Synthesis and Performance Characteristics of Filling-Grouting Composite Cement with Flame Retardant and Support Properties

Abstract

To prevent the spontaneous combustion of coal caused by heat accumulation from exothermic oxidation in the goaf, a water-absorbing gel was prepared using sodium carboxymethyl cellulose. The gel was characterized by Fourier transform infrared (FTIR) spectroscopy, X-ray diffraction (XRD), nuclear magnetic resonance (NMR), and atomic force microscopy (AFM) to analyze its functional groups, crystal structure, and surface morphology. The synthesized gel was then integrated into a cement-sand mixture along with nanosilica, bentonite, and boric acid to fabricate a novel composite cement that exhibits both flame-retardant and support capabilities. Key parameters of the resulting material, including its thermal stability, flame retardancy, and compressive strength, were systematically evaluated. Furthermore, molecular dynamics simulations were employed to elucidate the underlying flame-retardant mechanism.

1. Introduction

Goaf collapse and coal spontaneous combustion pose severe challenges to the safety and environmental sustainability of underground coal mining. The residual coal and gangue left in the goaf are exposed to air, undergoing slow oxidation. This process leads to the accumulation of heat and flammable gases, which can readily cause gas poisoning, suffocation, and even fire or explosion hazards. − Concurrently, the absence of adequate support in the goaf allows the overlying strata to subside continuously. This subsidence propagates upward, resulting in surface subsidence that forms sinkholes, thereby damaging the surface ecology, infrastructure, and buildings. −

In the realm of preventing and extinguishing coal spontaneous combustion in goafs, Jueraiti et al. addressed the issue of retardants loss by developing a high-viscosity composite material using molasses solution, a sugar industry waste, along with CaCl₂ and MgCl₂. Nitrogen injection is a common technique for goaf fire prevention. In this context, Gao et al. proposed a dual-period dynamic alternating injection strategy, which was demonstrated to enhance oxygen concentration dilution and cooling efficiency compared to traditional alternating methods. In another study, Gao et al. fabricated an organic hydrogel using sodium alginate and gelatin, which acts by forming an oxygen-isolating cover while utilizing the moisture within the gel for cooling. Xu et al. employed a surfactant to create a porous cement that achieves fire suppression through covering effect. Zheng et al. incorporated nanosilica and a foaming agent into a hydrogel to produce a gel foam, subsequently evaluating its effectiveness in cooling and suppressing the low-temperature oxidation of coal.

Regarding research on goaf backfilling to prevent strata subsidence, Wu et al. investigated a method combining waste rock filling with cemented roof filling and evaluated its performance. Wang et al. proposed a cementitious material prepared from coal-based solid wastes mineralized with CO₂, which serves the dual purpose of providing goaf support and sequestering carbon dioxide. Su et al. developed a backfill material using coal gangue, slag, and municipal solid waste incineration fly ash, and further assessed the heavy metal leaching rate of this waste-based material to ensure it poses no secondary environmental contamination. In a related study, Wu et al. incorporated gold mine overflow tailings and fly ash into cement for goaf backfilling, reporting that the compressive strength of the resulting waste-based cement could exceed 7 MPa.

In the field of multifunctional backfill materials for coal mine goafs, Jiang et al. developed a foamed cemented tailings backfill using chemical and mineral foaming agents. The resulting material achieves expansion support while also demonstrating excellent thermal insulation performance. Li et al. proposed a calcium aluminate cement-fly ash backfill material for underground coal gasification, demonstrating that the material maintains a compressive strength of approximately 5 MPa even after exposure to 1000 °C, attributed to secondary hydration and the formation of thermally stable phases such as C-A–S–H gel and AlO­(OH), thus offering both high-temperature resistance and load-bearing capacity. Liu et al. developed a modified magnesium–aluminum-based solid waste CO₂ backfill material with a maximum compressive strength of approximately 5 MPa and a carbon sequestration efficiency of 6.8%, achieving the dual functions of mechanical support and CO₂ fixation. These studies highlight a clear trend toward designing backfill materials that synergistically combine structural support with thermal management, fire resistance, or carbon storage for more sustainable and safer underground mining operations.

Current research on goaf filling materials has predominantly focused on single-function systems. For those materials that already satisfy the necessary functional requirements, reported studies have typically aimed either to further enhance the primary function or to add environmental attributes. However, studies on backfill materials that integrate both coal spontaneous combustion prevention and strata support functionalities are relatively rare. Furthermore, potential mutual interference between fire-prevention materials and support materials in practical applications remains unclear. To address this gap, this study combines organic polymer gel with inorganic minerals and cement. Silane coupling agent is employed as a bridging agent to integrate these components, resulting in a novel, functionally composite cementitious backfill material that exhibits both excellent flame retardancy and high compressive strength.

