Synergistic effect of chitosan derivative and DOPO for simultaneous improvement of flame retardancy and mechanical property of epoxy resin

In this work, the effects of a chitosan-based derivative (CSA), DOPO (9, 10-dihydro-9-oxa-10- phosphaphenanthreene-10-oxide) and CSA-DOPO additives on the flammable properties of EP (epoxy resin) composites were systematically studied, where CSA was synthesized by a facile condensation between chitosan (CS) and 9-anthralaldehyde. The mass ratio of CS and 9-anthralaldehyde in CSA was determined by elemental analysis and theoretical calculation. Under the 8% addition in EP, EP/2.66%/5.34%DOPO sample was the only one passing the UL-94 V-0 rating and exhibiting the highest LOI value of 36.4%. The cone calorimeter test (CC) showed that the total smoke emission value and the peak heat release rate of the EP/2.66%/5.34%DOPO decreased by 36.0% and 61.9%, and the residual char amount increased by 151%, respectively, when compared with EP. Moreover, the incorporation of CSA/DOPO effectively improved the flexural strength by 52.3%. According to the results obtained from Py-GC/MS analyses for EP and EP/2.66%CSA/5.34%DOPO samples, together with Raman spectra, XPS (X-ray photoelectron spectra) for their char residues, and the real time FTIR (Fourier-transform infrared) spectra at different pyrolysis temperatures and cone calorimeters, it was proposed that CSA/DOPO played roles in both gaseous and condensed phases, and the synergistic effect of CSA and DOPO significantly improved the flame retardancy and mechanical strength of EP.


Introduction
Epoxy resin has been frequently used in many fields owing to its superior chemical resistance, excellent adhesion and good mechanical properties (Zhu et al. 2020;Yu et al. 2016). However, the widespread use of epoxy resin is severely limited due to its own flammability (Jin et al. 2019). Adding flame retardants to EP is considered the most direct, efficient and economical method for increasing the fire resistance of epoxy resin materials .
Due to the flame retardancy function of DOPO in both gaseous and condensed phases, extensive work had been performed to synthesize DOPO-based derivatives for the purpose of upgrading the flame retardancy of EP (Levchik and Weil 2004). The preparation of DOPO-based derivatives was usually based on the chemical reactions between the P-H bond in DOPO and the active groups in other substances. Some of the derivatives exhibited excellent flame retardancy toward EP. For example, Carja and coworkers (Carja et al. 2014) synthesized the oligomer DOPO-PFR using DOPO, rterephthalaldehyde and phenylphosphonyl dichloride as raw materials. When DOPO-PFR with the phosphorus contents of 2% and 3% were added to EP system, the obtained epoxy matrix can reach UL-94 V-0 rating. Wang et al. (Wang et al. 2010(Wang et al. , 2011a used phosphorus oxychloride, pentaerythritol and DOPO-based dihydric phenolic compounds as raw materials to synthesize a series of DOPO-based oligomers. When these oligomers were added in EP using 4, 4'-diaminodiphenylmethane as a curing agent, the amount of combustible gas, the total heat release and the maximum heat release rate in the combustion process were significantly reduced, while the LOI value was greatly increased. Yang et al. 2015 used N-(4hydroxyphenyl) maleimide and cyanuric chloride to synthesize a maleimide functional group-containing triazine derivative TMT, which was then reacted with DOPO to synthesize the flame retardant DOPO-TMT. For a phosphorus content of 1.0 wt%, the LOI value of the modified EP reached 36.2% and passed the UL-94 V-0 test. Although some DOPO-based derivativemodified EP composites exhibited good flame retardancy, the complicated preparation process and the drawbacks of their mechanical properties limited their industrial applications. Easy preparation with lower cost and desirable mechanical properties is still the essential conditions for large scale production of EP. Aurelio et al. (Aurelio et al. 2020) modified epoxy resin with the combined action of a DOPO-derivative, melamine and silica and the obtained epoxy resin composite achieved UL 94-V-0 classification at low P-content (2.0 wt%). The exploration of EP composites with better flame retardancy, which can be achieved by the combined action of several substances, can lead to wider industrial applications.
