Enhanced Flame Retardancy of Silica Fume-Based Geopolymer Composite Coatings Through In Situ-Formed Boron Phosphate from Doped Zinc Phytate and Boric Acid

Abstract

Silica fume-based geopolymer composite coatings, an approach to using metallurgical solid waste, exert flame retardancy with ecological, halogen-free, and environmentally friendly advantages, but their fire resistance needs to be improved further. Herein, a silica fume-based geopolymer composite flame-retardant coating was designed by doping boric acid (BA), zinc phytate (ZnPA), and melamine (MEL). The results of a cone calorimeter demonstrated that appropriate ZnPA and BA significantly enhanced its flame retardancy, evidenced by the peak heat release rate (p-HRR) decreasing from 268.78 to 118.72 kW·m⁻², the fire performance index (FPI) increasing from 0.59 to 2.83 s·m²·kW⁻¹, and the flame retardancy index increasing from 1.00 to 8.48, respectively. Meanwhile, the in situ-formed boron phosphate (BPO₄) facilitated the residual resilience of the fire-barrier layer. Furthermore, the pyrolysis kinetics indicated that the three-level chemical reactions governed the pyrolysis of the coatings. BPO₄ made the pyrolysis E_α climb from 94.28 (P5) to 127.08 (B3) kJ·mol⁻¹ with temperatures of 731–940 °C, corresponding to improved thermal stability. Consequently, this study explored the synergistic flame-retardant mechanism of silica fume-based geopolymer coatings doped with ZnPA, BA, and MEL, providing an efficient strategy for the high-value-added recycling utilization of silica fume.

1. Introduction

Wood is a biological material that has been widely used in the fields of decoration, architecture, and furniture because of its renewable, environmental, and abundant resources [1,2]. However, wood is highly flammable [3], and its extensive use in daily life inevitably holds fire risk. The most direct and effective method to mitigate this risk is by brushing flame-retardant coatings on its surface [4], which minimizes the fire risk of flammable materials, benefits personnel evacuation and rescue, and reduces fire losses. Building coatings with environmental protection, ecology, low toxicity, and efficient fire resistance have become an important direction for the research and development of flame-retardant coatings.

Recently, inorganic geopolymer has emerged as a promising candidate for flame-retardant coatings due to its durability, halogen-freeness, environmental friendliness, and low-cost advantages [5,6]. Silica fume is a metallurgical by-product [7] with high emissions but low utilization efficiency that has been used as a flame-retardant coating. Shahidi et al. [8] investigated the flame retardancy of an intumescent flame retardant enhanced by silica fume and graphene oxide/talc. However, silica fume-based geopolymer coatings exhibit certain limitations in practical applications, such as insufficient flame retardancy [9] and low residual structural strength [10], failing to protect plywood effectively. Therefore, it is necessary to optimize its flame retardancy through a synergistic flame-retardant mechanism.

Currently, one of the most commonly used flame retardants is melamine (MEL) due to its lower cost and non-toxicity during combustion, without secondary pollution. During combustion, the MEL takes away heat and releases N₂, NH₃, and other non-flammable gases, diluting the concentration of O₂ and toxic gases. However, the flame retardancy of composite coating is not effectively improved by adding MEL alone. In recent years, numerous flame-retardant systems consisting of a variety of materials have been developed to seek synergistic flame-retardant effects [11,12]. For instance, combinations such as phosphorus/boron [13] and phosphorus/boron/nitrogen [14] have been demonstrated to achieve synergistic flame retardancy. Zhang et al. [15] improved the thermal stability and flame retardancy of furfurylated wood by introducing a multifunctional catalyst system comprising BA and ammonium dihydrogen phosphate. Huo et al. [16] incorporated hyperbranched oligomer-containing phosphorus/nitrogen/boron (BDHDP) in epoxy resin (EP), achieving a UL-94 V-0 level with 1.5 wt.% doped BDHDP. Consequently, it is a good strategy to seek a multi-element synergistic effect for improving the flame retardancy of geopolymer coatings.

Zinc phytate (ZnPA) is formed by chelating phytic acid and zinc ions, serving as an eco-friendly and efficient flame retardant due to its high phosphorus content [17]. Phosphorus-containing flame retardants facilitate catalytic charring and combustion prevention and generate phosphorus radicals capable of capturing active radicals, thereby quenching the combustion chain reaction [18]. Zhang et al. [19] used Zn²⁺ ions in ZnPA to chelate with components such as dopamine, DOPO, and fly ash, preparing an organic–inorganic hybrid flame-retardant coating. In this sense, ZnPA could form a zinc ion chelate cross-linked structure with the silica fume-based geopolymer to enhance the stability and continuity of the residues during fire.

