Introduction

Epoxy resin is one of the most versatile thermosetting polymeric materials, which has a wide application in various fields owing to its excellent adhesive properties, satisfactory chemical resistance and good thermal and mechanical stability [1,2,3,4,5]. However, with a limiting oxygen index (LOI) of only 19.5%, epoxy resin is highly flammable, and there is almost no residue left after sufficient combustion, which severely restricts its further application for use as reinforcing material in structural engineering of building [6]. Consequently, it is imperative to improve the flame retardance and fire resistance of epoxy resin.

Although a great deal of work [7,8,9,10,11] has been devoted to improving the flame retardancy of epoxy composites through physically adding various flame retardants, it still cannot meet the use requirements in the long-lasting fire or high temperature (over 900 °C) conditions. In order to overcome this problem, strategies to make polymer composite ceramifiable in the fire are reported [12,13,14,15,16,17], and those ceramifiable composites have similar properties of matrix materials at room temperature, but they could be converted to ceramic-like residues when they are fired at high temperatures.

Many studies [18,19,20,21,22] have been implemented to achieve a high fire resistance in a polymer composite, and the main aim is to improve shape stability, self-supporting performance or strength of fired composite residue. Glass frit with a low softening point as flux agent was often added to the ceramifiable polymer composite system. With the increase in temperature, glass frit will melt into liquid phase and then adhere to other components together to form a compact ceramic residue. This is a relatively effective method, but this kind of ceramifiable composite will have serious deformations due to the strong liquidity of the glass frit at high temperatures [18,19,20]. Glass fibers or carbon fibers were added to construct a skeleton as the supporting structure of residues [21, 22]. Another approach [23,24,25,26,27,28] is to increase the crystal content of the residue, and this strategy has been applied in the ceramifiable silicone rubber composites and EVA composites. For example, Shen et al. [23] improved the ceramifiable properties of silicone rubber composite through incorporation of ammonium polyphosphate, aluminum hydroxide, and mica, resulting in crystallization reaction among these additives. Gong et al. [25] prepared ceramifiable EVA composites by adding whitened and capsulized red phosphorus (WCRP) and silicate glass frit. It showed that the addition of WCRP promoted the generation of cristobalite crystal, which led to an improvement in the self-supporting property of the composite at high temperatures. The key to obtain ceramifiable polymer composites seems to need specific additives that could react into certain reinforcing phase at firing temperature, but this is not easy to achieve experimentally for different polymer matrix. To the best of our knowledge, ceramifiable polymer composite is rarely reported for thermosetting system such as epoxy.

On the one hand, ammonium polyphosphate (APP) is an excellent non-halogen flame retardant and widely used in flame-retarding polymer materials [29,30,31]. Moreover, ceramifiable property of polymer composite could also be improved due to the reaction between APP and other filler [17, 23, 24]. In our previous work [25], it was found that the interaction of phosphorous with some silicate compounds at high temperature is vital to improve fire resistance of the polymer composites. Thus, on the other hand, APP was used as a resource of phosphorus to induce the chemical reaction at high temperatures. In this study, a novel epoxy composite with high fire resistance was prepared by adding APP and silicate glass frit (SGF). The flame retardance of epoxy composite was tested by an oxygen index instrument and a cone calorimeter device. The fire resistance of the composite was investigated with the aid of shape stability test and flexural strength test. Furthermore, possible mechanisms for the phase separation and crystallization reactions at high temperatures were also investigated through TG, XRD, XPS, EDX, etc.

Experimental

Materials

The epoxy resin chosen here was liquid diglycidyl ether of bisphenol A (E51) obtained from Wuxi Bluestar Co., Ltd., and the curing agent methyl tetrahydrophthalic anhydride (MTHPA) was supplied by Zhejiang Alpha Chemical Technology Co., Ltd., and curing accelerator 2-ethyl-4-methylimidazole was purchased from Qingdao Haida chemical Co., Ltd. Silicate glass frit (SGF) was supplied by Kunshan Huaqiao Co., Ltd., which is composed of SiO2: 66.32 mass % Na2O: 19.76 mass % TiO2: 6.12 mass % F: 3.71 mass % others: 3.09 mass %. Ammonium polyphosphate (APP) was purchased from JLS Flame Retardants Chemical Co., Ltd.

