Ethylene-butyl acrylate copolymer (EBA, 3427AC) powder was supplied by Dupont Co., Ltd. (America). Ammonium polyphosphate (APP) was purchased from Zhenjiang Xingxing Flame-retardant Co., Ltd. (Zhenjiang, China). Silicone-containing macromolecular charring agent (Si-MCA) was synthesized in our laboratory according to our previous literature [32]. The chemical structures of EBA and Si-MCA are shown in Fig. 1. Maleic anhydride-grafted polyethylene (PE-g-MA) was obtained from Zhongjie Chemical Co., Ltd. (Guangzhou, China), and the grafting degree of MA was about 0.9 wt%.

EBA composites were prepared via melt compounding in a Haake batch intensive mixer (Rheomix 600, Karlsruhe, Germany) at 160°C with the rotor speed of 80 rpm for 8 min. PE-g-MA was used as a compatibilizer at a fixed content of 10 wt%, while the IFRs content was fixed at 25 wt% in all EBA composites. For convenience, the samples were designated as EBA/APP (with only 25 wt% APP) and EBA/APP-Si-xy, where xy represents the mass ratio of APP/Si-MCA. For example, EBA/APP-Si-31 indicates APP/Si-MCA = 3/1 in EBA composites. Notably, EBA/APP-Si-31-0 was also prepared under the same conditions; it contained APP/Si-MCA = 3/1 but without PE-g-MA.

The limited oxygen index (LOI) was measured on a JF-3 oxygen index meter (Jiangning, China) with sheet dimensions of 130 × 6.5 × 3.2 mm³, in accordance with ISO4589-1984. The vertical burning test was carried out according to the UL-94 (ANSI/ASTMD635-77) with sheet dimensions of 125 × 12.7 × 3.2 mm³. Thermogravimetric analysis (TGA) was conducted by a Q600 thermal analyzer (TA Co., New Castle, DE, USA) at a heating rate of 10°C min⁻¹ in air from 25°C to 650°C. Cone calorimeter testing (iCone, FTT, West Sussex, UK) was conducted according to ISO 5660-1. The sample dimension was 100 × 100 × 6 mm³; it was backed by aluminum foil and irradiated at a heat flux of 50 kW m⁻². The photographs of residual chars after cone calorimeter testing were taken by a digital camera. And the surface morphology was further observed by scanning electron microscopy (SEM, JSM-6700F, JEOL, Tokyo, Japan). The surface element analysis was characterized by X-ray photoelectron spectroscopy (XPS, K-Alpha, Thermo Fisher, Waltham, MA, USA) and Fourier Transform Infrared spectra (FTIR 460 PLUS spectrometer, Tokyo, Japan). The TGA-FTIR instrument consists of a thermogravimeter (TG209, Netzsch Instruments Co., Selb, Germany), a Fourier transform infrared spectrometer (Tensor 27, Bruker Optics Inc., Ettlingen, Germany), and a transfer tube with an inner diameter of 1 mm connecting the TGA and the infrared cell. The investigation was carried out from 30°C to 750°C at a heating rate of 20°C/min under a nitrogen flow of 100 mL/min. Uniaxial tensile tests were performed at room temperature with an Instron 1121 testing machine (Canton, MA, USA). Specimens were compression molded into sheets with 1 mm thickness, then cut into a dumbbell shape with gauge dimensions of 20 mm × 4 mm × 1 mm. The measurements were conducted at a crosshead speed of 50 mm/min. At least five runs for each sample were measured, and the averaged data were recorded as results.

Firstly, the flame retardancy of neat EBA and EBA/IFR composites was studied by LOI and UL-94 tests. These two methods are the most commonly employed techniques for examining the combustion behavior and fire-resistance capabilities [33–36]. As listed in Table 1, the LOI value of neat EBA was only 17.5%, while that of EBA/APP could reach 21.4%. Meanwhile, it was apparent that the combinations of APP/Si-MCA exhibited higher LOI values. With the decreasing mass ratios of APP/Si-MCA from 5/1 to 3/1, the LOI gradually increased from 24.5% to 31.2%. However, when the APP/Si-MCA ratio further decreased to 2/1, the LOI decreased to 30.6%. Moreover, although both APP and APP/Si-MCA had a positive effect on inhibiting the melt dripping of EBA, only EBA/APP-Si-31 and EBA/APP-Si-21could reach the UL-94 V0 rating. Notably, the extinction time (t₁ + t₂) of EBA/APP-Si-31 was shorter than that of EBA/APP-Si-21, indicating better flame retardancy. Thus, the optimal sample was EBA/APP-Si-31, which displayed the highest LOI value and the UL-94 V0 rating with the shortest extinction time among EBA samples.