Additionally, boric acid (BA) is another flame retardant widely used across various industries; it acts as an effective flame-retardant by forming a protective B₂O₃ glassy layer during combustion, which shields material from oxygen and heat [20]. Savas et al. [21] investigated the impact of zinc borate (ZnB) on the thermal and flame retardancy of polyamide 6 (PA6) composites containing aluminum hypophosphite (AlHP), where the as-formed BPO₄ enhanced the flame retardancy. Therefore, the synergism between BA and ZnPA could further form a zinc-chelated silicon phosphorus boron composite structure within the silica fume-based geopolymers during combustion in theory, thereby enhancing the flame retardancy.

Consequently, this research designed flame-retardant coatings integrating silica fume-based geopolymer with ZnPA/BA/MEL, characterized by X-ray diffraction (XRD), a cone calorimeter (CC), and scanning electron microscopy (SEM), respectively. It demonstrates that in situ-formed BPO₄ significantly enhances the residue strength and improves the flame retardancy of the condensed phase. Generally, the novelty of this thesis is in proposing a recycling strategy for high-value-added utilization of silica fume, probing the synergistic flame-retardant mechanism of ZnPA/BA/MEL-containing silica fume-based geopolymer coatings, and exploring the design of a halogen-free Si-P-N-C multi-element composite flame retardant system.

2. Experiment and Methods

2.1. Raw Materials

The silica fume was a gray powder with a SiO₂ content > 86 wt.% (mass percentage concentration), a density of 1.62 g/cm³, and a Blaine-specific surface area of 25 m²/g, purchased from Xi’an LinYuan company (Xi’an, China). ZnPA was procured from Guangzhou Jiale Chemical company (Guangzhou, China). Melamine (MEL) was obtained from Wuxi Yatai United Chemical company (Yantai, China). BA was supplied by Zhengzhou Del Boron Chemical company (Zhengzhou, China). Analytically pure polyacrylamide (PAM) was purchased from Tianjin Fuchen chemical reagent factory (Tianjin, China). Analytical pure sodium silicate (Na₂SiO₃·9H₂O) was purchased from Sinopharm chemical reagent company (Jinan, China). Potassium hydroxide (KOH) was purchased from Hongyan chemical reagent company (Tianjin, China). The silane coupling agent (KH-550, CAS no.: 919-30-2) was provided by Shandong Yousuo chemical technology company (Jinan, China). Polydimethylsiloxane (PDMS) was purchased from Dongguan Tianyu chemical company (Dongguan, China). The plywood was supplied by the timber processing plant of Xi’an in China with secondary flame retardancy.

2.2. Preparation of Geopolymer Composite Coating

Figure 1 illustrates the preparation steps of silica fume-based geopolymer composite coatings via the multi-step sol–gel method. Firstly, 0.5 g KH-550, 1 g MEL, and varying dosages of ZnPA were dissolved in 15 g of water, according to our previous research [22], as shown in

Table 1. The composition of coating samples.

Samples	Na2SiO3·9H2O/g	KOH/g	Silica Fume/g	KH-550/g	MEL/g	PAM/g	H2O/g	PDMS/g	ZnPA/g	BA/g
P0	14.21	5.61	30	0.5	1	0.2	45	0.25	0.48	0.0
P1	14.21	5.61	30	0.5	1	0.2	45	0.25	0.95	0.0
P2	14.21	5.61	30	0.5	1	0.2	45	0.25	1.43	0.0
P3	14.21	5.61	30	0.5	1	0.2	45	0.25	1.9	0.0
P4	14.21	5.61	30	0.5	1	0.2	45	0.25	2.38	0.0
P5	14.21	5.61	30	0.5	1	0.2	45	0.25	2.85	0.0
P6	14.21	5.61	30	0.5	1	0.2	45	0.25	2.38	0.0
B1	14.21	5.61	30	0.5	1	0.2	45	0.25	2.38	0.5
B2	14.21	5.61	30	0.5	1	0.2	45	0.25	2.38	1.0
B3	14.21	5.61	30	0.5	1	0.2	45	0.25	2.38	1.5
B4	14.21	5.61	30	0.5	1	0.2	45	0.25	2.38	2.0
B5	14.21	5.61	30	0.5	1	0.2	45	0.25	2.38	2.5
B6	14.21	5.61	30	0.5	1	0.2	45	0.25	2.38	3.0

. Magnetic stirring at 700 r·min⁻¹ at 60 °C was applied for pretreating ZnPA for approximately 30 min.