Sample preparation and pyrolysis

Epoxy composites were prepared as shown in Fig. 1. The preparation includes the following steps. Firstly, the mixture of epoxy resin and MTHPA (epoxy resin: MTHPA = 5:4, by mass ratio) was heated in oil bath at 85 °C under stirring. When the mixture was evenly mixed, 2-ethyl-4-methylimidazole was added to the above mixture with stirring. APP and SGF were then added with stirring for 20 min. Finally, the composite was cured in an oven at 145 °C for 4 h. The same method was used to prepare samples without APP as an experimental comparison. The formulations of samples are shown in Table 1. All samples were fired at 600, 700, 800, 900, and 1000 °C for 30 min in a muffle furnace, respectively.

Fig. 1
figure 1

Schematic showing the preparation of epoxy/SGF/APP (EPSA) composite

Full size image
Table 1 Formulations of samples
Full size table

Characterizations

Impact strength was carried out on an Izod impact test machine (HT-503, Dongguan Hengtong, China) according to ASTM D4812–2006.

Limited oxygen index (LOI) values were tested on a HC-2 oxygen index instrument (Jiangning, China) with sheet dimensions of 100 × 10 × 4 mm3 according to GB/T2406 standard; the vertical burning test (UL-94) was conducted using the vertical burning test instrument (CZF-2) (Nanjing Jion-glei Experimental Equipment Co., Ltd, Nanjing, Jiangsu, China) according to ASTMD 3801 standard.

Fire behavior of the samples was measured on 100  ×  100  ×  4 mm3 sheets according to ISO 5660 by a standard cone calorimeter (Fire Testing Technology Ltd. Sussex, Britain). Samples were burned horizontally at a heat flux of 50 kW m−2.

Linear shrinkage was calculated by measuring the length of the samples before and after firing at high temperatures. The linear shrinkage was calculated using Eq. (1):

$$L = \left( {L_\text{d} - L_\text{o} } \right)/L_\text{d} \times 100\%$$
(1)

where L is linear shrinkage (%), Ld is the dimensional length of sample after firing and Lo is the original length of the sample.

Universal testing machine (CMT-6203; MTS System Corporation, China) was used to test the flexural strength of residues by three-point bend mode, according to the GB/T9596-2006 standard.

Microstructure and elemental dispersion of samples were characterized by a scanning electron microscope (SEM; Japan Electronics Co. Ltd., China) equipped with an energy-dispersive X-ray (EDX; Thermal Scientific, USA) spectrometer.

XRD data were obtained from 5° to 65° at a scan rate of 10° min−1 to study the phase composition of the residues fired at different temperatures by an X-ray diffractometer (XRD; Rigaku Corporation, Japan).

Thermal gravimetric (TG; Netzsch instruments Co., Germany) analysis was conducted on a thermal analyzer from room temperature to 800 °C (nitrogen atmosphere, heating rate of 10 oC min−1).

X-ray photoelectron spectroscopy (XPS; TFS Escalab 250Xi, USA) to identify the electronic state of the residues.

Results and discussion

Morphology analysis

The morphology of residues of EPS and EPSA fired at different temperatures is shown in Fig. 2. After firing at 1000 °C, EPS had a significant deformation due to the high flow capacity of the melted SGF. By contrast, the original shape of EPSA fired at 1000 °C can still be maintained, indicating that the fire resistance of the composite is significantly improved due to the addition of APP. In addition, the carbonization of polymer matrix is considered a significant factor for the blackness of the residues [32]. It is worth noting that the surface of EPSA residues became flat and smooth with the increase in temperature, because SGF was melted into liquid phase at the surface of the sample.

