Preparation and Thermal Safety of Ammonium Polyphosphate Doping Silica Aerogels: Effects of Content and Polymerization

Abstract

Hydrophobic silica aerogels (SAs) have attracted much attention because of their excellent thermal insulation performance and have potential applications in energy conservation and emission reduction. However, the organic groups on its surface are flammable, which brings security risks and limits its application scope. In this study, two kinds of ammonium polyphosphate (APP) with different polymerization degrees, namely low-polymerization-degree APP (LAPP) and high-polymerization-degree APP (HAPP), were introduced into SA to prepare APP/SA composites, to improve the thermal safety of the materials. The results showed that APP with two polymerization degrees significantly delayed the initial decomposition and peak temperature of heat flow, and HAPP reduced the gross calorific value by 31.01% at most, which is 29.04% greater than that of LAPP, indicating that the effect of HAPP was slightly better than that of LAPP. With the increase in APP with two polymerization degrees, the density increased and the porosity decreased: LAPP system was 0.095–0.196 g/cm³ and 96.0–91.0%. Both made the thermal conductivity increase only slightly (up to 26.8 mW/m/K), but the sample still maintained excellent thermal insulation and hydrophobicity, which indicated that the addition of APP improved the thermal safety performance of SA while maintaining its basic excellent performance. This strategy provides an effective and simple way to improve the flame retardancy of SA, which makes SA more widely used in fields with strict requirements on thermal safety.

1. Introduction

Hydrophobic silica aerogels (SAs) are lightweight, highly porous solids composed of interconnected silica nanoparticles with nanoscale pores [1,2,3]. Their continuous nanoporous network imparts very low density (typically 0.03–0.50 g/cm³) [4], ultralow thermal conductivity (approximately 13–21 mW m⁻¹ K⁻¹) [5,6], high specific surface area (500–1200 m²/g) [7,8,9,10], and large mesoporous volume, making them the most effective thermal insulation materials available at present and widely used in building insulation, transportation, aerospace, and other fields [11]. After hydrophobic modification with organic groups such as trimethylsilyl [12,13], these materials can operate reliably in humid environments over extended periods and retain their structural integrity, exhibiting minimal performance degradation and no noticeable collapse or cracking of the porous framework [7].

The organic groups introduced during hydrophobic modification also pose inherent fire hazards [8,14]. Thermally unstable surface groups such as Si-CH₃ readily decompose under high temperatures or intense heat flux, releasing flammable gases and generating toxic carbon monoxide [15,16,17,18,19]. Consequently, enhancing the thermal safety and flame retardancy of SA, while preserving their low density, low thermal conductivity, and excellent hydrophobicity, has become a critical challenge in this field.

To reduce the flammability of SA, various flame-retardant strategies have been developed. Among them, the most effective and commonly used method is to fill it with a fire retardant, to reduce the flammability of materials and enhance the thermal safety of materials in various application scenarios. Inorganic hydroxide is a typical halogen-free flame-retardant additive. Zhang et al. [8] prepared Mg(OH)₂/SA, which retained low density and thermal conductivity and reduced the gross calorific value (GCV) by 15%. For aluminum hydroxide as a flame-retardant additive, the research of Gu et al. [20] shows that the flame-retardant efficiency of aluminum hydroxide is generally lower than that of magnesium hydroxide. Carbon-based nanomaterials as flame retardant dopants also show great potential in this respect. Li et al. [21] incorporated reduced graphene oxide (rGO) into SA. The formed dense carbon layers increase their initial and peak decomposition temperatures. Compared with SA, the GCV decreases by 12.4% at most, and the initial thermal decomposition temperature increases by about 50%, while maintaining low density and thermal conductivity. Wu et al. [22] prepared phytic acid-modified graphene oxide (PA-GO)/SA composites. Adding 0.45 wt% PA-GO delayed the initial decomposition and peak temperature of heat flow, and the GCV decreased by 13.8%. Phosphorus-containing compounds are another important category of SA flame retardants. Li et al. [23] used bio-based phytic acid (PA) as a catalyst and flame retardant to prepare SA with low phosphorus content. The test results showed that GCV could be greatly reduced by 51.6% by adding only 5% PA, and the initial decomposition temperature and peak heat flow temperature were significantly improved compared with the first two flame retardant additives. Park et al. [24] incorporated 9,10-dihydro-9-oxa-10-phosphaphenanthrene-10-oxide (DOPO)-functionalized vinyl trimethoxysilane (DOPO-VTMS) into polymethylsilsesquioxane (PMSQ) aerogels, which achieved a significant reduction in peak heat release rate (pHRR) and total heat release capacity (THR), while maintaining basic material properties.