Secondly, 14.21 g Na₂SiO₃·9H₂O, 5.61 g KOH, 30 g silica fume, and 30 g water were poured into another glass beaker [22]. The silica sol mixture was obtained by 10 min of stirring at 70 °C with 750 r·min⁻¹. Then, a pretreated ZnPA solution was injected into the silica sol and mixing was continued for another 20 min, with the speed increased to 1000 r·min⁻¹.

Finally, PAM as a thickener and PDMS as an antifoaming agent were doped into ZnPA/MEL-containing silica fume-based geopolymer coatings by stirring for 5 min. The dosages of ZnPA were 0 g (0.0 wt.%), 0.48 g (0.5 wt.%), 0.95 g (1.0 wt.%), 1.43 g (1.5 wt.%), 1.9 g (2.0 wt.%), 2.38 g (2.5 wt.%), and 2.85 g (3.0 wt.%). Then, the as-formed composite coating was covered evenly on the surface of the plywood (100 × 100 × 5 mm³) 3 times with an interval of about 20 min.

Based on the results of the ZnPA-containing geopolymer coatings, further addition of BA was carried out to prepare BA/ZnPA/MEL-containing silica fume-based geopolymer composite coatings. The preparation process was identical to that of the ZnPA-containing coating, as shown in Figure 1B. The dosages of BA were 0.5 g (0.5 wt.%), 1.0 g (1.0 wt.%), 1.5 g (1.50 wt.%), 2 g (2.0 wt.%), 2.5 g (2.5 wt.%), and 3.0 g (3.00 wt.%). The samples with a ZnPA dosage of 0.5 wt.%–3 wt.% were numbered as P1–P6. BA dosages of 0.5 wt.%–3 wt.% were referred to as B1-B6, respectively, and the coating without ZnPA or BA was denoted as the control P0.

2.3. Characterizations

2.3.1. Flame Retardancy Testing

The combustion performance of the sample was assessed by a cone calorimeter (CC, ZY6243, Zhongnuo Instruments Company, Dongguan, China) according to ISO 5660-1:2015 [23]. The irradiative heat flux of CC was 50 kW·m⁻² (approximately 715 °C), with a distance between the coating and the ignition needle of 25 mm. The following parameters were recorded in real time by CC, such as time to ignite (TTI, s), heat release rate (HRR, kW·m⁻²), peak heat release rate (p-HRR, kW·m⁻²), time to p-HRR (T_P, s), total heat release (THR, MJ·m⁻²), and weight loss (WL, g). Meanwhile, the following four parameters were used to assess the flame retardancy of the sample. The fire growth index (FGI) reflected the potential growth and intensity of a fire, which was calculated by Formula (1). The fire performance index (FPI) evaluated the fire performance of the materials, which was calculated by Formula (2). The average effective heat of combustion (AEHC) was the ratio of THR to WL. The flame retardancy index (FRI) was used to quantify the flame resistance of the materials, which was calculated by Formula (3). It was divided into three grades: “FRI < 1”, “1 < FRI < 10”, and “10 < FRI < 100”, which represented “poor”, “good”, and “excellent” flame retardancy, respectively [22].

$$FGI={\displaystyle \frac{p-HRR}{T_p}}$$

(1)

$$FPI={\displaystyle \frac{TTI}{p-HRR}}$$

(2)

$$FRI={\displaystyle \frac{THR\ast {(\frac{p-HRR}{TTI})}_{Control}}{THR\ast {(\frac{p-HRR}{TTI})}_{Composite}}}$$

(3)

2.3.2. Microstructure Testing

Infrared spectra of the samples before combustion were captured in the spectral range of 4000–500 cm⁻¹ by an infrared spectrometer Nicolet iS50 (Thermo, Singapore). The composition of the residual layer was analyzed by an X-ray diffractometer (D/MAX-2400) with Cu K_α (λ = 1.54056 Å) radiation at 40 kV and 40 mA within the range of 2θ = 10~50°, with a scanning step of 2°, which was preferentially ground into powder. The morphological structures of the barrier layer were observed using a scanning electron microscope (Gemini 500). Each sample needed to be sprayed with gold before scanning electron microscopy to improve the conductivity of the sample. Thermogravimetric analysis (TGA) of samples was conducted by the Mettler (Germany) thermal–gravimetric analyzer, from 30 to 1000 °C at a rate of 20 °C·min⁻¹ in a N₂ atmosphere.