Subsequently, thermogravimetric analysis (TGA) was used to investigate the thermal stability of EBA samples, which is able to indirectly evaluate the flame retardancy of polymer materials. As shown in Fig. 2a, the thermal decomposition temperature of EBA/APP shifted to a lower temperature range, which can be attributed to the low-temperature thermal decomposition characteristics of APP. In contrast, EBA/APP-Si-31 shifted to a higher temperature compared with EBA/APP, suggesting the combination of APP/Si-MCA enhanced the thermal stability of EBA. Meanwhile, Table 2 exhibits similar trends for T_5wt% and T_10wt%, which are the characteristic temperatures corresponding to 5 and 10 wt% mass loss, respectively. Furthermore, derivative thermogravimetry (DTG) curves in Fig. 2b also confirmed the trend of a decrease followed by an increase for T_max (corresponding to the maximum mass loss rate). Notably, the degradation rate at T_max was different. EBA/APP-Si-31 displayed the lowest degradation rate, thereby indicating the best thermal stability. Additionally, the higher amount of char residue from EBA/APP-Si-31 implied excellent carbonization ability of APP/Si-MCA, which is conducive to the formation of a high-quality char layer with better flame retardancy [28].

The flame retardancy of EBA samples was further investigated by cone calorimeter testing (CCT), which is helpful for providing important combustion parameters related to fire risk [37–39], such as time to ignition (TTI), heat and smoke release, CO production, mass loss, etc. Fig. 3a shows heat release rate (HRR)—combustion time curves for neat EBA, EBA/APP, and EBA/APP-Si-31, while their detailed fire parameters are summarized in Table 3. It was found that the TTI values of EBA/APP and EBA/APP-Si-31 became shorter, implying that they were easier to ignite than neat EBA. However, a significant decrease in the peaks of HRR (PHRRs) was observed. The PHRR values from EBA/APP and EBA/APP-Si-31 were 604 and 305 kW/m², respectively, which were reduced by 53.3% and 76.4% in comparison with neat EBA (1293 kW/m²). Furthermore, Fig. 3b illustrates their curves of total heat release (THR). Significantly, the slope of the THR curve became flatter, and the final THR value became smaller. Especially, the THR of EBA/APP-Si-31 decreased by 40.2% compared to neat EBA. These results suggested that APP-Si-31 was effective in reducing heat release and suppressing the fire spread in EBA composites.

Besides heat release, the smoke and CO releases are also important to evaluate fire hazard. It is reported that the acute toxicity of fire gases is mainly responsible for over 70% of people killed by fires [40–42]. As shown in Fig. 3c, the peak of the smoke production rate (PSPR) of EBA/APP-Si-31 was lower than that of neat EBA and EBA/APP. Meanwhile, the total smoke production (TSP) of EBA/APP-Si-31 was also lower than that of the others (Fig. 3d). Moreover, according to CO production curves in Fig. 3e, EBA/APP-Si-31 exhibited a dramatic reduction in the peak of CO production (PCOP). Correspondingly, here the less smoke and CO releases were favorable for decreasing the toxicity of fire gases and providing more escaping time.

Finally, Fig. 3f shows normalized mass loss curves of EBA samples with combustion time. Compared to neat EBA, EBA/APP and EBA/APP-Si-31 displayed relatively flatter curves with smaller slopes, implying their weaker fire intensity. Meanwhile, the percentage values of residual char were different (Table 3). As a non-charring polymer, the residual char of neat EBA was nearly zero. The char residue of EBA/APP was 13.6 wt% due to the solid product derived from APP. Interestingly, the residual char of EBA/APP-Si-31 was as high as 22.1 wt%, which was ascribed to the crosslinked carbonization of APP and Si-MCA to form an IFR char layer. As is known, once more char is formed in the condensed-phase region, the amount of flammable fuels entering the gas-phase region will be greatly reduced, resulting in decreased thermal and smoke releases during combustion.

To investigate the flame-retardant mechanism, TGA-FTIR was employed to analyze the gas-phase products of neat EBA and EBA/APP-Si-31. As shown in Fig. 4a,b, it was found that most gas products were similar, but EBA/APP-Si-31 had weaker absorption intensities. Meanwhile, the FTIR spectra of their gaseous products at different temperatures were comparatively studied. When the temperature exceeded 400°C, the characteristic absorption peaks of H₂O (3810–3640 cm⁻¹), C-H (2970–2940 cm⁻¹), CO₂ (2370–2340 cm⁻¹), C=O (1749 cm⁻¹) and C-H (1510 cm⁻¹) were detected (Fig. 4c). These peaks should be assigned to the thermal decomposition products of EBA. In the case of EBA/APP-Si-31, these characteristic absorption peaks of EBA appeared with weaker intensities, and a new peak for P-O at 958 cm⁻¹ was observed (Fig. 4d). The peak intensities for hydrocarbons, carbonyl compounds and C-H were greatly decreased, indicating most carbon elements were fixed in solid chars. Specifically, the almost disappeared peak for CO and the newly appeared P=O peak confirmed the reduced toxicity and the flame retardant effect from the P element. In addition, the FTIR spectra of gaseous products at the maximum degradation stage were analyzed. As shown in Fig. 4e, weaker peaks for these molecular fuels and the characteristic peaks for P=O (1239 cm⁻¹), P-C (1057 cm⁻¹), and P-O-C (925 cm⁻¹) were visible [43,44]. Here, the presence of P-containing groups was attributed to PO•/HPO• radicals released from APP during combustion, which could effectively quench reactive H• and HO• radicals, thereby inhibiting combustion reactions.