As a natural biomass polymer material (Dash et al. 2011;Yang et al. 2018), chitosan (CS) can be used as a charring agent due to the abundant hydroxyl groups in its structure (Chen et al. 2016). However, it has the disadvantages of low initial decomposition temperature, poor thermal stability and low flame retardant efficiency (Prabhakar and Song 2020;Hu et al. 2012). The amine groups and hydroxyl groups possessed by chitosan can be used as active groups for chemical modification (Hirano et al. 2003;Huang et al. 2011). The salicylaldehyde modified chitosan combined with polyphosphate with a total content of 25% in polyurethane exhibited better flame retardancy (Liu et al. 2017). A chitosan derivative modified by gradually grafting -PO(OH) 2 , melamine and organic montmorillonite was used as an intumescent flame retardant (IFR) for EP ). There were some examples of chitosan being utilized as a flame retardant component of epoxy resins. It has been reported (Rao et al. 2019) that the introduction of ZrO 2 NPs, chitosan, boric acid, ammonium polyphosphate and melamine into EP could significantly increase the LOI value and reach a UL-94 V-0 rating. However, flame retardants derived from CS had rarely been reported due to their lower reaction activity. In our previous studies, a flame retardant (CCD) was synthesized by a two-step chemical reaction using chitosan, cinnamaldehyde and DOPO as raw materials, where CCD exhibited improved flame retardancy (Chen et al. 2020b). We have also tried to synthesize other DOPO modified chitosan. However, it was challenging to obtain the designed products. It was deduced that the successful synthesis of CCD can be ascribed to the additional -C = C-double bonds in its precursor, which can inset into the reaction media and react with DOPO more easily. In the synthesis of other DOPO modified chitosans, the active groups in their precursors are -C = N-bonds, which are closely attached with CS units. The space steric hindrance and insolubility of the precursors makes it difficult for them to further react with DOPO. This may be the reason why no other reports of DOPO-chitosanmodified chitosan are found in the literature.
Based on the charring function of chitosan derivative and the flame retardancy of DOPO, a Schiff base compound CSA was prepared by the condensation between the active amine groups of chitosan and aldehyde of 9-anthralaldehyde. The effects of different CSA and DOPO dosages in EP on the combustion and mechanical properties of EP composites were studied, and the possible flame retardant mechanism of CSA/ DOPO was proposed based on the results of TG-FTIR, cone calorimeter, Raman spectroscopy, XPS and Py-GC/MS analysis. The superior properties of EP/ 2.66%CSA/5.34%DOPO in terms of both flame retardancy and mechanical properties had not been previously observed in only chitosan-containing flame retardants for EP. Moreover, the prepared EP composites had the advantages of facile preparation, lower cost and improved flame retardancy and mechanical properties.

Synthesis of CSA
The CSA was synthesized according to the referenced method with some modifications (Kumar et al. 2009) as shown in Fig. 1a. First, chitosan (8.06 g, 0.05 mol) was dispersed in 150 ml methanol in a three-necked flask equipped with a condenser. A mixture of 9-anthralaldehyde (20.73 g, 0.1 mol) and 150 ml of tetrahydrofuran was added dropwise through a constant pressure funnel within 60 min. The resulting mixture was heated to 75°C and refluxed for 12 h. Then it was cooled and filtered. The obtained filter cake was washed three times with a mixture of methanol and tetrahydrofuran (volume ratio 1:1). Finally, the product was vacuum dried at 60°C for 6 h.

Preparation of EP mixtures
The preparation of CSA/DOPO flame-retardant epoxy resin material was conducted as following ( Fig. 1b): Firstly, DGEBA and CSA (or CS) were mixed uniformly in a two-necked flask. After being heated under vacuum at 130°C for approximately 40 min, a translucent liquid was obtained. Then, DOPO was added in a certain mass ratio, and stirred to be homogeneous. DDM was added under vigorous stirring and vacuum conditions after the temperature dropping to 95°C. Finally, the mixture was poured into a clean mold, cured for 2 h at 100°C and another 2 h at 150°C. The compositions of the modified EP samples were listed in Table 1.

Methods
FTIR spectra of CS, CSA and the char residue for EP/ 2.66%CSA/5.34%DOPO were recorded on a Thermo Nicolet 5700 FTIR spectrophotometer using the KBr disk in a wavenumber range of 4000-400 cm -1 .