Compared with inorganic hydroxide and carbon-based nanomaterials in the SA system, a phosphorus-containing flame retardant is easier to significantly reduce the total heat release and increase the characteristic temperature of decomposition and heat flow at a lower dosage, which shows more advantageous flame-retardant potential. Based on this understanding, we will focus on ammonium polyphosphate (APP), a typical phosphorus-containing flame retardant. APP is a widely used halogen-free intumescent flame retardant [25,26,27], and its unique thermal decomposition behavior and charring characteristics provide a new opportunity to solve the above limitations. APP decomposes at high temperature, releasing polyphosphoric acid and ammonia. Polyphosphoric acid promotes the carbonization of the matrix to form a dense carbon layer, while ammonia gas in turn dilutes combustible gas, thus achieving a synergistic flame-retardant effect involving gas phase and condensed phase mechanism [28,29]. The thermal stability of APP depends largely on its degree of polymerization (DP) and crystal form: Compared with type I APP (DP = 30–100, hereafter denoted as LAPP), type II APP with high APP (DP > 1000, hereafter denoted as HAPP) has higher decomposition temperature and lower water solubility [30,31].

These studies [25,26,27,28,29,30,31] show that APP with excellent flame retardancy and an adjustable polymerization degree is a promising flame retardant. However, the effects of APP content and different polymerization degrees on the structure, basic physical and chemical properties, thermal oxidative decomposition behavior, and combustion energy release of SA have not been systematically clarified. In this study, ammonium polyphosphate with a low polymerization degree and a high polymerization degree was introduced into SA with different loading amounts to clarify the relationship between the structure/dosage of ammonium polyphosphate and the microstructure, basic physical and chemical properties, and thermal decomposition behavior of aerogel. It aims to provide a feasible, economical, and effective strategy for improving the thermal safety of SA and expanding its application in high-performance insulation systems.

2. Results and Discussion

2.1. Density and Porosity

As illustrated in Figure 1a,b, the density and porosity of SA varied systematically with the loading of LAPP or HAPP. For the LAPP series, the bulk density increased from 0.095 to 0.196 g/cm³, whereas the porosity decreased from 96.0% to 91.0%. For the HAPP series, the density increased from 0.125 to 0.206 g/cm³, the porosity decreased from 95.4% to 90.6%, and both parameters tended to level off at APP loadings above 15.38%. These trends arose from the higher density of APP and its pore-filling and structure-collapsing effects, consistent with the SEM observations.

The monotonic increase in density and the concomitant decrease in porosity were governed by a two-step mechanism: (i) geometric pore filling, which was more effective for LAPP because of its higher diffusivity into medium and small pores; and (ii) skeleton rearrangement coupled with neck thickening during drying, promoted by charging screening and capillary forces. The effective skeleton density also increased, further amplifying the apparent reduction in porosity. For the HAPP series, chain entanglement and early aggregation at high loadings hindered further penetration, leading to the gradual stabilization observed above 15.38%. Nevertheless, both series retained the characteristic low-density and high-porosity features of SA.

2.2. Pore Size Distribution

The pore structure characteristics of SA and APP/SA were further characterized by nitrogen adsorption−desorption isotherms, as shown in Figure 2a. The adsorption isotherms of all samples could be classified as type IV isotherms defined by IUPAC [32,33], which rose slowly in the middle relative pressure region, and H3 hysteresis loops appeared in the range of relative pressure (P/P_o) greater than 0.5, indicating that all samples were mainly composed of mesopores, and there were slit-like channels formed by particle accumulation [34]. Compared with SA, the overall morphology of the isotherm remained the same after the introduction of APP with different polymerization degrees and dosages, indicating that the incorporation of APP did not change the mesoporous nature of aerogels. As far as adsorption capacity was concerned, the nitrogen adsorption capacity of SA was the highest in the high relative pressure region. With the increase of LAPP content from 8.33% to 21.43%, the adsorption capacity at the end of isotherm gradually decreased, indicating that some pores were preferentially filled or blocked by LAPP, and the available pore volume decreased. In HAPP-doped samples, the adsorption capacity of HAPP-8.33%/SA with a low dosage was close to that of the LAPP sample. When the dosage was increased to 15.38%, the adsorption capacity dropped obviously and then rose to a certain extent at 21.43%, which was consistent with the change trend of BET specific surface area and pore volume, suggesting a tendency toward more open interparticle voids at high HAPP loadings. This effect may be associated with the aggregation or bridging of high-polymerization APP within the particle network rather than a definitive pore-network reconstruction.