3. Results

3.1. Flame Retardancy of Samples

Figure 2a shows the HRR curve for the ZnPA-containing samples, which gradually shifted to the right with the increasing content of ZnPA. When the ZnPA dosage was 2.38 g, p-HRR exerted the lowest value, decreasing from 268.78 (P0) to 156.35 kW·m⁻² (P5), and T_P was delayed to 404 s, indicating that an appropriate ZnPA content imparted greater flame retardancy. However, doping in excess of 2.85 g ZnPA led to a rise in p-HRR and the early emerged peak at 144 s because the excess ZnPA accelerated the coating catalytic decomposition and uneven dispersion in the coatings.

Figure 2a’ displays the HRR curve for the BA-containing samples. The HRR curve decreased at first and then increased. Compared with P5, B3 exhibited a lower p-HRR value of 118.73 kW·m⁻² and T_P was delayed from 404 s to 440 s. When the BA content exceeded 1.50 wt.%, p-HRR gradually rose, indicating that the excess BA led to a decrease in the flame retardancy due to the uneven dispersion of the doped BA in the coatings.

Figure 2b,b’ show the smoke temperature of the samples, which gradually decreased with the increasing ZnPA content, as shown in Figure 2b. Among these, P5 presented the lowest smoke temperature, reaching a peak value of 88.66 °C at 438 s. This indicated that the doped ZnPA gave enhanced flame retardancy to the composite coating. Regarding the 3.00 wt.% ZnPA-containing sample, the smoke temperature rose to 92.39 °C, corresponding to poor flame retardancy. As shown in Figure 2b’, B3 displayed the lowest smoke temperature at 494 s, decreasing from 88.66 (P5) to 84.68 °C. This was attributed to BA dehydrating during combustion, which had an endothermic and cooling effect. However, samples with a BA content exceeding 1.50 wt.% exhibited a higher smoke temperature in comparison to that of B3. The smoke temperature increased with the increasing BA content, which was completely consistent with the results of HRR.

Figure 2c,c’ present the residual weight value of the samples under an externally constant heat flux of 50 kW·m⁻². The lowest final residual weight was observed for P0 (35.29%). As the content of ZnPA increased, the residual weight also increased. P5 displayed the maximum final residual weight of 46.09% at 600 s, indicating that the incorporated ZnPA caused the composite coating to have a reduced weight loss rate. As shown in Figure 2c’, the residual weight increased and then dropped with the increasing BA dosage. Compared to P5, B3 exhibited a lower weight loss of 50.97% at 650 s. The residual weight decreased sequentially for B4, B5, and B6, indicating that a higher BA content was detrimental to improving the flame retardancy of the composite coating.

Table 2. Combustion parameters of samples in CC.

Samples	TTI/s	TP/s	p-HRR/kW·m−2	FPI/s·m2·kW−1	FGI/kW·m−2·s−1	WL/g	THR/MJ·m−2	AEHC/kW·kg−1	FRI
P0	158	314	268.78	0.59	0.86	30.93	36.67	1.19	1.00
P1	219	308	242.31	0.90	0.79	29.16	31.27	1.07	1.80
P2	201	319	234.90	0.86	0.74	29.93	35.93	1.20	1.49
P3	232	321	223.93	1.04	0.70	26.04	27.92	1.07	2.31
P4	205	364	213.97	0.96	0.59	24.56	27.79	1.13	2.15
P5	228	404	156.35	1.46	0.39	27.12	27.79	1.02	3.27
P6	78	348	215.41	0.36	0.62	31.37	40.44	1.29	0.56
B1	266	414	147.49	1.80	0.36	26.84	22.42	0.84	5.02
B2	294	423	131.89	2.23	0.31	29.46	23.44	0.80	5.93
B3	336	440	118.72	2.83	0.27	24.30	20.83	0.86	8.48
B4	327	453	123.31	2.65	0.27	23.90	22.62	0.95	7.31
B5	318	443	128.29	2.48	0.29	25.01	20.68	0.83	7.48
B6	260	431	159.23	1.63	0.37	26.51	27.63	1.04	3.69

summarizes the combustion parameters of the samples. The doped ZnPA and BA extended the TTI. Among all samples, B3 exhibited the longest TTI of 336 s and the highest flame retardancy, evidenced by the highest FPI of 2.83 s·m²·kW⁻¹ and the lowest FGI of 0.27 kW·m²·s⁻¹. FRI was 8.48, indicating a ‘good’ flame retardant grade compared to P0. Conversely, when the content of BA exceeded 1.50 wt.%, the parameters exhibited a contrary trend. For instance, the TTI of B5 shrank to 318 s compared to B3, the FPI declined to 2.48 s·m²·kW⁻¹, and FGI climbed to 0.29 kW·m⁻²·s⁻¹ with the lower FRI of 7.48, corresponding to diminished flame retardancy. Therefore, the appropriate amount of doped ZnPA and BA generated synergism, leading to the formation of a dense and resilient siliceous layer, corresponding to enhanced flame retardancy.