The flame-retardant mechanism in the condensed phase was further investigated by analyzing the chars after cone calorimeter test, which were observed using a digital camera and SEM. In the case of EBA/APP, it exhibited a thin char layer with broken surfaces (Fig. 5a). In contrast, EBA/APP-Si-31 presented a dense char layer with interconnected surfaces (Fig. 5b). According to SEM images (Fig. 5c,d), the char layer of EBA/APP was composed of big holes, while the char of EBA/APP-Si-31 was a complete block without holes and cracks. These results indicated the char layer of EBA/APP-Si-31 had a better barrier effect in a fire.

XPS was used to evaluate the elemental composition of residual chars. As shown in Fig. 6a, EBA/APP presented the characteristic peaks of P2s and P2p, which result from the thermal decomposition of APP. Besides P2s and P2p, the new peak of Si2p was detected in EBA/APP-Si-31, indicating that the Si-MCA participated in the construction of the char layer. In addition, the char residues were analyzed by FTIR (Fig. 6b). The typical bands from APP derivates were visible, including NH₄⁺, P=O, P-O and P-C [28]. In the case of EBA/APP-Si-31, strong bands for C=N (1641 cm⁻¹) and Si-O (1022, 766, 478 cm⁻¹) were observed, which were attributed to the decomposition of Si-MCA [32].

On the basis of the above analysis, Fig. 7a shows the chemical reaction between APP and Si-MCA. The decomposition of APP could release phosphoric acid [45], while Si-MCA could produce Si–OH. Their dehydration reactions (such as esterification) promoted the formation of cross-linking chains and further constructed intumescent char (C=N and P–O–Si). Meanwhile, lots of incombustible gases (NH₃, CO₂ and H₂O) were released. Moreover, a possible flame-retardant mechanism in EBA system was proposed. As illustrated in Fig. 7b, the combination of APP and Si-MCA as IFRs played important roles in gas phase and condensed phase during combustion. In gas zone, the thermal decomposition of APP produced PO•/HPO• radicals to quench H•/HO• radicals [28], and these non-flammable gases were helpful for diluting flammable gases and oxygen concentrations. In the condensed phase zone, the derivates from APP and Si-MCA constructed a continuous and compact char layer, which displayed excellent barrier effect to reduce the infiltration of heat and O₂ from outside as well as flammable gases from inside. Thus, the excellent flame retardancy in EBA composites with the LOI of 31.2%, the V0 rating in UL-94 and 76.4% reduction in PHRR was achieved.

As a useful polymer composite material for practical applications, it is highly desirable to achieve a balance between flame retardancy and mechanical performance [46–49]. The mechanical properties of EBA composites containing APP or APP/Si-MCA were evaluated by tensile testing. Fig. 8a shows the stress-strain curves of neat EBA and EBA composites, while the normalized values for Young’s modulus, tensile strength and elongation at break are summarized in Fig. 8b. With the addition of 25 wt% APP, the Young’s modulus of EBA/APP increased due to the reinforcing effect of rigid inorganic particles. However, the tensile strength and elongation at break significantly decreased in comparison with those of neat EBA. As for EBA/APP-Si-31 with the same total IFR content, it exhibited better mechanical properties than EBA/APP. The Young’s modulus, tensile strength, and elongation at break increased by 6.2%, 7.6% and 12.3%, respectively. These results indicated the synergistic effect of APP/Si-MCA on improving the mechanical performance of EBA composites.

To investigate the effect of PE-g-MA on mechanical properties of the EBA system, EBA/APP/Si-MCA composites without PE-g-MA (denoted as EBA/APP-Si-31-0 in experimental section) were prepared under the same preparation conditions as EBA/APP-Si-31. It was found that all mechanical parameters of EBA/APP-Si-31-0 were lower than those of EBA/APP-Si-31 (Fig. 8a,b), including the Young’s modulus (377.3 vs. 404.1 MPa), tensile strength (3.81 vs. 4.65 MPa), and elongation at break (249.4% vs. 344.9%). These results confirmed that PE-g-MA acted as an effective compatibilizer in EBA/IFR composites, which could improve the dispersion of fillers and the matrix-filler interaction [50,51]. As a result, EBA/APP-Si-31 presented relatively good mechanical properties with higher stiffness and toughness than EBA/APP-Si-31-0.