The contents of carbon, hydrogen and nitrogen in CS and CSA were analyzed by a Vario MACRO cube elemental analyzer and phosphorus content in the char residue of EP/2.66%CSA/5.34%DOPO was detected additionally.
The surface morphologies studies of the char residual were performed on JEOL JSM-5900LV under vacuum at an acceleration voltage of 20 kV.
The thermal decomposition behaviors of CS, CSA and epoxy resin thermosetting polymer samples were recorded by a thermogravimetric analyzer (NETZSCH STA449F3). The test was conducted by heating from 30 to 790°C under nitrogen. The heating rate was 10°C/min.
The LOI value was obtained on a JF-3 oxygen index instrument, the sample size was 130 9 6.5 9 3.2 mm 3 . The UL-94 rating was measured with the sample size 130 9 13 9 3.2 mm 3 . The cone calorimeter (CC) test was carried out according to standard ISO 5660 with the sample size of 100 9 100 9 3.2 mm 3 under an external heat flux of 50 kW/m 2 .
X-ray photoelectron spectra of the char residues left behind after cone calorimeter tests were recorded on an ESCALAB 250xi (America) using AlKa radiation (1486.6 eV) as an excitation source.
Raman spectroscopy was recorded on Lab-RAMHR800 laser Raman spectrometer to characterize the types of the carbon under 532 nm helium-neon laser line at room temperature.
Pyrolysis products of EP composites were analyzed by Py-GC/MS in helium atmosphere. Firstly, the sample was pyrolyzed in a cracker (CDS 5200) at 600°C, and then the volatile products were transported to the Perkin-Elmer Clarus 680 GCSQ8MS gas chromatography-mass spectrometer via helium.
The flexural and tensile tests of the samples were recorded on CMT4104 universal testing machine (SANS) with a dumbbell shape of 75 9 4 9 2 mm 3 and a rectangle of 80 9 10 9 4 mm 3 , separately, according to the standard of GB/T 9341-2008 and GB/ T 1040.2-2006. Five samples were measured each time, and the average value was taken.
TG-FTIR was used to analyze the FTIR spectra of the gas components from the TGA, and the real time FTIR analysis of the pyrolysis gases was conducted on a Thermo Fisher IS50, the transfer line of gases from TGA to FTIR was heated at 280°C.

Results and discussion
Characterization of CSA Figure 2 shows the FTIR spectra of chitosan and CSA. For chitosan, there was a wide frequency band in the range of 3400-3800 cm -1 , which was due to the stretching vibration of the O-H bond and the N-H bond, while the peak centered at 2907 cm -1 was assigned to the C-H bond. There were two obvious characteristic peaks at 1157 cm -1 and 1602 cm -1 , which were attributed to the tensile vibration of the sugar structure and amino groups, respectively. For CSA, a new stretching vibration absorption peak at 1645 cm -1 was detected, which was attributed to -C = N (Dubey et al. 2018), indicating the existence of Schiff base in CSA. It was noted that there was no aldehyde characteristic peak appeared at approximately 1680 cm -1 , revealing that CSA did not contain unreacted 9-anthralaldehyde. The chemical composition of CSA was further characterized by elemental analysis. According to the C, N and H atomic contents of CS and CSA given in Table S1, the C contents in CS and CSA were 44.58% and 68.04%, respectively. Compared with CS, H and N contents in CSA decreased by 1.13% and 3.55%, respectively. The reason for the increase of C content and the reduction of H and N contents can be attributed to the introduction of nitrogen-free anthracene unit with low H content and high C content in CSA.
Chitosan contains amino groups which can form Schiff base compounds with aldehyde substances under certain conditions. The reaction was generally carried out in the mixture of methanol and tetrahydrofuran, and the generated products were found to be insoluble in ethanol. At the end of the reaction, the products were obtained by washing with the mixed solvents for several times. With chitosan as a polymer, it can be treated as a mixture with different degree of polymerization. When it reacted with the aldehyde material, there might be either multiple amino groups or none from the chitosan molecules to be involved in the reaction. Considering that chitosan and the final product were not soluble in ethanol, the unreacted chitosan may mix with the final product. In addition, CS is also a polymer, the reacted -NH 2 groups in each CS may be different. To confirm the composition of the product, the following calculations were used to calculate the mass ratio of chitosan and the proportion of aldehydes involved in the reaction.