3.2. FTIR Spectra of Samples

Figure 3 presents the obtained FTIR spectra of the composite coatings before burning. The wide peak at 3398 cm⁻¹ was assigned to stretching vibrations of -OH and N-H [24]. Among them, -OH was mainly derived from geopolymer and BA and N-H were mainly derived from MEL. The broad peak at 1600 cm⁻¹ was assigned to the C=N absorption peak of the triazine ring [25]. The asymmetric vibrations of the Si-O-Si bond were detected at 1120 cm⁻¹, whereas the symmetric vibrations were located at 765 cm⁻¹ [26], attributed to the hydrolysis of silane coupling agents and alkali-activated geopolymer sols. P-O-C symmetric vibrations occurred at 980 cm⁻¹ [27]. After the addition of BA, the absorption peaks of the B3 and B6 samples at 617 cm⁻¹ showed a strengthening, attributed to the deformation vibration of the atoms in the B-O bond in BA [28,29]. Generally, the spectra of samples exhibited similar curves without obvious differences, except for the peaks at 980 cm⁻¹ and 617 cm⁻¹. Although the dosage of ZnPA and BA was very low, presenting tiny peaks, they could improve the flame retardancy, as evidenced by the results of the CC. The specific reasons will be explained later in Section 4.

3.3. XRD Analysis

Figure 4 depicts the X-ray diffraction (XRD) patterns of the raw materials. Figure 5 presents the XRD of the residual silica fume-based geopolymer coatings after combustion. A broad hump observed between 15 and 35° at 2θ suggested a high content of amorphous silicates, which contributed to enhanced flame retardancy. Overlaid upon this broad hump, discrete peaks corresponded to quartz (SiO₂, PDF no. 46-1441, no. 04-0379), graphite (G, PDF no. 05-0625), zinc metaphosphate (Zn(PO₃)₂, PDF no. 01-0587), boric oxide (B₂O₃, PDF no. 46-1045), and boron phosphate (BPO₄, PDF no. 12-0380). Notably, the diffraction peaks at 29.12° and 40.56° were attributed to the hexagonal structure of BPO₄. However, peaks corresponding to melamine and its derivatives were not detected, possibly due to their complete decomposition at high temperatures.

Essentially, ZnPA decomposed into phosphoric acid and Zn(PO₃)₂ while releasing CO₂ through a dehydration process [17], as shown in reaction (4). Subsequently, the partially generated phosphoric acid dehydrated into polyphosphoric acid, which then further dehydrated into P₂O₅ [30], as shown in reactions (5) and (6). BA underwent dehydration at high temperatures, converting into B₂O₃ [31], as shown in reaction (7). The generated P₂O₅ further reacted with B₂O₃ and transformed into BPO₄ [32], as shown in reaction (8), holding a high melting point of 1200 °C and excellent stability [33]. The XRD analysis confirmed that the doped ZnPA and BA generated in situ reactions.

$$C_6{H}_6{O}_{24}{P}_{6}Z{n}_6\to H_{3}P{O}_4+Zn{(P{O}_3)}_2+C{O}_2+H_{2}O$$

(4)

$$n{H}_{3}P{O}_4\to H{(P{O}_3)}_n+n{H}_{2}O$$

(5)

$$H{(P{O}_3)}_n\to n{P}_2{O}_5+H_{2}O$$

(6)

$$2B{(OH)}_3\to B_2{O}_3+3H_{2}O$$

(7)

$$P_2{O}_5+B_2{O}_3\to 2BP{O}_4$$

(8)

3.4. Appearance

Figure 6 and Figure 7 illustrate the appearance of the composite coatings before burning. The coating without ZnPA and BA appears slightly whitened in Figure 6a. Because MEL is an alkaline compound [34], it resulted in excessive alkali content on the coating. With the increasing dosage of ZnPA, the whitening phenomenon diminished, as shown in Figure 6b–g, and Figure 6f presents a flat and smooth surface. The excess ZnPA caused a slightly rough surface, as shown in Figure 6g, due to its uneven dispersion. After the addition of BA, the white powders were dispersed on the surface, as shown in Figure 7a–e. This phenomenon became obvious with the increasing BA dosage. When the content of BA reached 3.0 wt.%, a rough surface with white small particle aggregation appeared, as shown in Figure 7e.

3.5. Residual Appearance of Samples

All samples formed a siliceous layer [35], as shown in Figure 8 and Figure 9. This layer effectively inhibited flame propagation and trapped the transfer of heat and mass, thereby protecting the underlying plywood against complete combustion. As shown in Figure 8a, the siliceous layer was inherently brittle and fragile. As shown in Figure 8b–e, the coating hardly swelled with a low content of ZnPA.