2. Experimental Methods

2.1. Materials

Silicate cement, sand, sodium carboxymethyl cellulose (CMC), potassium persulfate (KPS), 2-acrylamide-2-methylpropylsulfonic acid (AMPS), N,N′-methylene bis­(acrylamide) (MBA), 3-Methacryloxypropyltrimethoxysilane (KH570), nanosilica, boric acid, and bentonite were purchased from Tianjin Kemie Chemical Reagents Co., Ltd.; The Silicate cement, sand, nanosilica, and bentonite are technical grade, and other substances are AR).

2.2. Preparation Process

2.2.1. CMC-co-AMPS

Five g of CMC was dissolved in 300 mL of deionized water. 1.0 mol % of the initiator KPS was added. The mixture was mechanically stirred at 60 °C in a water bath for 10 min to generate CMC radicals. Subsequently, 1.5 g of the monomer AMPS and 0.8 mol % of the cross-linker MBA were introduced into the reaction system. The reaction was allowed to proceed under continuous mechanical stirring at a constant temperature of 60 °C for 1 h, yielding a CMC-co-AMPS hydrogel. The resulting hydrogel was dried to a constant weight at 80 °C in a vacuum drying oven, then ground into powder for subsequent use.

2.2.2. CMC-co-AMPS Cement

Cement and sand were mixed at a mass ratio of 1:3. Then, CMC-co-AMPS (6 wt % of total solids) and silane coupler KH570 (0.05 wt % of total solids) were added to the mixture. An appropriate amount of water was introduced, and the blend was mechanically stirred to obtain the CMC-co-AMPS modified cement composite (CMC-co-AMPS cement).

2.2.3. CBBS Cement

Cement and sand were first mixed at a mass ratio of 1:3. CMC-co-AMPS (6.00 wt % of total solids), nanosilica (2.00 wt % of total solids), bentonite (5.00 wt % of total solids), boric acid (1.00 wt % of total solids) and silane coupler KH570 (0.05 wt % of total solids) were added. An appropriate amount of water was then introduced, and the blend was mechanically stirred until a homogeneous mixture was achieved, yielding the composite backfill grouting cement, designated as CBBS cement.

2.3. Single-Factor Experiment

To investigate the effects of monomer, initiator, and cross-linker concentrations on the properties of the CMC-co-AMPS material during its synthesis, a single-factor experiment was conducted. The water absorption rate of the fully dried product was used as the evaluation metric to identify the reactant formulation that yields the maximum absorption ratio. The sample was weighed every 2 h until its mass change was less than 2% over a consecutive 4 h period. Each group was tested five times, and the average results were recorded along with error analysis.

$$\mathrm{W}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{a}\mathrm{b}\mathrm{s}\mathrm{o}\mathrm{r}\mathrm{p}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{e}=\frac{m_2-m_1}{m_1}$$ 1

where m₁ is the weight of the fully dried CMC-co-AMPS, m₂ is the weight of the CMC-co-AMPS after complete water absorption.

2.4. Response Surface Methodology

In the preparation of the composite cement, the introduction of water-absorbing and flame-retardant materials such as CMC-co-AMPS, bentonite, and boric acid may affect the mechanical properties of the cement. For instance, boric acid can adsorb onto the surface of cement or its hydration products, or form coordination complexes that hinder hydration reactions; it is therefore often used as a retarder. However, excessive boric acid reduces the compressive strength of cement, and the optimal dosage is approximately 0.25–1.00 wt %. − Nanosilica, on the other hand, enhances mechanical properties through a synergistic hydration mechanism, although this enhancement also exhibits an upper limit of approximately 2.00 wt %. − Based on these findings from the literature, the dosages of boric acid, nanosilica and KH570 were set at 1.00, 2.00, and 0.05 wt %, respectively. The basic caracteristics of the bentonite as shown in Table .

1. Characteristics of the Bentonite Used in the Experiments.

characteristics	cation exchange capacity	swelling ratio	specific surface area	particle size pass through 200-mesh sieve
	90 meg·100 g–1	27 mL·2 g–1	450 m2·g–1	98%

To investigate the effects of CMC-co-AMPS and bentonite dosages on the properties of the composite cement, a response surface analysis using the Central Composite Design method was conducted. The compressive strength and water absorption rate of cement specimens cured for 28 days were used as the response indicators. The experiments factors as shown in Table .

