to the control.With further comparison, the LOI values of C-N-D composites were observed to increase from 27.6% to a peak value of 31.6%, and were then followed by a decrease to 25.1% with the substitution of DOPO by Ni(OH) 2 .This change can be attributed to the synergistic flame-retardant effect between DOPO and Ni(OH) 2 for composite fire safety improvement.
The combustion and heat release behavior of the composites at elevated temperatures can be characterized by cone calorimetry testing.The key parameters for time to ignition (TTI), heat release rate (HRR), total heat release (THR), and total smoke production (TSP) were analyzed.As shown in Table 2, the TTI values of the composite with GF veils indicated the significant delay for ignition when compared to the control, which was attributed to the incorporation of the synergistic flame-retardant agent and the thermal stability of the "Skin" layer with glass fiber.With the addition of Ni(OH) 2 into the flame-retardant agent, the ignition time of the composites can reach up to 24 s.This is due to the synergistical effect from both DOPO and Ni(OH) 2 in flame-retardant GF veils.The charcoal layer formed by DOPO was able to prevent flames in initial ignition [26,31], while the H 2 O released from Ni(OH) 2 decomposition could absorb the heat from the pyrolysis of resin in the "Skin" region, thus this synergistical effect enhanced the effect of the synergistic effect under the appropriate ratio [32].
For the heat release behavior of composites with GF veils shown in Figure 4a, the value of the peak heat release rate (PHRR) for the C-N0-D0 composite without flame retardant agents was 594.18 kW/m 2 , which was reduced by 6.3% compared with that of the control (634.16kW/m 2 ).This indicated that the GF veils on the "Skin" of the composite also could effectively reduce the heat transfer and reduce the related release rate.As such, the introduction of flame-retardant agents into GF veils should be expected to improve the flame-resistant capability.
can be characterized by cone calorimetry testing.The key parameters for time to ignition (TTI), heat release rate (HRR), total heat release (THR), and total smoke production (TSP) were analyzed.As shown in Table 2, the TTI values of the composite with GF veils indicated the significant delay for ignition when compared to the control, which was attributed to the incorporation of the synergistic flame-retardant agent and the thermal stability of the "Skin" layer with glass fiber.With the addition of Ni(OH)2 into the flameretardant agent, the ignition time of the composites can reach up to 24 s.This is due to the synergistical effect from both DOPO and Ni(OH)2 in flame-retardant GF veils.The charcoal layer formed by DOPO was able to prevent flames in initial ignition [26,31], while the H2O released from Ni(OH)2 decomposition could absorb the heat from the pyrolysis of resin in the "Skin" region, thus this synergistical effect enhanced the effect of the synergistic effect under the appropriate ratio [32].
For the heat release behavior of composites with GF veils shown in Figure 4a, the value of the peak heat release rate (PHRR) for the C-N0-D0 composite without flame retardant agents was 594.18 kW/m 2 , which was reduced by 6.3% compared with that of the control (634.16kW/m 2 ).This indicated that the GF veils on the "Skin" of the composite also could effectively reduce the heat transfer and reduce the related release rate.As such, the introduction of flame-retardant agents into GF veils should be expected to improve the flame-resistant capability.In addition, the PHRR value of the C-N0-D5 composite (578.19 kW/m 2 ) with DOPO was further decreased when compared to the control, which could be due to the organic DOPO that could protect the composites from heat flux through the formation of a dense char layer [33], whilst the C-N5-D0 composite with only Ni(OH)2 presented a lower value of 537.03 kW/m 2 for PHRR, attributed to the absorption of heat released from resin matrix pyrolysis via the decomposition of Ni(OH)2 [34].Furthermore, the mixture of flame-retardant agents for GF veils was observed to positively improve the flame resistance.It was In addition, the PHRR value of the C-N0-D5 composite (578.19 kW/m 2 ) with DOPO was further decreased when compared to the control, which could be due to the organic DOPO that could protect the composites from heat flux through the formation of a dense char layer [33], whilst the C-N5-D0 composite with only Ni(OH) 2 presented a lower value of 537.03 kW/m 2 for PHRR, attributed to the absorption of heat released from resin matrix pyrolysis via the decomposition of Ni(OH) 2 [34].Furthermore, the mixture of flameretardant agents for GF veils was observed to positively improve the flame resistance.It was evident that C-N3-D2 composites showed the lowest value of PHRR (435.55 kW/m 2 ) when compared to other composites due to the synergistic effect between Ni(OH) 2 and DOPO.The Ni(OH) 2 particles in the GF veil simultaneously decomposed with the releasing H 2 O molecules to reduce heat release, while DOPO reacted with epoxy during combustion to form a char layer, preventing heat transfer into the "Core" region of the composite [35].Thus, the combustion behavior of the composite could be synergistically reduced with heat absorption and thermal insulation.
Meanwhile, from the total heat release (THR) of the composites as shown in Figure 4b, it can be seen that the THR value for the composite with pure GF veils (C-N0-D0) was observed to decrease by 6.1% when compared to the control.In addition, for flame retardant composites, the THR value was reduced from 95.22 MJ/m 2 (C-N5-D0) to 75.24 MJ/m 2 (C-N0-D5) with the substitution of Ni(OH) 2 by DOPO, indicating that the mixture of DOPO and Ni(OH) 2 could only reduce the PHRR value effectively with formation of a thermal barrier for composite surface protection.Whereas, for the THR value for composites related to a composite full combustion process, DOPO was able to release phosphorus-containing free radicals during combustion, capturing highly reactive hydrogen and oxygen free radicals in the gas phase, thereby interrupting the combustion chain reaction resulting in reduction of total heat release [36].
Total smoke production (TSP) represents the key factor endangering life safety.As shown in Figure 4c, the composite C-N0-D0 without flame retardant agents exhibited a TSP value of 13.40 m 2 , which was an increase of 8.3% compared to the control (12.54 m 2 ).This was attributed to the fact that the GF veils effectively protect the composites during the thermal insulation process, also resulting the partial combustion of epoxy due to insufficient oxygen exchange, thus releasing extra smoke particles.However, with an increase of Ni(OH) 2 in GF veils, the TSP value for composites was observed to decrease and reach the lowest value of 10.09 m 2 for the C-N3-D2 composite.This could be related to Ni(OH) 2 during combustion, capturing the carbon of smoke via a catalytic reaction to convert them into non-toxic CO 2 [35].
In order to further investigate the synergistic behavior of the two flame-retardant agents in the GF veil for composite fire protection, the scanning electron microscopy (SEM) analyses of the composites after cone calorimeter tests were carried out in this study.As shown in Figure 5, without the addition of flame-retardant agents, the surface of the control was relatively smooth without obvious residual char due to the absence of glass fibers.In contrast, on the surface of the C-N0-D0 composite with pure GF veil random glass fibers could be seen accompanied by a certain amount of loose residue char.Different from this, it can be clearly seen from the composites with treatment of DOPO for GF veil that the expanded char layer was formed on the surface of the composite due to the action of DOPO, which forms a dense and stable char layer with glass fibers for heat insulation and reduction of heat transfer.However, with the addition of Ni(OH) 2 to substitute DOPO, the dense structure of residual char was observed to be sparse with reduction of residual char on the composite surface.
evident that C-N3-D2 composites showed the lowest value of PHRR (435.55 kW/m 2 ) when compared to other composites due to the synergistic effect between Ni(OH)2 and DOPO.The Ni(OH)2 particles in the GF veil simultaneously decomposed with the releasing H2O molecules to reduce heat release, while DOPO reacted with epoxy during combustion to form a char layer, preventing heat transfer into the "Core" region of the composite [35].Thus, the combustion behavior of the composite could be synergistically reduced with heat absorption and thermal insulation.
Meanwhile, from the total heat release (THR) of the composites as shown in Figure 4b, it can be seen that the THR value for the composite with pure GF veils (C-N0-D0) was observed to decrease by 6.1% when compared to the control.In addition, for flame retardant composites, the THR value was reduced from 95.22 MJ/m 2 (C-N5-D0) to 75.24 MJ/m 2 (C-N0-D5) with the substitution of Ni(OH)2 by DOPO, indicating that the mixture of DOPO and Ni(OH)2 could only reduce the PHRR value effectively with formation of a thermal barrier for composite surface protection.Whereas, for the THR value for composites related to a composite full combustion process, DOPO was able to release phosphorus-containing free radicals during combustion, capturing highly reactive hydrogen and oxygen free radicals in the gas phase, thereby interrupting the combustion chain reaction resulting in reduction of total heat release [36].
Total smoke production (TSP) represents the key factor endangering life safety.As shown in Figure 4c, the composite C-N0-D0 without flame retardant agents exhibited a TSP value of 13.40 m 2 , which was an increase of 8.3% compared to the control (12.54 m 2 ).This was attributed to the fact that the GF veils effectively protect the composites during the thermal insulation process, also resulting the partial combustion of epoxy due to insufficient oxygen exchange, thus releasing extra smoke particles.However, with an increase of Ni(OH)2 in GF veils, the TSP value for composites was observed to decrease and reach the lowest value of 10.09 m 2 for the C-N3-D2 composite.This could be related to Ni(OH)2 during combustion, capturing the carbon of smoke via a catalytic reaction to convert them into non-toxic CO2 [35].
In order to further investigate the synergistic behavior of the two flame-retardant agents in the GF veil for composite fire protection, the scanning electron microscopy (SEM) analyses of the composites after cone calorimeter tests were carried out in this study.As shown in Figure 5, without the addition of flame-retardant agents, the surface of the control was relatively smooth without obvious residual char due to the absence of glass fibers.In contrast, on the surface of the C-N0-D0 composite with pure GF veil random glass fibers could be seen accompanied by a certain amount of loose residue char.Different from this, it can be clearly seen from the composites with treatment of DOPO for GF veil that the expanded char layer was formed on the surface of the composite due to the action of DOPO, which forms a dense and stable char layer with glass fibers for heat insulation and reduction of heat transfer.However, with the addition of Ni(OH)2 to substitute DOPO, the dense structure of residual char was observed to be sparse with reduction of residual char on the composite surface.Based on the combustion behavior and microstructure characterization of the composites, the flame-retardant mechanism of carbon fiber-reinforced composites can be analyzed.Typically, composite material combustion involves a gas phase and a condensed phase, and the smoke is normally generated via partial combustion of the resin in the gas phase [37].As can be seen from Figure 6, the pyrolysis of epoxy resin can release a large number of combustion radicals (H• and OH•) with toxic smoke, while the flame-retardant agents in the composite skin layer will decompose as well.DOPO can induce resin matrix carbonization, thus promoting the formation of the initial char layer in the condensed phase.At the same time, the phosphorus-oxygen radicals (PO-and -PO 2 ) decomposed from DOPO capture the H• and OH• radicals, which can inhibit or terminate the pyrolysis chain reaction, thus achieving the purpose of protecting the lower layer of the "Core" region [38].
ant agents in the composite skin layer will decompose as well.DOPO can induce resin matrix carbonization, thus promoting the formation of the initial char layer in the condensed phase.At the same time, the phosphorus-oxygen radicals (PO-and -PO2) decomposed from DOPO capture the H• and OH• radicals, which can inhibit or terminate the pyrolysis chain reaction, thus achieving the purpose of protecting the lower layer of the "Core" region [38].
Additionally, the endothermic reaction with water molecular release can be observed from the decomposition of Ni(OH)2, which is able to dilute oxygen and other flammable gases and absorbs heat via H2O evaporation, finally reducing the temperature and inhibiting the combustion behavior of the composites [39].On the other hand, Ni(OH)2 is able to catch and react with carbon particles of smoke released from partial combustion, then transfer them to non-combustible gases such as H2O and CO2, which not only further dilutes the combustible gases, but also reduces the smoke particles [40].In addition, the Ni and NiO generated during the decomposition of Ni(OH)2 can be used as an insulation layer with a high melting point, which further stabilizes the surface structure of the char layer with improved thermal insulation [41].Therefore, based on the flame-retardant study, efficient synergistic flame-retardant composites can be achieved via a flame-retardant GF veil design for composite surface protection, and the flame-retardant agent's mixture of DOPO and Ni(OH)2 with acceptable ratio thus exhibited excellent flame resistance, thermal insulation, and smoke-suppression behavior.