The surface of the layer was unevenly convex, which affected the flame retardancy of the samples and their heat release rate. A continuous swelling layer was observed with the increasing content of ZnPA. Especially for the P5 in Figure 8f, the siliceous layer was uniformly expanded, continuous, and intact, but small cracks could be observed on the surface of the layer. When the content of ZnPA reached 3.0 wt.% in Figure 8g, cracks were observed, which were averse to the formation of the intact layer.

However, the integrity of the siliceous layer gradually improved when the BA was doped into the coatings, as shown in Figure 9a,b. The unbroken, smooth, and robust homogeneous barrier layer was generated for B3 with the addition of 1.5 wt.% BA, as shown in Figure 9c, and a smooth surface was observed, indicating that a resilient swelling siliceous layer was formed. When the BA content exceeded 1.5 wt.%, an inhomogeneous barrier layer emerged, as shown in Figure 9d–f, leading to diminished flame retardancy.

3.6. SEM of Residues

Figure 10 illustrates the microscopic morphology (500× and 2000×) of the residues after the combustion of the composite coatings. Figure 10a,a’ reveal sparse fibrous barriers covering the fractured and discontinuous siliceous layer. This layer was easily penetrated by gases produced from pyrolysis and could not effectively prevent the transfer of heat and flammable gases during firing.

When ZnPA was doped into the coating, as shown in Figure 10b,b’, dense fibrous barriers appeared on the surface of the barrier layer, delaying flame propagation and preventing volatile gases and small liquid molecules from entering the combustion zone. When adding an excessive dosage of ZnPA, the coverage of the fibrous decreased, as shown in Figure 10c,c’, which adversely affected the flame retardancy.

Figure 10d,d’ show the formation of column-like BPO₄ on the surface of the B3 siliceous layer. During combustion, BA and ZnPA decomposed into phosphate and borate, respectively, which further reacted to form column-like BPO₄. The integration of BPO₄ with the siliceous layer enhanced the quality of the barrier layer. Notably, BPO₄ typically adopts a cristobalite-like structure [36]. According to our results, the formed BPO₄ exhibited a column-like structure, a discrepancy that may stem from varying conditions during BPO₄ crystal growth, leading to a diversity of crystal structures [37]. The column-like BPO₄ increased the surface area of the siliceous layer, hindering heat transfer and protecting the underlying plywood from combustion. Meanwhile, the porous surface of the barrier layer facilitated the volatiles’ release into the surrounding air. However, an excess BA resulted in reduced BPO₄ on the surface of the siliceous layer, exhibiting an irregular and amorphous structure, leading to poor flame retardancy. Therefore, the appropriate amount of ZnPA (2.5 wt.%) and BA (1.5 wt.%) led to the in situ formation of BPO₄, enhancing the toughness of the barrier layer, presenting higher flame retardancy.

3.7. Thermal Performance Analysis

According to the thermal performance analysis, 5% weight loss temperature (T_d5%), 10% weight loss temperature (T_d10%), 20% weight loss temperature (T_d20%), the initial degradation temperature (T_i), the final degradation temperature (T_f), and the corresponding maximum weight loss temperature (T_max) and carbon residue rate of the samples are expressed in

Table 3. Pyrolysis parameters of samples.

Samples	a Td5%/℃	b Td10%/℃	c Td20%/℃	d Ti	e Tf	f Tmax/℃	Char Residue/wt.%
P0	133.3	210.0	718.0	30	998.3	250.7	78.61
P5	151.3	243.7	774.3	30	999.3	245.0	78.40
P6	113.3	194.0	515.0	30	998.3	244.3	76.81
B3	80.7	160.7	393.0	30	997.0	252.3	74.14
B6	125.0	243.0	721.3	30	999.0	258.7	77.64

^a Temperature of 5% weight loss (T_d5%). ^b Temperature of 10% weight loss (T_d10%). ^c Temperature of 20% weight loss (T_d20%). ^d Initial decomposition temperature (T_i). ^e Final decomposition temperature (T_f). ^f Temperature of the maximum rate of weight loss (T_max).

. The comparison reveals that B3 had better thermal stability because the reaction of ZnPA and BA formed BPO₄ with high thermal stability [38], thereby blocking the heat transfer further.

Figure 11a shows the TGA curves for the geopolymer coatings, depicting temperature variations at a constant heating rate of 20 °C/min. With the addition of ZnPA and BA, continuous weight loss was observed in all samples during the heating process. This may be attributed to the thermal decomposition of ZnPA, which produced phosphorus compounds, accelerating the dehydration of the composite coatings [39]. The formation of BPO₄ was catalyzed [40], prompting the decomposition of the siliceous layer, consequently reducing the weight loss of the residues.