The BJH pore size distribution curves of each sample are shown in Figure 2b, and the characteristic pore sizes of all samples fall within 2–50 nm, confirming their mesoporous nature. For pristine SA, the distribution is not strictly unimodal but exhibits a dominant peak centered at about 5.2 nm with a discernible shoulder/broad secondary feature, indicating the coexistence of two pore-size levels within the mesopore range. After LAPP incorporation, the bimodal-type characteristics were still observed, while the dominant peak gradually shifted toward larger pore sizes. Specifically, the most probable pore size increased from about 5.5 nm for LAPP-8.33%/SA to about 6.5–6.8 nm for LAPP-15.38%/SA and LAPP-21.43%/SA, accompanied by an evident broadening of the distribution. This evolution suggests that LAPP preferentially occupies part of the smaller mesopores and modifies the pore-size balance between the two pore populations, thereby shifting the dominant pore size and widening the overall distribution.

In contrast, the HAPP-doped samples show a distribution that is closer to a unimodal-dominant pattern, with the main peak located at larger pore sizes. The most probable pore sizes of HAPP-8.33%/SA, HAPP-15.38%/SA, and HAPP-21.43%/SA are about 10 nm, 18 nm, and 12–13 nm, respectively. Compared with the SA and LAPP series, the HAPP series presents a more pronounced dominant peak at larger pore sizes, suggesting that high-polymerization APP tends to form larger interparticle voids or pore domains through aggregation or bridging within the silica particle network. Despite these differences in distribution shape and peak position, the pore sizes of all samples remain mainly concentrated in the mesoporous range, indicating that APP incorporation alters the pore-size hierarchy but does not change the mesoporous nature of the SA framework.

The specific pore structure parameters of each sample are listed in

Table 1. Pore parameters of SA, LAPP/SA, and HAPP/SA.

Sample	BET Surface Area (m2/g)	Total Pore Volume a (cm3/g)	Average Pore Size b (nm)	Vpore c (cm3/g)	Dpore d (nm)
SA	882.5	2.6	8.8	8.7	39.5
LAPP-8.33%/SA	768.5	1.8	7.1	8.9	46.6
LAPP-15.38%/SA	735.5	2.5	9.7	6.5	35.2
LAPP-21.43%/SA	665.2	2.2	9.6	4.7	27.9
HAPP-8.33%/SA	689.2	2.2	10.0	5.5	31.8
HAPP-15.38%/SA	584.9	2.0	10.2	4.6	31.1
HAPP-21.43%/SA	681.2	2.8	12.3	4.4	25.7

^a,^b: The pore volume and average pore diameter were calculated from nitrogen desorption; ^c,^d: The V_pore and D_pore were obtained according to Equations (4) and (5) with a 10% relative uncertainty [35].

. The BET specific surface area of SA was as high as 882.5 m²/g, and the total pore volume and average pore diameter were 2.6 cm³/g and 8.8 nm, respectively, which reflected the typical high specific surface area and porous skeleton. After APP doping, the BET specific surface area of all composite aerogels was lower than that of SA. With the increase in APP doping, the specific surface area of LAPP series gradually decreased from 768.5 m²/g to 665.2 m²/g; HAPP series varied between 689.2 and 584.9 m²/g, indicating that the presence of APP partially occupied or blocked the original pores, resulting in the decrease in the exposed specific surface area. In terms of pore volume and average pore diameter, the total pore volume and average pore diameter of LAPP-15.38%/SA reached the local maximum of this series, about 2.5 cm³/g and 9.7 nm, respectively, while the total pore volume of HAPP-21.43%/SA was the highest, and the average pore diameter also increased to 12.3 nm, showing a trend of “slightly decreasing at first and then significantly increasing” with the increase in HAPP content. In addition, the theoretical pore volume V_pore and the average pore diameter D_pore calculated by Formulas (4) and (5) also showed a downward trend with the overall content of APP; for example, V_pore decreased from 8.7 cm³/g of SA to about 4.4 cm³/g of high-content samples, and D_pore also decreased from 39.5 nm to about 25.7 nm, which further proved that excessive APP cause partial skeleton shrinkage and densification. Appropriate APP doping regulated the pore structure by filling pores and generating secondary mesopores between particles. When APP content was too high, its agglomeration and skeleton collapse effect dominated, thus significantly reducing the specific surface area and altering the pore size distribution, which was consistent with the phenomenon of skeleton coarsening and agglomeration observed by subsequent SEM.

2.3. Microstructures

As shown in Figure 3a, SA exhibits a typical three-dimensional network composed of interconnected spherical particles, forming a continuous bead-like mesoporous framework [36]. This morphology is consistent with the BJH pore size distribution in Figure 2b, where SA presents a dominant mesopore population in the small-pore region, as well as with the high porosity reported in Table 1.

With the incorporation of LAPP (Figure 3b–d), the particle packing becomes progressively more compact as the LAPP content increases, accompanied by an apparent reduction in the open pore features between adjacent particles. In parallel, the BJH pore size distribution of the LAPP series in Figure 2b shows bimodal/shoulder-like characteristics rather than a strictly unimodal pattern, indicating a coexistence of pore populations with different characteristic sizes. This pore-structure evolution agrees with the pore-related parameters in Table 1, where the pore-structure decreases and the pore structure descriptors vary accordingly. These observations suggest that LAPP, owing to its relatively smaller molecular size and higher mobility, can enter the silica framework more effectively and partially occupy or narrow a fraction of pores, leading to local densification of the aerogel network [37].