Mechanical Characterization
The addition of organic or inorganic flame-retardant agents into a composite matrix has been reported to disrupt the fiber-matrix interface and structure stability.As such, this study introduced the sandwich structure design for flame retardant composites, where the GF veils with different flame-retardant agents act as a skin layer to protect the composite from heat flux, and the flame-retardant agents are only concentrated in the skin Additionally, the endothermic reaction with water molecular release can be observed from the decomposition of Ni(OH) 2 , which is able to dilute oxygen and other flammable gases and absorbs heat via H 2 O evaporation, finally reducing the temperature and inhibiting the combustion behavior of the composites [39].On the other hand, Ni(OH) 2 is able to catch and react with carbon particles of smoke released from partial combustion, then transfer them to non-combustible gases such as H 2 O and CO 2 , which not only further dilutes the combustible gases, but also reduces the smoke particles [40].In addition, the Ni and NiO generated during the decomposition of Ni(OH) 2 can be used as an insulation layer with a high melting point, which further stabilizes the surface structure of the char layer with improved thermal insulation [41].
Therefore, based on the flame-retardant study, efficient synergistic flame-retardant composites can be achieved via a flame-retardant GF veil design for composite surface protection, and the flame-retardant agent's mixture of DOPO and Ni(OH) 2 with acceptable ratio thus exhibited excellent flame resistance, thermal insulation, and smoke-suppression behavior.