Figure 11b presents the DTG of the sample coatings. The whole pyrolysis process could be categorized into four stages: (I) 30–198 °C, (II) 198–439 °C, (III) 439–731 °C, and (IV) 731–1000 °C. In the first stage of 30–198 °C, weight loss was primarily attributed to the loss of free water and bound moisture from the coating. In the temperature range of 198–439 °C, the primary reactions involved the decomposition of ZnPA and MEL. MEL released inert gases such as NH₃ [41]. ZnPA decomposed within this temperature range, forming pyrophosphate, metaphosphate, and metal phosphate [42]. After adding BA, BA dehydrated and formed vitreous boron oxide. The third stage occurred at 439–731 °C, primarily due to the thermal decomposition of polyphosphoric acid and the formation of the blocking layer. The final stage occurred between 731 and 1000 °C, mainly attributed to the decomposition of the blocking layer and the formation of BPO₄. Moreover, some unreacted boride and pyrolysis products of ZnPA volatilized at high temperatures. In summary, ZnPA and BA formed BPO₄ at high temperatures within the siliceous coatings, improving the flame retardancy of the composite coating.

3.8. Pyrolysis Kinetics

Pyrolysis kinetics evaluates a material that undergoes decomposition or chemical reactions during heating to obtain the reaction mechanism, activation energy, and pyrolysis kinetics parameters. Using the modified Coats–Redfern integral method, the thermal decomposition kinetics of the coating were calculated using Formulas (9)–(13). A plot of the ln[G(α)/T²] − 1/T relationship was generated. The activation energy E_α and correlation coefficient R² were obtained by the linear regression fitting of the curve.

$$f\left(\alpha \right)={\left(1-\alpha \right)}^n$$

(9)

$$G\left(\alpha \right)={\left(1-\alpha \right)}^{-2}$$

(10)

$${\displaystyle \frac{d\alpha }{dT}}={\displaystyle \frac{A}{\beta }}\exp (-{\displaystyle \frac{E_{\alpha }}{RT}})f(\alpha )$$

(11)

$$\ln [{\displaystyle \frac{G(\alpha )}{T^2}}]=\ln ({\displaystyle \frac{AR}{\beta E_{\alpha }}})-{\displaystyle \frac{E_{\alpha }}{RT}}$$

(12)

$$\alpha ={\displaystyle \frac{m_0-m_t}{m_0-m_f}}$$

(13)

n is the reaction order of combustion. A is the preexponent factor in units of min⁻¹. _Eα is the catalytic activation energy, measured in kilojoules per mole (kJ·mol⁻¹). R is the ideal gas constant, 8.3145 J·mol⁻¹·K⁻¹. β is the heating rate of 20 °C·min⁻¹. T is the thermodynamic temperature in units of K. α is the weight conversion rate of the sample during pyrolysis, %. m₀ represents the initial mass of the sample. m_t is the real-time mass of the coating at time. m_f corresponds to the final remaining mass.

By performing calculations and screening among the 29 commonly used mechanism functions [43], a three-level chemical reaction model (F3) was chosen to plot the curves, as illustrated in Figure 12. This meant that the reaction process could be controlled by regulating reactant design, condition control, equivalence ratio, etc., and the doped ZnPA and BA could affect the pyrolysis. The pyrolysis was mainly divided into the following four stages, as shown in

Table 4. Kinetic parameters of samples calculated using F3.

Sample	Temperature	Intercept	Slope	Adj.R2	Eα/kJ·mol−1
P5	60–198 °C	−11.02	−190.19	0.95	1.58
	198–439 °C	−7.35	−1948.92	0.99	16.20
	439–731 °C	−6.13	−2800.53	0.95	23.28
	731–940 °C	2.20	−11,339.87	0.94	94.28
B3	60–198 °C	−10.40	−333.17	0.95	2.77
	198–439 °C	−7.19	−1916.77	0.99	15.94
	439–731 °C	−7.13	−1929.49	0.93	16.04
	731–940 °C	5.91	−15,284.84	0.93	127.08
B6	60–198 °C	11.32	−61.67	0.94	0.51
	198–439 °C	−7.24	−2062.03	0.99	17.14
	439–731 °C	−7.15	−2172.26	0.98	18.06
	731–940 °C	3.42	−12,986.68	0.91	107.98

: (IV) 60–198 °C, (III) 198–439 °C, (II) 439–731 °C, and (I) 731–940 °C. Stage IV was mainly the volatilization of small molecular substances. In comparison to P5, the E_α of B3 decreased to 15.94 kJ·mol⁻¹ in stage III and 16.04 kJ·mol⁻¹ in stage II. This was because the thermal decomposition of ZnPA and BA catalyzed the thermal degradation and carbonization of the coating. The E_α of B3 was 127.08 kJ·mol⁻¹ in the range of 731–940 °C, indicating that ZnPA and BA reacted with the geopolymer to form a dense siliceous barrier layer with BPO₄, preventing plywood from coming into contact with air and enhancing the flame retardancy.