Figure 3e–h presents the morphology of the HAPP-doped samples. Compared with the LAPP system, HAPP-doped samples show more pronounced particle aggregation and less uniform pore morphology, and molten-like agglomerates appear locally with increasing HAPP content. Consistently, the BJH pore size distribution of the HAPP series in Figure 2b is closer to a unimodal-dominant pattern with an overall shift toward larger characteristic pore sizes, which matches the SEM-observed coarsened interparticle regions and aggregation features. Meanwhile, the corresponding porosity and pore-structure parameters in Table 1 support the overall trend that HAPP induces stronger structural rearrangement and localized aggregation while maintaining the mesoporous nature of the framework.

Figure 3d–f shows that pristine LAPP exhibits a smooth bulk morphology, whereas Figure 3g–i indicates that HAPP has a denser, layered particle structure, confirming the distinct packing characteristics associated with different polymerization degrees [30]. Overall, both LAPP/SA and HAPP/SA retain the inherent three-dimensional porous framework of SA. However, the combined evidence from SEM morphology, pore size distribution, and porosity-related parameters demonstrates that increasing APP content promotes pore-level evolution through pore filling, densification, and local aggregation, while preserving the structural integrity of the aerogel skeleton [8].

Figure 3j,k compares the EDS spectra of APP/SA prepared with LAPP in Figure 3j and HAPP in Figure 3k. In both cases, a distinct P signal was detected together with the characteristic Si signal from the silica framework, confirming the presence of APP-derived phosphorus within the aerogel matrix. The main peaks in the revised spectra were assigned to C Kα (~0.28 keV), O Kα (~0.52 keV), Si Kα (~1.74 keV), and P Kα (~2.01 keV). The appearance of phosphorus in both spectra, while the silica signature remained unchanged, indicated that APP was effectively incorporated without disrupting the Si-based backbone. The comparable P response across the two APP types at the tested condition further suggested that the doping process was robust with respect to APP polymerization degree. These compositional results were consistent with the SEM-observed morphology and with the trends in properties, jointly supporting the successful introduction of APP into SA.

2.4. Thermal Insulation

As shown in Figure 4a, the thermal conductivity of the LAPP series increased slightly but steadily with APP loading, from 22.8 mW m⁻¹ K⁻¹ for SA to 25.4 mW m⁻¹ K⁻¹ at 21.43%. Figure 4b shows a more pronounced increase for the HAPP series, where the value rose from 23.0 mW m⁻¹ K⁻¹ to 26.8 mW m⁻¹ K⁻¹ at 21.43%, approaching the thermal conductivity of air. These results indicated that APP incorporation enhanced the overall heat-transfer capability of the composites, with the effect becoming more significant at higher APP loadings and polymerization degrees.

The increase in thermal conductivity arose from the combined effects of pore filling, skeleton densification, and the formation of continuous solid-phase conduction paths. As APP particles were introduced into the SA network, they occupied part of the pore volume and promoted closer contact between silica particles, leading to a gradual decrease in porosity and improved connectivity within the solid framework. This densification allowed heat to be transmitted more efficiently through the condensed phase. At the same time, aggregation and neck thickening of silica particles further reinforced the conductive network, particularly in the HAPP-containing composites. The longer molecular chains and higher viscosity of HAPP facilitated the formation of solid bridges during drying, enhancing the continuity of thermal pathways. Although APP addition slightly increased thermal conductivity, all samples retained low absolute values, thereby preserving the intrinsic thermal-insulation performance of SA.

Overall, the increase in thermal conductivity with APP loading reflected a trade-off between improved structural integrity and reduced porosity. HAPP exerted a stronger influence because of its denser molecular structure and greater tendency to form aggregated conductive networks. These results suggested that careful control of both APP polymerization degree and loading was essential to balance flame-retardant performance and thermal-insulation efficiency in APP/SA.

2.5. Surface Chemistry and Hydrophobicity

Figure 5 shows the infrared spectra of SA and composite aerogels doped with different polymerization degrees and different amounts of APP to characterize the composition of surface functional groups and the introduction of APP. The main absorption peaks of each sample were the same, but with the change in APP doping amount and polymerization degree, the intensity of some characteristic peaks changed regularly.