Mechanical Characterization
The addition of organic or inorganic flame-retardant agents into a composite matrix has been reported to disrupt the fiber-matrix interface and structure stability.As such, this study introduced the sandwich structure design for flame retardant composites, where the GF veils with different flame-retardant agents act as a skin layer to protect the composite from heat flux, and the flame-retardant agents are only concentrated in the skin region so that the side-effect on the resin matrix as well as the fiber-matrix interface could be avoided.
As can be seen in Figure 7a and Table 3, the flexural strength and flexural modulus of control composites without GF veils were 604.93 MPa and 53.61 GPa, respectively, whereas the decrease of flexural strength and modulus were observed for composites with GF veils.This was attributed to the sandwich structure of the composites, where the strength of the glass fiber structure of the skin layer is lower than that of the carbon fiber-reinforced core region.Thus, the stresses concentrated in the surface layer should be much higher than that in the core region during flexural testing, whereas the skin layer with glass fiber reinforcement is more prone to fracture failure resulting in lower flexural properties [42].With substitution of DOPO by Ni(OH) 2 , the flexural strength and modulus were found to decrease from 508.77 MPa and 51.92 GPa for the C-N0-D5 composite to 470.22 MPa and 46.53 GPa for the C-N5-D0 composite, respectively.This could be due to the inorganic Ni(OH) 2 particles in the glass fiber veil disrupting the structure stability resulting in a poor interface of the skin layer for flame retardant composites [43].For interlaminar shear strength shown in Figure 7b, the composites with GF veils exhibited consistent values when compared with that of the control.This indicated that that the flame-retardant veils could protect composites without structural influence on the core region by the addition of flame retardant agents.
For the impact strength shown in Figure 7b, composite C-N0-D0 was found to increase from 920.23 KJ m −2 to 1128.24KJ m −2 when compared to the control, indicating that GF veils covering composite surfaces could improve the composite's toughness.Furthermore, with the addition of DOPO and Ni(OH)2, a decrease in impact strength was evaluated for C-N-D composites.When DOPO was replaced by Ni(OH)2, the interlaminar shear  For interlaminar shear strength shown in Figure 7b, the composites with GF veils exhibited consistent values when compared with that of the control.This indicated that that the flame-retardant veils could protect composites without structural influence on the core region by the addition of flame retardant agents.
For the impact strength shown in Figure 7b, composite C-N0-D0 was found to increase from 920.23 KJ m −2 to 1128.24KJ m −2 when compared to the control, indicating that GF veils covering composite surfaces could improve the composite's toughness.Furthermore, with the addition of DOPO and Ni(OH) 2 , a decrease in impact strength was evaluated for C-N-D composites.When DOPO was replaced by Ni(OH) 2 , the interlaminar shear strength decreased from 1093.34 KJ m −2 to 1014.26KJ m −2 for C-N0-D5 and C-N5-D0 composites, respectively.This was also confirmed in a preliminary study [44], where the interface of the skin layer could be disrupted with the addition of flame retardant particles, resulting in poor surface structure stability with lower surface mechanical protection.However, due to better compatibility with epoxy for DOPO, the substitution of DOPO by Ni(OH) 2 could result in a disbanding failure of the fiber-matrix interface [45].
As seen in Figure 8, the composites in this study were all brittle fracture, and the fracture surface was observed to be relatively flat, which was a typical fracture mode of resin matrix composites.However, it was also found that the glass fiber structure of the skin layer has slipping and pulling out behavior during fiber fracture due to the addition of flame-retardant agents.
strength decreased from 1093.34 KJ m −2 to 1014.26KJ m −2 for C-N0-D5 and C-N5-D0 composites, respectively.This was also confirmed in a preliminary study [44], where the interface of the skin layer could be disrupted with the addition of flame retardant particles, resulting in poor surface structure stability with lower surface mechanical protection.However, due to better compatibility with epoxy for DOPO, the substitution of DOPO by Ni(OH)2 could result in a disbanding failure of the fiber-matrix interface [45].
As seen in Figure 8, the composites in this study were all brittle fracture, and the fracture surface was observed to be relatively flat, which was a typical fracture mode of resin matrix composites.However, it was also found that the glass fiber structure of the skin layer has slipping and pulling out behavior during fiber fracture due to the addition of flame-retardant agents.To address the impact of flame-retardant agent additives on the structural performance of composites, several classic studies can be referenced.For instance, Khalili et al. [46] developed a type of fire-retardant fiber mat for epoxy composites, which demonstrated excellent flame retardancy for surface protection.However, they also reported a reduction in mechanical performance.Cong et al. [11] also employed expandable graphite for composite surface protection, but the smoke emission due to partial combustion of the epoxy matrix was an impressive issue limiting composite application.In this study, the design of the sandwich structure using a flame-retardant glass fiber veil was confirmed as an impressive solution to balance the flame retardancy, mechanical performance, and smoke suppression.When compared to the control, the C-N3-D2 composites exhibited impressive flame retardancy (LOI increased by 42.9%, PHRR decreased by 31.3%, and TSP decreased by 19.5%), whilst the maximum reduction of THR was observed for the composite C-N0-D5 by 24.16%.