Fig. 2
figure 2

Surface morphology of EPS (left) and EPSA (right) fired at different temperatures

Full size image

Mechanical properties of epoxy composites and residues

Mechanical properties of epoxy composites

The impact strength of samples EP, EPS, and EPSA is shown in Table 2. For the EP, the impact strength is 25.70 kJ m−2. When SGF and APP were added to epoxy resin, the impact strength decreased, and this may result from the poor compatibility between the epoxy resin and inorganic fillers [33, 34]. The impact strength of EPS and EPSA was similar; both are larger than 4 kJ m−2.

Table 2 Impact strength of EP and epoxy composites
Full size table

Mechanical properties of ceramic residues

Composites residues of sample EPS fired at 800, 900, and 1000 °C adhered to the surface of refractory bricks. Hence, the flexural strength of EPS residues was not tested. The flexural strength of the EPSA fired at various temperatures is shown in Fig. 3. When the temperature increased from 600 to 900 °C, an increase in the flexural strength of EPSA residues is observed, which may result from more liquid phase produced by melting SGF, and that is beneficial to generate denser residue structures. The flexural strength of the EPSA residue was significantly improved at the temperature from 600 to 700 °C. However, there is an obvious decrease in the flexural strength of the ceramic residues at the temperature from 900 to 1000 °C, which may result from the reduction in the liquid phase in the residue.

Fig. 3
figure 3

Flexural strength of EPSA fired at different temperatures

Full size image

Linear shrinkage analysis

Figure 4 shows the linear shrinkage of EPSA fired at various temperatures. The linear shrinkage of EPSA increased with increasing temperature. The melting temperature of SGF is around 660 °C, when the firing temperature is higher than that, SGF will melt into liquid phase and could fill the pores formed by the decomposition of EP and APP, [27] which leads a significant increase in the linear shrinkage of the composite. There is a more significant increase in the linear shrinkage of samples fired from 900 to 1000 °C, possibly due to crystallization reaction [24, 35]. The crystallization reaction led to the reduction in liquid phase in the residues. Liquid phase as binding phase of residues can improve the density of the ceramic-like residue. Thus, decreased liquid phase led to a decrease in the flexural strength of the ceramic residue.

Fig. 4
figure 4

Line shrinkage of EPSA fired at different temperatures

Full size image

Combustion performance of epoxy composites

The LOI and UL-94 results of samples EP, EPS, and EPSA are shown in Table 3. The epoxy resin was easily flammable and its LOI value was 19.5%. When SGF was incorporated into EP, the LOI of sample EPS increased to 28.5%, indicating that SGF had an influence on the flammability of epoxy resin. When APP was added to EPS, the LOI of sample EPSA slightly decreased to 24.5%. It can be seen from Table 3 that EP has serious melt dripping. Samples EPS and EPSA, however, are out of burning drip. Unfortunately, EPS and EPSA exhibit no rating without self-extinguish in the UL-94 vertical burning test. Obviously, SGF and APP promoted the flame retardance of the epoxy resin.

Table 3 LOI and UL-94 results of EP and epoxy composites
Full size table

Cone calorimeter test was also applied to assess the combustible behaviors of samples. The cone calorimeter results are shown in Fig. 5 and Table 4. Compared to epoxy resin, the ignition time for samples EPS and EPSA was extended, illustrating that the fire safety of samples EPS and EPSA is higher than that of epoxy resin. Cone calorimeter result confirmed that epoxy resin was flammable. After ignition, a very sharp peak of heat release rate (PHRR) of 944.95 kW m−2 and its total heat release (THR) and total smoke production (TSP) were 112.05 MJ m−2 and 47.45 m2, respectively. For sample EPS, the PHRR and THR were 359.73 kW m−2 and 54.28 MJ m−2, respectively. For sample EPSA, the PHRR and THR were 534.91 kW m−2 and 59.83 MJ m−2, respectively. Obviously, both the PHRR and THR values of samples EPS and EPSA decreased significantly in comparison with the corresponding values of epoxy resin. The TSP values of samples EPS and EPSA were only 21.01 m2 and 16.57 m2, respectively, which is much lower than that of epoxy resin. Obviously, SGF and APP exhibited a high efficiency in reducing the smoke release.