In the Schiff base condensation reaction of chitosan with a substance containing aldehyde group, the chemical reaction of an amino group with an aldehyde group can be expressed as following: The masses of 9-anthralaldehyde reacting with chitosan and total chitosan added in the reaction system were set as m 1 and m 2 , respectively. The molar mass of 9-anthralaldehyde was M 2 , then the mole number of R 2 -CHO was m 2 /M 2 . The product mass (m) is equal to the total mass of the reactants deducting the mass of the produced water molecules, where water molecules are the only substance, which can be washed away by solvents. The product mass can be theoretically calculated according to the Eq. (1): where m 1 and m 2 can be obtained from the following equations: where C% and H% represent the percentage of C and H in the product, respectively, which can be obtained directly by measuring the element content of the product; C 1 %, H 1 % and N 1 % were the percentages of C, H and N in chitosan, which can be obtained directly by measuring the element content of chitosan. Among them, C 2 % and H 2 % were the percentage content of C and H in 9-anthralaldehyde molecule, which can be directly calculated by the molecular formula of the aldehyde. By solving Eqs.
(1-4), the mass ratio of chitosan (m 1 ) and the mass of 9-anthralaldehyde (m 2 ) involved in the reaction can be quantitatively determined. This value was calculated to be 1.0536, which was obtained from the following: since the carbon content was the highest in the sample, the relative measurement error of carbon was relatively smaller than that of H and N in the sample. Therefore, the carbon content obtained by the elemental analysis and the theoretically inferred product quality calculation formula were selected to calculate the mass ratio (m 1 /m 2 ). The theoretical values of H and N were calculated by substituting them into the theoretical content calculation formulas of hydrogen and nitrogen, which were compared with the actual measured percentage contents of H and N. The results showed that the calculated theoretical value was consistent with the actual measured value, indicating that the calculated ratio of m 1 to m 2 was correct. Therefore, the mass ratio of chitosan to 9-anthraldehyde involved in the reaction was obtained. The elemental analysis data of chitosan and CSA and the calculated values for CSA were listed in Table S1.

Thermal stability
In order to understand the thermal decomposition behavior of CSA and CSA/DOPO modified epoxy resins, CSA, EP, EP/8%CSA, EP/8%DOPO and EP/ 2.66%CSA/5.34%DOPO were studied by thermogravimetric analyzer. Figure 3 showed the TGA and DTG curves of all the samples, and the corresponding results were listed in Table S2. The highest residual char of 34.6% at 750°C and the lowest weight-loss rate at maximum weight loss temperature (T max ) indicated the excellent char-forming property of CSA (Fig. 3). However, the initial decomposition temperature T 5% and T max% of CSA were the lowest ones among the tested samples, revealing the poorer thermal stability of CSA when compared with other EP composites. It was found that EP/8%CSA exhibited the highest T max and char residue among the EP composites, revealing that the introduction of CSA improved the thermal stability and char-forming property of EP. In addition, the char residue of EP/ 8%DOPO was only 12.6% at 750°C, which was the lowest in the tested samples. So, it can be concluded that the introduction of DOPO into EP greatly decreased the char-forming properties of EP. For EP/ 2.66%CSA/5.34%DOPO, it was observed that the residue char amount was increased by 28.8% relative to EP/8% DOPO and reduced by 12.7% relative to EP/ 8% CSA, exhibiting a comparable char-forming property with EP.

Flame retardancy performance
The LOI value and UL-94 test were adopted as the frequently-used methods for evaluating the flame retardancy of materials. The test results were summarized in Table 1. The EP sample had a LOI value of 22.6%, which was the lowest among all the samples. Except for EP, all the other samples containing CSA, DOPO or CSA and DOPO showed no dripping in the combustion process. The LOI value of the EP/ 1.66%CSA/3.34%DOPO was 34.5%, which increased by 52.6% compared with the EP sample. When the total content of CSA/DOPO was increased to 8wt% with the mass ratio of CSA to DOPO as 1:2, the LOI value of EP/2.66%CSA/5.34%DOPO samples was 36.4% enhancement of 61.1%, 31.9% and 6.4% relative to the EP, EP/8%CSA and EP/8%DOPO samples, respectively. The results showed that the synthesized flame retardant synergist CSA had a good synergistic effect with DOPO on the flame retardant properties of EP composites. In the UL-94 test, the EP samples were completely burned out and no ratings were obtained. Furthermore, EP/2.66%CSA/ 5.34%DOPO and EP/2.01%CSA/5.99%DOPO passed the V-0 rating, while EP/8%DOPO only reached V-1 rating and EP/8%CSA exhibited no rating. This result indicated that the improvement in the flame retardant properties of EP was contributed by the synergistic function of CSA and DOPO.