4. Discussion

In this research, an alkali-activated silica fume-based geopolymer coating was optimized by incorporating MEL, ZnPA, and BA to enhance its flame retardancy. Various characterization techniques, including CC, XRD, SEM, and TGA, were employed to analyze the coating. The results indicate that the optimal combination of BA and ZnPA significantly improved the flame retardancy. The flame-retardant mechanism of ZnPA/BA/MEL-containing silica fume-based geopolymer coatings is summarized as follows.

Firstly, the condensation reactions and the formation of coordination bonds were crucial in enhancing flame retardancy. During the combustion, a condensation reaction occurred, which transformed Si[OH]₄ into the Si-O-Si network structure [44], as shown in reaction (14). This silicon network structure contributed to the heat dissipation of the coating.

Secondly, zinc ions from ZnPA reacted with the Si-O-Si structure to form crosslinking networks [45], as shown in reaction (15). This crosslinking enhanced the strength of the residues [19]. Additionally, ZnPA enhanced flame retardancy by generating phosphorus-containing free radicals that quenched the pyrolysis process [46]. The production of phosphoric acid, polyphosphoric acid, and metal phosphates catalyzed the formation of residual layers.

Furthermore, BA helped in reducing the surface temperature of the coating by producing B₂O₃ and water. Furthermore, zinc ions and boron could enhance charring [47,48]. Moreover, MEL accelerated the production of polyphosphoric acid [49], releasing inert gases, such as NH₃. These gases diluted the combustible gases and O₂ in the vicinity, reducing the combustion intensity.

(14)

(15)

Thirdly, the synergistic interaction between BA and ZnPA promoted the formation of highly thermally stable BPO₄. BPO₄ formed a ceramic shielding layer atop the siliceous layer, as shown in Figure 10. This additional layer significantly enhanced the physical barrier properties of the siliceous layer. Moreover, BPO₄ had a filling effect, strengthening the structural integrity of the residual layer and forming a dense residual layer [50]. This layer acted as a barrier to heat transfer, thereby protecting the underlying plywood. In addition, BPO₄ increased the surface area of the siliceous layer. This was similar to the findings of Decsov et al. [51], who changed the specific surface area of 2-hydroxypropyl-β-cyclodextrin (HP-β-CD) microfibers, leading to an increase in the flame-retardant index of the composites. Thus, BPO₄ introduced more active sites on the surface of the barrier layer. It adsorbed flammable organic volatiles, preventing their release and diffusion during combustion and therefore reducing the rate of flame spread.

Consequently, the synergistic flame-retardant mechanism of silica fume-based composite coatings, as illustrated in Figure 13, primarily involved the quenching effect, dilution action, heat absorption, and catalytic charring of coatings. The interpenetrating network structure and BPO₄ promoted the generation of a continuous and robust siliceous layer that acted as a barrier. However, the complex reaction mechanism among ZnPA, BA, MEL, and geopolymer, as well as the smoke suppression performance of the composite coating, will be investigated in our subsequent research.

5. Conclusions

In this study, the flame retardancy of ZnPA/BA/MEL-containing silica fume-based geopolymer coatings was preliminarily investigated. The following conclusions were drawn:

(1) An appropriate amount of ZnPA (2.5 wt.%) and BA (1.5 wt.%) enhanced the flame retardancy of silica fume-based geopolymer coating, evidenced by the FPI increasing from 0.59 (P0) to 2.83 (B3) s·m⁻²·kW⁻¹, the FGI decreasing from 0.86 (P0) to 0.27 (B3) kW m⁻²·s⁻¹, and the FRI of 8.48.

(2) The in situ-formed boron phosphate (BPO₄) derived from the doped ZnPA and BA facilitated interpenetrating networks within siliceous gels, improving the residual resilience of the fire-barrier layer, which could shield, fill, and adsorb heat during firing.

(3) According to pyrolysis kinetic calculation, in situ-formed BPO₄ made the E_α climb from 94.28 (P5) to 127.08 kJ·mol⁻¹ (B3) with temperatures of 731–940 °C, corresponding to improved flame retardancy.