Fig. 5
figure 5

Cone calorimetric curves of EP, EPS, and EPSA, HRR (a), THR (b), and SPR (c)

Full size image
Table 4 Cone calorimetric data of EP and ceramifiable epoxy composites
Full size table

The char layer and ceramic-like layer could contribute to the flame retardation of epoxy composite, which would be an important reason for better flame retardance of samples EPS and EPSA than that of epoxy resin [36, 37]. Unfortunately, although the flame retardance of epoxy composite only adding SGF is much better than that containing SGF and APP from the data of LOI and CONE, the fire resistance of EPS cannot meet the usage requirement in some special fields. This may be due to the release of NH3 and H2O in the pyrolysis process of sample EPSA, which damaged the retardant layer to some extent [17].

Thermogravimetric analysis

The TG curves of pure SGF, pure APP, EP, EPS, and EPSA are shown in Fig. 6, and the corresponding data are summarized in Table 5. During the thermal degradation process of the pure SGF, it can be observed that the mass loss was less than 1%. Unlike the pure SGF, the thermal degradation of APP consisted of two mass loss steps. The first step of thermal degradation temperature of pure APP started at 295 °C and ended at 469 °C, owing to the release of NH3 and H2O. The second step from 544 to 683 °C was attributed to the decomposition and volatilization of P-O compounds [24, 38, 39]. It can be seen that the thermal degradation process of EP started at 278 °C and ended at 444 °C, and the mass of the residue was close to zero. Comparing the thermal decomposition temperature characteristics of EP and pure APP, it can be concluded that the EP was decomposed at a certain temperature accompanied by partial APP decomposition.

Fig. 6
figure 6

Thermal behaviors of SGF, APP, EP, EPS, and EPSA under nitrogen atmosphere

Full size image
Table 5 Thermal characteristics of SGF, APP, EP, EPS, and EPSA
Full size table

Phosphoric acid was first formed after the first step decomposition of APP according to Jiang’s research,36 and the char decomposed by EP was coated with phosphoric acid. Moreover, the initial decomposition temperature of EPS was 328 °C much higher than 278 °C of EP. It indicates that the initial thermal decomposition temperature of EPS was delayed, which is attributed to the presence of silica in the SGF [40,41,42]. Further, comparisons between the experimental and calculated mass remaining of EPS and EPSA are shown in Table 2. The experimental mass remaining of EPS was higher than the calculated mass, suggesting that the addition of SGF can promote the char yield of EP. Similarly, the experimental mass remaining of EPSA was also higher than the calculated mass, indicating that the APP in composite could be able to further increase the mass of the residue.

Theoretically, EPSA sample would undergo two steps of thermal decomposition process due to the addition of APP, and the second step was caused by the decomposition and volatilization of P-O compounds. However, thermal degradation of EPSA sample only consisted of one step in the real experiment, which indicates that the thermal stability of the composite was improved. As the temperature was further increased, the char was further retained in the residue due to the effect of adhesion of molten SGF, leading to the black residues as shown in Fig. 2. Moreover, phosphoric acid and the molten SGF on the surface of the composite lead to the production of dense ceramic-like barrier char and prevent combustible volatiles from escaping and burning in the fire zone.

SEM analysis of ceramic residues

Figure 7 shows the surface morphology of EPSA fired at high temperatures. The surface of EPSA fired at 600 °C showed porous structure, which is due to the release of NH3, H2O and CO2 from the thermal degradation of APP and EP [23, 25]. However, the porous structures for EPSA fired at 800 °C obviously decreased, because the pores were filled with sufficient molten SGF liquid phase, leading to the smooth surface of the EPSA residue. Notably, the flexural strength of EPSA residues increased with the more compact structure of the residue. Compared with the surface morphology of EPSA fired at 800 °C, no significant liquid phase is observed on the surface of residue fired at 1000 °C. The residue exhibited a rough surface, which is not conducive to improve its mechanical properties, resulting in a decrease in flexural strength.