In addition, the combustion behavior of EP materials was tested with a cone calorimeter (CC). Figure 4 showed the heat release rate (HRR), total heat release rate (THR), smoke generation rate (SPR) and total smoke generation (TSP) curves of EP and EP/ 2.66%CSA/5.34%DOPO samples. The CC test results were listed in Table 2. It showed that the TTI value of EP/2.66%CSA/5.34%DOPO samples increased from 59 s in EP to 82 s. The improvement of the ignition resistance of the matrix was attributed to the release of volatile products produced by the early decomposition of CSA/DOPO, which could capture or dilute oxygen. Relative to EP, EP/2.66%CSA/5.34%DOPO had much higher residual char amount with an increment of 151% and lower peak heat release rate (PHRR) with a reduction of 36.0%.
In the TGA test, the residue char of EP/2.66%CSA/ 5.34%DOPO was slightly lower than that of EP in N 2 atmosphere. However, the obvious increase of the residue char amount of EP/2.66%CSA/5.34%DOPO in the CC test can be attributed to the existence of oxygen, which was the only difference between these two tests. It indicated that oxygen could improve the formation of more residue char for EP/2.66%CSA/ 5.34%DOPO, which boosted the fire resistance ability. FIGRA was defined as the maximum value of HRR/t p , which was equal to PHRR/t p in accordance with general practice. FIGRA was an important parameter for evaluating fire risk (Schartel and Hull 2007). It can be seen from Table 2 that the FIGRA value of EP/ 2.66%CSA/5.34%DOPO was about 40% lower than that of EP sample, indicating a significant function in reducing the fire risk. More importantly, it was worth noting that the peak smoke production rate (PSPR) and total smoke release (TSP) values of EP/CSA2.66%/ 5.34%DOPO were much lower than those of the EP sample, illustrating that the combination of CSA2.66% and 5.34%DOPO can significantly improve the smoke suppression effect of EP. Av-EHC represents the average amount of heat energy released per unit mass of the sample when burned, which can reflect the degree of gas phase combustion in the test (Xu et al. 2016). As shown in Table 2, the average value of EHC for EP/2.66%CSA/ 5.34%DOPO thermosets decreased from 29.3 MJ/kg for EP to 21.7 MJ/kg, indicating that EP/2.66%CSA/ 5.34%DOPO had lower degree of gaseous phase combustion. The superior behavior of EP/2.66%CSA/ 5.34%DOPO both in smoke suppression and flame retardancy could be ascribed to the following reasons: 1) the lower thermal stability of CSA promotes the escape of some non-flammable gases such as NH 3 and CO 2 during the early stage, which then diluted combustible gases, 2) P containing radicals derived from the DOPO units in EP/2.66%CSA/5.34%DOPO help to capture the radicals produced from chain reactions (Schartel et al. 2008), 3) the much higher char residue also plays a role in isolation and quenching heat transfer. Analysis of char residue Figure 5 showed the macrostructure and microstructure of the char residues left behind after the CC test. Figure 5a showed that the residual carbon left behind after EP combustion exhibited noncontinuous fiberlike structure with large pores and poor quality. The char residue in this form cannot isolate the exchange of the produced combustion gases or protect the substrate from burning. With increasing CSA and DOPO contents in EP, it was found that the char residue became denser with higher combustion char residue. It indicated that the addition of CSA/DOPO could increase the mass and compactness of residual char, and played a role in condensed phase flame retardant effect.