Firstly, in the C-H region of 2880–2980 cm⁻¹, there were obvious absorption bands in each curve, corresponding to the symmetric and antisymmetric stretching vibration and shear bending vibration of the C-H bond in the surface-CH group [38]. SA and APP/SA maintained strong absorption in this region, and the peak position did not shift, which indicated that the organic hydrophobic groups represented by Si-CH remained stable after being doped with APP, which was the chemical basis for the material to maintain high hydrophobicity. In general, no obvious peak shift was observed in this region among different samples, suggesting that the Si–CH-related hydrophobic groups were retained after APP incorporation; the minor intensity differences are mainly attributed to the semi-quantitative nature of FTIR rather than an abnormal change in Si–CH content.

At 1630 cm⁻¹, all samples showed a weak absorption peak, which was attributed to the O-H bending vibration of adsorbed water or a small amount of residual hydroxyl [39]. Compared with SA, the peak intensity of APP was slightly increased, especially in HAPP-15.38%/SA and HAPP-21.43%/SA, which indicated that phosphorus-containing APP molecules were hydrophilic and adsorbed a little water on the skeleton surface, but the overall peak intensity remained weak, indicating that the overall hydrophobicity of composite aerogels was still good.

At 1260 cm⁻¹, each sample exhibited a clear absorption peak, which was attributed to the bending vibration of the Si-C bond [40]. The position of the peak was basically unchanged before and after doping, but there was a slight difference in intensity, showing that the introduction of APP did not destroy the Si-C bond structure, and the silicone group on the SA surface still existed stably after recombination, which confirmed the characteristics of the composite material to maintain a hydrophobic surface from another side.

In the range of 900–1000 cm⁻¹, obvious absorption peaks were observed in LAPP-8.33%/SA and above, while there was almost no characteristic absorption at the corresponding position of SA, which were attributed to the vibration of the P-O-P bond in polyphosphate [41]. With the increase in APP content, the peak intensity gradually increased, especially in HAPP-15.38%/SA and HAPP-21.43%/SA, indicating that the enrichment degree of phosphorus-containing groups in aerogel skeleton increased with the increase in APP content, which proved that APP was successfully introduced into SA porous network. At the lowest wavenumber region of 450 cm⁻¹, all samples showed obvious absorption peaks, corresponding to the bending vibration of Si-O-Si bonds [42]. The position of the peak was basically the same in different samples, indicating that the doping process of both LAPP and HAPP did not destroy the basic structure of silica skeleton, and the three-dimensional network of SA remained complete and continuous after recombination. Based on the analysis and comparison of the above-mentioned several iconic absorption peaks, APP doping, on the one hand, introduced and enhanced the phosphorus-containing characteristic peaks such as P-O-P in the infrared spectrum, which verifies the successful loading of APP in SA skeleton. On the other hand, the positions of vibration peaks of key skeletons such as Si-C and Si-O-Si were not changed, which shows that the network structure of SA hydrophobic groups on the surface were preserved, so the composite aerogel still maintained good hydrophobic characteristics while obtaining phosphorus-containing flame retardant components.

The water contact angles of the SA and its composites doped with LAPP or HAPP at different loadings are presented in Figure 6. Pristine SA exhibited excellent hydrophobicity with a static water contact angle of 144.1°. Upon incorporation of LAPP, the contact angles of the LAPP/SA slightly decreased to 133.8°, 135.8°, and 137.3° for doping amounts of 8.33%, 15.38%, and 21.43%, respectively. A similar trend was observed for the HAPP/SA, where the contact angles were 140.6°, 138.1°, and 139.9° corresponding to HAPP loadings of 8.33%, 15.38%, and 21.43%, respectively. Overall, all doped-samples maintained contact angles above 133°, indicating that the composites still possessed strong hydrophobic characteristics, with water droplets adopting a near-spherical shape on the surfaces.

It was well established that surface hydrophobicity was primarily governed by chemical composition and surface roughness [43]. Since the contact angle measurements were performed on compressed powder pellets under identical conditions, the influence of macroscopic roughness could be considered comparable across all samples. The slight reduction in contact angle upon APP incorporation was therefore mainly attributed to the introduction of hydrophilic phosphate and ammonium groups from both LAPP and HAPP, which partially replaced or shielded the hydrophobic siloxane and alkyl groups derived from TEOS-based modification [44]. Notably, at equivalent doping levels, the HAPP/SA consistently exhibited higher contact angles than their LAPP/SA counterparts, suggesting that the HAPP exerted a milder impact on surface hydrophilicity, possibly due to its lower density of terminal hydrophilic moieties per unit mass [45]. Nevertheless, the retention of contact angles well above 130° across the entire doping range confirmed that the hydrophobic nature of the SA matrix is largely preserved even after APP incorporation up to 21.43%.