Fig. 7
figure 7

Surface morphology of sample EPSA fired at 600, 800, and 1000 °C

Full size image

XPS and XRD analysis for the residues at various temperatures

In order to investigate the interaction of the additives during firing process, the residue powder of EPSA fired at 800 °C, SGF powder and APP powder were tested by XPS. It is well known that silicon exists in the form of bridging silicon and non-bridging silicon. The Si2p XPS spectra of SGF powder and the ceramic residue powder of EPSA fired at 800 °C are shown in Fig. 8a, b. In Fig. 8a, the Si2p peaks at 102.3 and 102.8 eV corresponded to Si–O and Si–O–Na in SGF, which was a kind of silicon-terminal oxygen bond. However, for the residue powder of EPSA fired at 800 °C, there was a new Si2p peak appeared at 103.5 eV corresponding to the SiO2, as shown in Fig. 8b. Si–O in SiO2 was a kind of silicon-bridge oxygen bond. Hence, the chemical state of the silicon changed at high temperatures after adding APP [43, 44].

Fig. 8
figure 8

XPS spectra of SGF (a), APP(c) and the residue of EPSA (b, d) fired at 800 °C

Full size image

To confirm the effect of APP on the changing of chemical state of silicon, the XPS test was performed for APP. In Fig. 8c, P2p peaks at 134.2 and 135.7 eV corresponded to PO4 and P2O5, while for the residue powder of EPSA fired at 800 °C, the binding energy of P2p (135.7 eV) disappeared. But a new P2p peak appeared at 133.2 eV attributing to the P2O74− [45]. It illustrated that the APP participated in the ceramization process. According to our previous research [24, 26], the sodium binding with non-bridge oxygen ions (such as ≡ Si–O–) would be captured by phosphorus at high temperatures (Eq. 2). This promoted silicon-terminal oxygen bond changed into silicon-bridge oxygen bond (Eq. 3).

$${\text{Si}} - {\text{O}} - {\text{Na}}^{ + } + {\text{P}}_{2} {\text{O}}_{5} \to {\text{Si}} - {\text{O}} - {\text{Si}} + {\text{Na}}_{4} {\text{P}}_{2} {\text{O}}_{7}$$
(2)
$${\text{Si}} - {\text{O}} - {\text{Si}}\mathop{\longrightarrow}\limits{\text{High Temperature}}{\text{SiO}}_{2} ({\text{Cristobalite}})$$
(3)

XRD was adopted to study the phase composition of SGF (Fig. 9) and EPSA (Fig. 10) fired at different temperatures. Figure 9 shows that there was only an amorphous “hump” can be observed, indicating that the phase composition of SGF fired at various temperatures was basically the same and no crystalline phase was shown [18, 24, 47]. For EPSA as shown in Fig. 10, some diffraction peaks appeared at positions of 20.07°, 26.45°, and 33.07°, fitting to the Na4P2O7 phase [24]. It is worth noting that some new diffraction peaks appeared at 21.81, 28.28, 31.43, and 36.13° with increasing temperature, which confirmed the formation of cristobalite phase [48]. Further, the characteristic diffraction peaks at 21.81°, 28.28°, 31.43°, and 36.13° corresponded to (101), (111), (102), and (200) lattice planes of cristobalite crystal, respectively [49]. Therefore, the addition of APP led to a crystallization reaction.

Fig. 9
figure 9

XRD patterns of SGF fired at different temperatures

Full size image
Fig. 10
figure 10

XRD patterns of sample EPSA fired at different temperatures

Full size image

EDX analysis

Surface element composition of EPSA fired at 600, 800, and 1000 °C was analyzed by EDX, and the elemental mapping results are presented in Fig. 11. It can be observed that the content of silicon decreased gradually with the increasing temperature, whereas the content of phosphorus and sodium increased gradually. This indicated that phosphorus element migrated to the surface of the residue with increasing temperature. The migration of phosphorus was attributed to the movement of phosphoric acid-char layer, and the movement of phosphoric acid-char layer was caused by the release of NH3 and H2O generated by decomposition of APP and epoxy resin [50, 51]. According to our previous research [24], the bond energy of P–O was higher than the bond energy of Si–O. The Na2O-rich area formed because O of Na–O was easily captured by P2O5, resulting in the sodium in the SGF migrated into the phosphate phase during the migration of phosphorus. Therefore, elemental migration promoted the formation of Si-rich regions in SGF. The Si–O non-bridging oxygen bond in the Si-rich region converted into Si–O bridging oxygen bond at high temperatures, which was beneficial to promote the amorphous silica convert into crystal in this region [51].