The residual carbon after CC test was systematically studied by elemental analysis, infrared spectroscopy and Raman spectroscopy. As shown in Fig. 6, two significant peaks at 1594 cm -1 and 1374 cm -1 can be assigned to the G band and the D band, respectively. In general, the G-band is attributed to sp 2hybridized carbon atoms, and the D-band belongs to amorphous carbon (Mauere 2005). Therefore, the graphitization degree of the material can be reflected by the integral intensity ratio (I D /I G ) of the Dand G-bands. With the decreased I D /I G , the graphitization degree increases, leading to better flame retardant performance (Sadezky et al. 2005). Figure 6 shows that the I D /I G value (3.93) of EP was greater than that of EP/CSA2.66%/5.34%DOPO sample (3.52), indicating that the char layer formed by the  Char residue (%) 11.9 29.9 Fig. 5 The digital and SEM images of the char residues from EP (a, a-1,a-2), EP/1.66%CSA/3.35%DOPO (b-1, b-2) and EP/ 2.66%CSA/5. 34%DOPO (c, combustion of EP/CSA2.66%/5.34%DOPO has a higher degree of graphitization than that of the char residue of EP, which is essential to better prevention on heat transfer in the combustion process. Fig. S1 showed the FTIR spectra of the char residues from EP/2.66%CSA/5.34%DOPO and EP. The two curves exhibited similar main absorption peaks at 2921 cm -1 and 1619 cm -1 , corresponding to the vibration of the aromatic ring. The signal peaks at 1080 cm -1 and 3422 cm -1 belonged to C-O and O-H bonds, respectively (Liu et al. 2006). There were three new absorption peaks at 1159 cm -1 , 1236 cm -1 and 1509 cm -1 in the char residues of EP/2.66%CSA/ 5.34%DOPO, which were assigned to the stretching vibrations of P-O-C, P = O and C-P, respectively (Chen et al. 2020b). The results indicated that some phosphorus compounds existed in the char residue of EP/2.66%CSA/5.34%DOPO, which were not found in EP.
The results of elemental analysis of the char residues were listed in Table S3. Compared to EP char residue, the C content in the char residue of EP/ 2.66%CSA/5.34%DOPO increased by 6.25%, and its O and N contents were decreased by 27.9% and 40.3%, respectively. The P content of 0.48% in the char residue of EP/2.66%CSA/5.34%DOPO was 37.7% lower than that in its precursor, which was 0.77% calculated from the addition mass of DOPO in the sample and P percentage in DOPO. The decrease of P content in the char residue illustrated that some P containing substances have escaped into gaseous phase in the combustion, which was proved by the Py-GC/MS analysis of the sample. Generally, the P element in the condensed phase would exist in the form of polyphosphoric acid and its derivatives, promoting the dehydration and carbonization and preventing further combustion.
Additionally, XPS was further employed to analyze the composition of the various components of the char residue. Detailed data of C 1s , N 1s , O 1s , P 2s chemical bonds were listed in Table S4. As shown in Fig. 7, in the C 1s spectrum, the peaks at 284.4 eV were assigned to C-H and C-C bonds in the aliphatic and aromatic components. Similarly, the peaks of C-N bond and C-OH bond were observed at 285.4 eV, and those of C-C and C-O bonds were observed at 288.7 eV. Compared with EP, a new weak peak at 289.9 eV was observed, which was attributed to O = C-O bonds (Rao et al. 2019). It can be seen that the peak areas at 288.4 eV, 285.4 eV and 288.7 eV in the char residue of EP/ 2.66%CS/5.34%DOPO are different with the relative ones in EP, indicating that the introduction of CSA and DOPO in EP changed the compositions in the char residues (Levchik et al. 2005). In the N 1s spectrum, the peaks at 398.6 eV and 400.2 eV were assigned to the C-N and C = N bond, respectively. In the O 1s spectrum of EP, the peak at 533.2 eV was attributed to C-OH and C-O-C groups, while peak at 532.2 eV was assigned to C = O. For the EP/2.66%/ 5.34%DOPO, the peaks at 532.2 and 533.2 eV were also observed, however, peak at 533.2 eV was bigger than that the same peak at EP, which can be attributed to the formation of C-O-P or C-O-C, P-O-P groups in the char residue of EP/2.66%/5.34%DOPO ). In the P 2s spectrum, the peak at 133.4 eV was attributed to P = O, and the peak at 134.2 eV was assigned to P-OH and P-O-C bond, indicating that P containing substances were generated during the