2.6. Thermal Safety

As shown in Figure 7g, thermal oxidative decomposition of SA occurred earlier in air atmosphere, with an initial temperature of 247.9 °C and an exothermic peak temperature of 267.7 °C, and its decomposition was mainly due to thermal oxidative degradation of Si-CH₃ groups [15]. After adding APP, the initial decomposition temperature and exothermic peak temperature of the sample increased significantly, which indicated that APP could effectively delay the thermal oxidation process of SA and improve its thermal stability. For the addition of LAPP, 8.33% LAPP/SA in Figure 7a has two adjacent exothermic peaks, the initial temperature of the first stage is 425.9 °C with a peak temperature of 433.3 °C, and the initial temperature of the second stage is 446.9 °C with a peak temperature of 455.3 °C. In Figure 7b, With the LAPP content increasing to 15.38%, the two stages continued to move to high temperature, with the initial temperature and peak temperature increasing to 448.6 °C and 455.5 °C in the first stage, respectively, and 473.9 °C and 488.2 °C in the second stage, respectively. As shown in Figure 7c, when it is further increased to 21.43%, the initial temperature of the main exothermic process rises to 465.3 °C, and the peak temperature rises to 474.5 °C. At the same time, the exothermic process is relatively more concentrated, indicating that the higher content of LAPP can continuously delay the temperature range of the main thermal oxidation reaction. [46].

The change in the HAPP system is shown in Figure 7d–f. The initial temperature of 8.33% HAPP/SA is 396.3 °C and the peak temperature is 407.6 °C. When the content of HAPP increases to 15.38%, the initial temperature and peak temperature increase to 409.6 °C and 422.4 °C, respectively. It is worth noting that when the content of HAPP reaches 21.43%, the improvement in thermal stability is more significantly enhanced, the initial temperature is increased to 496.9 °C, and the peak temperature is increased to 516.1 °C, indicating that high content of HAPP has a stronger delaying effect on thermal oxidative decomposition in a high-temperature region.

Comparing the effects of the two APPs with the same dosage, it can be found that when the dosage is 8.33% and 15.38%, the decomposition starting temperature and exothermic peak temperature of LAPP/SA are higher than those of the corresponding HAPP/SA, which shows that LAPP is more effective in improving thermal stability at low to medium dosages. However, when the dosage is 21.43%, the initial temperature and peak temperature of HAPP/SA are obviously higher than those of LAPP/SA, which indicates that the stabilizing effect of HAPP is more dominant under the condition of high dosage. Overall, LAPP is more conducive to rapidly raising the decomposition temperature range at a low dosage, while HAPP can further postpone the main thermal oxidation process to a higher temperature range at a high dosage. This difference reflects that the polymerization degree and dosage jointly determine the action range and strength of APP to improve the thermal stability of SA.

2.7. Gross Calorific Value

As shown on Figure 8a, the GCV of pure SA was 12.47 MJ/kg. With the addition of ammonium polyphosphate, the GCV decreased continuously, indicating that ammonium polyphosphate effectively weakened the exothermic potential of the system. For the LAPP/SA system, the GCV was 11.32 MJ/kg at a dosage of 8.33%, corresponding to an absolute decrease of 1.15 MJ/kg and a reduction rate of 11.53%. At a dosage of 15.38%, the GCV was 10.57 MJ/kg, the absolute reduction was 1.89 MJ/kg, and the reduction rate was 15.2%. At a dosage of 21.43%, the GCV was 8.85 MJ/kg, the absolute reduction was 3.62 MJ/kg, and the reduction rate was 29.04%. For the HAPP/SA system, the GCV values were 10.87, 10.14, and 8.60 MJ/kg at the same dosages, and the corresponding absolute reductions were 1.59, 2.33, and 3.86 MJ/kg, with reduction rates of 12.8%, 18.71%, and 31.01%, respectively. At the same dosage, the GCV of HAPP/SA was consistently lower than that of LAPP/SA, and the reduction rate of HAPP/SA reached 31.01% at 21.43%, which was higher than that of LAPP/SA at 29.04%, indicating that increasing the polymerization degree of ammonium polyphosphate enhanced the reduction in the total calorific value of SA.

Figure 8b further reveals the source of GCV reduction. The red dotted line indicated the theoretical reduction rate was caused only by mass dilution, and its values were consistent with the addition amounts, namely 8.33%, 15.38%, and 21.43%. When the experimental reduction rate was higher than the dotted line, it showed that ammonium polyphosphate not only contributed through dilution by replacing combustible components but also introduced additional physical and chemical effects. By comparison, the reduction rates of HAPP/SA at the three dosages were 12.8%, 18.71%, and 31.01%, respectively, all of which were higher than the respective dilution line by 4.47, 3.33, and 9.58 percentage points, indicating that HAPP provided a clear non-dilution contribution to GCV reduction over the entire dosage range. The reduction rates of LAPP/SA at 8.33% and 21.43% were 11.53% and 29.04%, respectively, which were 3.20 and 7.61 percentage points higher than the dilution line, respectively, indicating that significant non-dilution contributions also existed at low and high addition levels. At 15.38%, the reduction rate was 15.2%, which was essentially consistent with the dilution line of 15.38%, indicating that the GCV reduction was mainly dominated by mass dilution, whereas the additional physical and chemical effects were not prominent.