Fig. 11
figure 11

EDX spectra of EPSA fired at 600 °C (a), 800 °C (b), and 1000 °C (c)

Full size image

However, the P-Na region that migrated to the surface converted to phosphate phase, and phase separation provided the possibility of crystal formation. Moreover, the epoxy composite can still maintain its original shape without melting collapse at high temperatures due to the precipitation of crystals.

In order to further investigate the elemental distribution of the residue, the internal zone and surface zone of the EPSA fired at 1000 °C was tested as shown in Fig. 12. The spectrum (Fig. 12a) showed that internal zones were mainly composed of silicon and oxygen, and the mass ratio of the two elements was close to the mass ratio of Si to O in SiO2. There were still a small amount of phosphorus and sodium elements presented, which could be caused by incomplete phase separation. Meanwhile, the elements of the surface zones were mainly phosphorus and sodium, and the mass ratio of phosphorus to sodium was close to the mass ratio of P and Na in Na4P2O7. It was also inevitable that a small amount of silicon remained on the residue surface.

Fig. 12
figure 12

The elemental content of internal (a) and surface (b) of the EPSA fired at 1000 °C

Full size image

Phase separation and crystallization reaction mechanisms

The process of phase separation and crystallization reaction within the EPSA is proposed in Fig. 13. Firstly, the thermal decomposition of epoxy resin and APP in EPSA occurred at the beginning of firing, which produced H2O, NH3, and CO2. Then, phosphoric acid was formed during the decomposition of APP, and the char decomposed by epoxy resin was coated by the formed phosphoric acid. Successively, the phosphoric acid-char layer structure was produced. As the temperature was further increased, phosphorus oxides were generated. The generation of NH3 and H2O in the composite as well as the thermal driving force promoted the migration of the phosphoric acid-char layer. The phosphoric acid-char layer began to migrate toward the surface of the residue due to the driving force of the gases and heat, and the content of phosphorus element on the surface was increased. Meanwhile, the sodium in the SGF migrated into the phosphate phase during the process of phosphoric acid-char layer migration, because the bond energy of P–Na was higher than the bond energy of Si–Na. The P–Na region that migrated to the surface converted to a phosphate phase, forming a sodium pyrophosphate phase. Finally, a Si-rich region was formed in the SGF after the sodium migrated to the surface to form a phosphate phase. The Si–O non-bridging oxygen bond in the system was converted into Si–O bridging oxygen bond at high temperature, so the amorphous silica in this region was converted into cristobalite phase. This process showed that phase separation provided the conditions for the crystallization reaction.

Fig. 13
figure 13

Schematic illustrations of phase separation and crystallization reaction mechanisms of epoxy composite fired at high temperatures

Full size image

Conclusions

In this work, a novel epoxy composite with flame retardance and high fire resistance was prepared by adding SGF and APP into the epoxy resin. The addition of SGF and APP as well as the formation of residue improved the flame retardance of the epoxy composite. The improvement of flame retardance and high fire resistance of the epoxy composite mainly benefited from the phase separation and crystallization reactions between APP and SGF during the composite firing process. The flexural strength of composite residue formed at different temperatures (600–1000 °C) kept at a relatively high level, and the original shape of the composite cannot deform seriously at high temperatures. The essential precondition for the phase separation was that the interplay between the epoxy resin matrix and APP at high temperature. The essential reason for the crystallization reactions was that sodium from SGF was captured by phosphoric acid during the phosphoric acid migration process. The migration of sodium made the transformation of amorphous silica to cristobalite crystal easier at high temperatures.