formation of residual carbon since there was no P-containing peak shown in the char residue of EP. Fig. 6 Raman spectra of char residues for EP (a) and EP/2.66%CSA/5.34%DOPO (b) Fig. 7 C 1s , N 1s , O 1s and P 2s spectra of char residues for EP and EP/2.66%CSA/ 5.34%DOPO 123 Py-GC/MS analysis of EP and EP/2.66%CSA/ 5.34%DOPO According to reports in the literature, DOPO undergoes thermal decomposition during the combustion process to release free radicals containing P, thereby having a flame retardant effect in the gas phase (Li et al. 2020;. Py-GC/MS was carried out for the EP sample and EP/2.66%CSA/ 5.34%DOPO to dissect the gas flame-retardant mechanism at 500°C. Fig. S2 exhibited the pyrolysis diagram of EP, and the main cracked products were listed in Table S5. As shown in Table S5, the pyrolysis products of EP samples were mainly derivatives of DDM and epoxy resin. Among them, ÁOH radicals and ÁH radicals produced during thermal cracking were the main reasons for the inflammability of thermosetting epoxy resins. The pyrolysis spectrum of EP/2.66%CSA/ 5.34%DOPO was shown in Fig. 8. The identified specific products were listed in Table 3. Some characteristic products of the flame retardant pyrolysis of EP/2.66%CSA/5.34%DOPO were as following: methylaminoacetic acid (peak 1), L-alanylglycine (peak 2), paromomycin (peak 3), 4-hydroxy Phenyl phosphate (peak 4), o-cresol (peak 5), 2-isopropylphenol (peak 6), o-phenylphenol (peak 7), dibenzofuran (peak 8), fluorene (peak 9) and 9-hydroxy-4methoxyacridine (peak 10). The existence of these substances could be explained by the structure of CSA/DOPO flame retardant. The pyrolysis products 4 and 7-9 can be attributed to the decomposition of DOPO, accompanied by the production of PO and PO 2  (Qian et al. 2011). These phosphorus-containing radicals could trap active free radicals during the combustion process and terminate radical chain reactions (Jellinek et al. 1977). Peak 1-3 were mainly corresponding to the cracking of chitosan fragments in CSA. The peaks with retention time more than 22.0 min in the pyrolysis spectrum of EP/ 2.66%CSA/5.34%DOPO were also found in that of EP, indicating that the substances related to these peaks come from similar components with EP. The alkyls groups and amino groups containing compounds can generate non-combustible gases, such as CO 2 , H 2 O and NH 3 , leading to diluted combustion gases and reduced heat release. On the basis of the above analysis, the flame retardant mechanism of EP/2.66%CSA/5.34%DOPO could be summarized as following: with the increase of temperature, the gradual decomposition of DOPO and CSA promoted the release of POÁ radicals and non-combustible gases such as CO 2 , H 2 O and NH 3 , playing the quenching and diluting function in the gaseous phase, respectively. In the condensed phase, the polyhydroxyl and aromatic rings in CSA and the phosphorus-containing compounds generated by DOPO decomposition can accelerate the formation of a more uniform and dense coke layer, so as to trigger the thermal oxygen barrier effect.
Particularly, as shown in Fig. 9 (c, d), the peak at 1130 cm -1 in the FTIR spectra of the gaseous products of EP/2.66%CSA/5.34%DOPO was attributed to the Ph-P bond . What's more, the peak at 1260 cm -1 was assigned to Ph-O-C, which probably overlapped with the characteristic absorbance of ether bonds Wang et al. 2011d;Qian et al. 2011). It could be seen that there were phosphorus-containing fragments in the pyrolysis gas of EP/2.66%CSA/5.34%DOPO samples, indicating that the CSA/DOPO flame retardancy system plays an active role in the gas phase, which was consistent with the conclusion of Py-GCMS.
In order to further understand the changes of pyrolysis products, Fig S3 shows the changes in the absorbance of EP and EP/2.66%CSA/5.34%DOPO pyrolysis products over time (a) and the curve obtained from the first derivative (b). As can be seen from the figure that the pyrolysis products of EP/ 2.66%CSA/5.34%DOPO began to be released at about 30.2 min, while the pyrolysis products of EP began to release at around 31.4 min. Additionally, it could be observed from Fig. 9c, d that the thermal cracking of EP/2.66%CSA/5.34%DOPO promoted CO 2 release and the early decomposition of of EP/2.66%CSA/ 5.34%DOPO was contributed to the introduction of CSA/DOPO in EP.