Conclusions
In this study, the organic flame retardant DOPO and the inorganic flame-retardant Ni(OH)2 were used as synergistic flame retardants, which were combined with glass fibers to construct flame-retardant glass fiber veils as the composite skin structure to protect carbon fiber-reinforced epoxy composites.
It was evident from flame retardancy analysis that DOPO with phosphorous could promote formation of a dense char layer to prevent heat transfer and gas exchange between pyrolysis and oxygen, whilst Ni(OH)2 could be found to capture carbon particles of smoke generated from partial combustion, thus suppressing smoke emission.Through comparison amongst composites in this study, the C-N3-D2 composite exhibited the highest LOI value of 32.3%, and lowest value for PHRR of 435.55 KW/m 2 and TSP of 10.09 m 2 , whilst the C-N0-D5 composite presented the lowest value of THR as 75.24MJ/m 2 .In addition, except for partial effects on the flexural strength and modulus, the interlaminar To address the impact of flame-retardant agent additives on the structural performance of composites, several classic studies can be referenced.For instance, Khalili et al. [46] developed a type of fire-retardant fiber mat for epoxy composites, which demonstrated excellent flame retardancy for surface protection.However, they also reported a reduction in mechanical performance.Cong et al. [11] also employed expandable graphite for composite surface protection, but the smoke emission due to partial combustion of the epoxy matrix was an impressive issue limiting composite application.In this study, the design of the sandwich structure using a flame-retardant glass fiber veil was confirmed as an impressive solution to balance the flame retardancy, mechanical performance, and smoke suppression.When compared to the control, the C-N3-D2 composites exhibited impressive flame retardancy (LOI increased by 42.9%, PHRR decreased by 31.3%, and TSP decreased by 19.5%), whilst the maximum reduction of THR was observed for the composite C-N0-D5 by 24.16%.