Overall, the reduction in the GCV of SA by APP did not simply rely on the mass dilution effect of “reducing SA content”. In particular, at an addition level of 21.43%, the excess of the experimental reduction rate relative to the dilution line was significantly enlarged, and the excess for HAPP was higher than that for LAPP, indicating that ammonium polyphosphate with a higher polymerization degree was more likely to form a stable polyphosphate condensed phase during heating and to become immobilized on the silica skeleton. This process promoted condensed-phase transformation and residue formation via acidic species and enabled spreading over the pore−wall surface to form a barrier layer that inhibited the release of combustible volatiles and the diffusion of oxygen, thereby further reducing the effective heat-release capacity on a mass basis. Such additional contributions were usually associated with enhanced condensed-phase processes, including promoting solid-phase residue formation and structural densification, weakening the formation and release of combustible volatiles and increasing energy consumption during thermal decomposition, such that the total calorific value of the unit-mass system was reduced beyond the theoretical prediction based on “non-combustible filler replacement”.

3. Conclusions

We fabricated APP-modified SA composites by incorporating two APP grades with distinct polymerization degrees into the SA matrix, and the work primarily aimed at enhancing the thermal safety of the resulting aerogels. The results showed that the microstructure of APP/SA was almost unchanged compared with that of pure SA, and it maintains excellent properties of low density, low thermal conductivity, and hydrophobicity. It was found that the introduction of APP effectively delayed the thermal oxidative decomposition of methyl silicon groups on SA, thus improving the thermal stability of APP/SA. It was further found that the thermal safety performance of HAPP was better than that of LAPP. However, LAPP was beneficial to fill holes and consider thermal insulation and structural integrity, while polymeric APP significantly improved thermal stability and combustion energy suppression at the expense of a small amount of thermal insulation performance, which is a more radical choice to improve thermal safety. This work provides a simple method to improve the thermal safety of SA, which has great potential in more fields of thermal insulation application.

4. Materials and Methods

4.1. Raw Materials

Tetraethyl orthosilicate (TEOS), ethanol (EtOH, 99.7%), and n-hexane (97.0%) were purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). APP (PD = 30–100 or >1000) was purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). Nitric acid (HNO₃, 36–38%) and ammonia water (NH₃·H₂O, 25–28%) were used as the acid and basic catalysts, respectively. Trimethyl Chlorosilane (TMCS, 99.0%) used in the experiments were purchased from Aladdin (Shanghai, China). Deionized water was made by an ultra-pure water machine (Direct-Q 3UV, Merck Millipore, Burlington, MA, USA).

4.2. Preparation of APP/SA Composites

Figure 9 schematically illustrates the preparation procedure of APP/SA composites. In a typical batch, TEOS (5.75 mL) was mixed with ethanol (15.0 mL) and deionized water (1.0 mL) (ethanol/water volume ratio = 15:1) under magnetic stirring. The pH of the mixture was adjusted to 2–3 by adding 0.1 mol L⁻¹ HNO₃ (0.30 mL), and the resulting sol was hydrolyzed in a 45 °C water bath for 12 h.

After hydrolysis, ammonium polyphosphate (APP) with different polymerization degrees (LAPP or HAPP) was introduced into the hydrolyzed sol at predetermined loadings of 0.2, 0.4, or 0.6 g per batch, followed by ultrasonic-assisted stirring for 30 min to ensure homogeneous dispersion. Subsequently, the mixture was transferred to magnetic stirring, and 0.5 mol L⁻¹ aqueous ammonia (NH₃·H₂O, 0.50 mL) was slowly added dropwise, followed by further stirring for 3 min. Gelation typically occurred within 20 min. The obtained wet gel was then kept static at room temperature for 3–4 h.

To remove residual water and strengthen the gel network, the wet gel was subjected to solvent exchange using ethanol and n-hexane. First, the supernatant (containing excess water) was decanted, and ethanol was added to fully immerse the gel (45 mL per batch). The gel was cut into small pieces and aged in ethanol at 45 °C for 18 h, during which the ethanol was replaced once after 9 h. During each ethanol replacement, the gel pieces were further cut into smaller fragments to facilitate solvent diffusion and water extraction.

Next, ethanol was replaced by n-hexane (45 mL per batch) for solvent exchange at 45 °C for 18 h, and the n-hexane was replaced once after 9 h. After solvent exchange, surface silylation was performed using a silylating agent in n-hexane. Specifically, the gel was immersed in TMCS/ n-hexane by adding TMCS (6.0 mL) into n-hexane (44 mL), and the modification was conducted at 45 °C for 24 h.