Mechanical properties
In order to explore the effect of CSA/DOPO flame retardant on the mechanical properties of epoxy resin materials, the measurements of unnotched impact strength, tensile strength and flexural strength were conducted on EP, EP/8%DOPO and EP/2.66%CSA/ 5.34%DOPO. The results were presented in Table 4. It was observed that the mechanical properties increased accordingly from EP to EP/8%DOPO and EP/ 2.66%CSA/5.34%DOPO. A prominent increase of 52.3% was discovered for flexural strength of EP/ 2.66%CSA/5.34%DOPO. The improved mechanical properties of the modified EP can be attributed to the reaction activity of CSA and DOPO with the epoxy groups in EP during the curing procedure. The -OH and -NH 2 in CSA and -P-H groups in DOPO can participate in the ring opening reaction of epoxy groups in EP at high temperature, which were shown in Scheme 1, the formed strong interactions greatly improved the compatibility of CSA and DOPO with EP. Moreover, the hydrogen bond interactions among the polar groups such as -OH and -NH-in the modified EP as well as the p-p stacking among the aromatic rings are also beneficial to increase the mechanical strength of EP.
To the best of our knowledge, EP/2.66%CSA/ 5.34%DOPO was the first report for the CS-containing flame retardants, which simultaneously improve the flame retardancy and mechanical properties of EP. The relative data of flame retardancy and mechanical properties for the CS-containing flame retardants in EP was listed in Table S6, where the relative references (Chen et al. 2020a;Shi et al. 2018;Yu et al. 2020b;Zhang et al. 2019;Zhang et al. 2021), have been cited to support the information Compared with CCD modified EP composite (EP/10%CCD) (Chen et al. 2020b), EP/2.66%CSA/5.34%DOPO exhibits comparable flame retardancy with higher LOI value and enhanced mechanical properties. It was worth mentioning that the preparation of the flame retardant in EP/2.66%CSA/5.34%DOPO was much easier compared to EP10%CCD. CSA was synthesized by Schiff base condensation, which was similar with the first step of the CCD synthesis. However, a further reaction between the Schiff base compound and DOPO was needed to produce CCD, while the flame retardant in EP/2.66%CSA/5.34%DOPO can be achieved by simple mixing of CSA and DOPO.

Conclusions
In this work, a flame retardant synergist (CSA) was successfully synthesized, and the EP/CSA/DOPO thermosetting material obtained by compounding DOPO and CSA with EP demonstrated good flame Fig. 9 The real time FTIR spectra and 3D TG-FTIR images of EP (a, c) and EP/2.66%CSA/5.34%DOPO (b, d) samples at different pyrolysis temperature retardant properties. Based on the UL-94 test, EP/ 2.66%CSA/5.34%DOPO passed a V-0 rating and had the LOI value of 36.4%, with an increase of 61.06% compared with 22.6% for EP. In comparison with the EP, EP/2.66%CSA/5.34%DOPO composites exhibit the following merits, 1) the PHRR, PSPR and TSP values were significantly reduced, 2) the mechanical properties were improved, which were attributed to the reaction activities of the functional groups in CSA and DOPO with the epoxy groups in EP during the curing procedure, and 3) the char residue amount increased by 151%. The analyses of TG-FTIR, CC tests and Py-GC/MS of EP composites further supported the synergistic significance of CSA and DOPO on flame retardant properties. The released POÁradicals together with non-combustible gases such as CO 2 , H 2 O and NH 3 in the gas phase formed a more uniform and dense coke layer which then facilitate the thermal oxygen barrier effect. It can also be concluded that simple mixing of substances with respective functions based on a specific mass ratio can greatly magnify the relative properties of the synthesized composites. EP 54 ± 4 8 6 ± 3 2 1 ± 1 EP/8%CS 52 ± 2 8 9 ± 3 2 1 ± 3 EP/8%DOPO 58 ± 3 9 7 ± 2 2 2 ± 2 EP/2.66%CSA/5.34%DOPO 63 ± 3 131 ± 2 2 8 ± 2 Scheme 1 The possible reactions of CSA, DOPO, DDM and DGEBA in the curing process