Conclusions
In this study, the organic flame retardant DOPO and the inorganic flame-retardant Ni(OH) 2 were used as synergistic flame retardants, which were combined with glass fibers to construct flame-retardant glass fiber veils as the composite skin structure to protect carbon fiber-reinforced epoxy composites.
It was evident from flame retardancy analysis that DOPO with phosphorous could promote formation of a dense char layer to prevent heat transfer and gas exchange between pyrolysis and oxygen, whilst Ni(OH) 2 could be found to capture carbon particles of smoke generated from partial combustion, thus suppressing smoke emission.Through comparison amongst composites in this study, the C-N3-D2 composite exhibited the highest LOI value of 32.3%, and lowest value for PHRR of 435.55 KW/m 2 and TSP of 10.09 m 2 , whilst the C-N0-D5 composite presented the lowest value of THR as 75.24MJ/m 2 .In addition, except for partial effects on the flexural strength and modulus, the interlaminar shear strength and impact strength of the flame-retardant composite were observed to remain consistent.
Thus, the sandwich structure design incorporating fire-retardant glass fiber veils effectively balances flame retardancy, smoke suppression, and mechanical performance in composites.This approach shows significant potential for transportation fields such as aircraft and aerospace.Future studies should aim to optimize the fabrication process for large-scale production and explore the application of this technology in other highperformance domains, potentially widening the impact of this approach.The interaction