After modification, the gel was decanted and washed with fresh n-hexane twice (2 × 45 mL) to remove residual reagents. Finally, the samples were dried at 120 °C for 4 h to obtain APP/SA composites, denoted as XAPP-n/SA, where X represents the polymerization degree of APP (LAPP or HAPP) and n represents the APP loading per batch (8.33%, 15.38%, or 21.43%).

4.3. Methods of Characterization

The microstructure was observed using a field-emission scanning electron microscope (SEM, ZEISS Sigma 300, Oberkochen, Germany). The elemental composition and content were analyzed via an energy-dispersive X-ray spectroscopy (EDS, EDX-720, Shimadzu Corporation, Kyoto, Japan). The tap density (ρ_t) of pure SA and APP/SA composites was measured with a tap density tester (ZS-202, Liaoning Instrument Research Institute, Shenyang, China), using a 10 mL graduated cylinder vibrated continuously at 300 rpm for 10 min. The porosity of the APP/SA composites was determined using Equations (1)–(3):

$$Porosity=(1-{\displaystyle \frac{{\rho }_t}{{\rho }_s}})\times 100\%$$

(1)

$${\displaystyle \frac{1}{{\rho }_s}}={\displaystyle \sum _{i=1}^n\frac{{\omega }_i}{{\rho }_i}}$$

(2)

$${\omega }_{APP}={\displaystyle \frac{m_{APP}}{m_{APP}+m_{SA}}}$$

(3)

where ρ_s represents the skeletal density of the APP/SA composites, and ω_i and ρ_i denote the mass fraction and skeletal density of each constituent, respectively. It should be noted that the tap density (ρ_t) measured by the tap density tester is only used to characterize the packing/stacking state of the powder, and it is not used in Equations (1)–(3). In Equations (1)–(3), ρ_i refers exclusively to the skeletal (true) density of each component. The density of APP is reported to be approximately 1.9 g/cm³, while that of SA is 2.2 g/cm³ according to the literature [47,48]. The mass fraction of APP, expressed as ω_APP, is determined by Equation (3). Here, m_SA (around 2.2 g) refers to the mass of pure SA, which was obtained from repeated experiments conducted under identical conditions without the addition of APP.

The nitrogen adsorption−desorption isotherms were measured using an automatic surface area and porosity analyzer (Quanta chrome, AUTOSORB IQ, Boynton Beach, FL, USA) at 77 K (liquid nitrogen temperature). The Brunauer−Emmett−Teller (BET) method [49] was employed to calculate the specific surface area, and the Barrett−Joyner−Halenda (BJH) [50] method was used to determine the pore size distribution. However, during the nitrogen adsorption process, the skeletons of SA may undergo deformation, which could affect the accuracy of the test results [51]. To obtain more precise pore structure data, the specific pore volume (V_pore, cm³/g) and average pore diameter (D_pore) were calculated according to Equations (4) and (5) [1]:

$$V_{pore}={\displaystyle \frac{1}{{\rho }_t}}-{\displaystyle \frac{1}{{\rho }_s}}$$

(4)

$$D_{pore}={\displaystyle \frac{4V_{pore}}{S_{BET}}}$$

(5)

where ρ_t, ρ_s, and S_BET refer to the tap density, the skeletal density, and the BET surface area, respectively.

The thermal conductivity of the composites was measured at room temperature by the transient hot-wire method using a constant thermal conductivity analyzer (TC3000E, XIATECH, Xi’an, China). To reduce the effects of particle-size distribution and powder compaction, all samples were tested in powder form under identical packing conditions: the same sample cell was used, powders were filled to the same volume and leveled to obtain a comparable packing state, and the probe was fully embedded in the packed powder bed. Each sample was measured at least three times, and the reported value is the average.

The chemical group was confirmed by the Fourier-transform infrared spectroscopy (FTIR, Nicolet 8700, Nicolet, Madison, WI, USA).

The sample powder was evenly spread on a piece of tape and subsequently pressed and flattened to construct a test sample with a smooth surface. A contact angle meter (Model JC2000D1, Shanghai Zhongchen Instruments, Shanghai, China) was used to conduct the hydrophobicity characterization test. During the operation, a 5 μL water droplet was dispensed onto the smooth surface of the powder sample. After the water droplet shape stabilized, ImageJ image processing software (version:1.54f) was used to analyze the water droplet images, and then, the contact angle value of the sample was calculated to complete the hydrophobicity characterization.

The thermal stability analysis was assessed using TG-DSC (SDT Q650, TA Instrument, New Castle, NH, USA). The GCVs of APP/SA were measured by using an oxygen bomb calorimeter (Yuan Fa AM-C1009, Yuanfa Instrument, Changsha, China).