Figure 1 .
Figure 1.Manufacturing process of flame-retardant GF veil and composites.

Figure 1 .
Figure 1.Manufacturing process of flame-retardant GF veil and composites.

Figure 2 .
Figure 2. The (a) epoxy resin contact angle and (b) FTIR analysis for retardant GF veils.

Figure 2 .
Figure 2. The (a) epoxy resin contact angle and (b) FTIR analysis for retardant GF veils.

Figure 3 .
Figure 3.The (a) thermogravimetric analysis and (b) limited oxygen index of composites with flame retardant GF veils.

Figure 3 .
Figure 3.The (a) thermogravimetric analysis and (b) limited oxygen index of composites with flame retardant GF veils.

Figure 4 .
Figure 4.The cone calorimetry test of (a) heat release rate (HRR), (b) total heat release (THR) and (c) total smoke production (TSP) for flame-retardant composites.

Figure 4 .
Figure 4.The cone calorimetry test of (a) heat release rate (HRR), (b) total heat release (THR) and (c) total smoke production (TSP) for flame-retardant composites.

Figure 5 .
Figure 5. Morphology of composites after cone calorimetry test and SEM morphology of surface coke layer after combustion.

Figure 6 .
Figure 6.The flame-retardant mechanism of composites in this study.

Figure 6 .
Figure 6.The flame-retardant mechanism of composites in this study.

Figure 7 .
Figure 7.The mechanical properties analysis of (a) flexural properties, (b) interlaminar shear strength and impact strength for flame retardant composites.

Figure 7 .
Figure 7.The mechanical properties analysis of (a) flexural properties, (b) interlaminar shear strength and impact strength for flame retardant composites.

Figure 8 .
Figure 8. Scanning electron microscope image of fracture surface for composites.

Figure 8 .
Figure 8. Scanning electron microscope image of fracture surface for composites.

Table 1 .
Detailed formulation data of flame-retardant composite materials.

Table 2 .
Vertical combustion test and cone calorimeter test results of composites.

Table 2 .
Vertical combustion test and cone calorimeter test results of composites.

Table 3 .
The mechanical properties of composites.

Table 3 .
The mechanical properties of composites.