2. Response Surface Experiments Factors.

name	units	low	high	-alpha	+alpha
CMC-co-AMPS	wt %	3	9	1.75736	10.2426
Bentonite	wt %	3	9	1.75736	10.2426

2.5. Characterization Experiments

To elucidate the reaction process of the CMC-co-AMPS synthesis and the microstructural characteristics of the resulting material, a comprehensive characterization was performed using Fourier-transform infrared (FTIR) spectroscopy, solid-state ¹H nuclear magnetic resonance (¹H NMR), X-ray diffraction (XRD) (scanning range: 5–80°, scanning speed: 4°·min^–1), atomic force microscopy (AFM), and scanning electron microscopy (SEM), which provided insights into the material’s functional groups, chemical bonding, crystal structure, and mesoscale morphology.

2.6. Performance Tests

2.6.1. Thermal Stability Test

To evaluate the thermal stability of the materials within the temperature range from coal oxidative self-heating to the onset of combustion, four distinct samples were prepared according to the method described in Section : (I) ordinary cement, (II) composite cement with CMC-co-AMPS and boric acid, (III) composite cement with CMC-co-AMPS, boric acid, and nanosilica, and (IV) the final CBBS cement. The pyrolysis characteristics of these cement composites were subsequently analyzed using simultaneous thermogravimetric analysis and differential scanning calorimetry (TGA-DSC). The testing temperature range is from room temperature to 650 °C, with a heating rate of 10 °C·min^–1. And according to the oxygen concentration range for coal spontaneous combustion in the goaf, a gas atmosphere of 10% oxygen balanced with nitrogen was employed.

2.6.2. Water Absorption Test

The water absorption capacity of the filling material is one of its key characteristics for achieving flame retardancy and fire extinguishing. Therefore, the water absorption ratio of the synthesized composite cement was tested, and the calculation method for the water absorption ratio is based on eq , where m₁ is the weight of the dried cement, m₂ refers to the weight of cement that has been absorbed water. The sample was weighed every 2 h until its mass change over a consecutive 4 h period fell below 2%, at which point the test sample was considered to have reached absorption saturation. This threshold was selected because actual engineering materials do not require excessively precise values, and also because experimental errors are generally within the 2% range. Under this criterion, all samples achieved water absorption saturation after 16 h of soaking, after which the mass remained largely stable. Each group was tested five times, and the average results were recorded along with error analysis.

2.6.3. Coal Oxidation Inhibition Test

The inhibition of temperature and carbon monoxide (CO) generation during the process of coal oxidation are key indicators for evaluating the flame-retardant performance of materials. Since the inhibitory effect of the composite cement on coal spontaneous combustion primarily originates from the CMC-co-AMPS and boric acid components, the retarding efficiencies of CMC-co-AMPS, boric acid, CBBS cement, and the composite gel were individually assessed. According to the mass ratio specified in Section , the specific composition ratios of the composite gel and the CBBS cement are presented in Table .

3. Composition Ratios of the Composite Gel and CBBS Cement for the Flame Retardant Test.

sample	CMC-co-AMPS	boric acid	cement	sand	bentonite	nanosilica	KH570
composite gel	86.6%	13.4%	-	-	-	-	-
CBBS cement	6.00%	1.00%	21.49%	64.46%	5.00%	2.00%	0.05%

Coking coal was crushed and sieved to obtain 80-mesh particles, and its proximate analysis is presented in Table . Subsequently, 50 g of the coal sample was placed in a beaker, 5 g of CMC-co-AMPS, boric acid, CBBS cement, and the composite gel was added respectively, and the blend was mechanically stirred until fully mixed. A thermocouple was embedded into the mixture, which was then dried to a constant weight at 30 °C under vacuum. After that, the prepared sample was placed in a programmed temperature chamber, and the temperature evolution along with CO generation was monitored during the heating process, which was conducted at a rate of 1 °C·min^–1 under an air flow of 80 mL·min^–1. Each group was tested five times, and the average results were recorded along with error analysis.

4. Proximate Analysis Data of the Coal Sample for Experiment.

sample	Mad (%)	Aad (%)	Vad (%)	FCad (%)
coking coal	0.88	11.63	31.58	55.91

2.6.4. Compressive Strength Test

The backfill material is subjected to vertical pressure from the overlying strata in the goaf under gravitational force. Consequently, the compressive strength of the material is a critical parameter reflecting its support performance. The prepared cement mixtures were poured into silicone molds and demolded after curing for 3, 7, 14, and 28 days respectively, to obtain cubic specimens with a side length of 7 cm. The cement specimens were cured under standard conditions: initially in a moist curing box at 20 °C and 95% relative humidity for 24 h, followed by demolding and further curing in a constant-temperature water tank at 20 °C. The maximum compressive strength of these specimens was then determined using a universal testing machine at a compression rate of 0.01 mm·s^–1. Each group was tested five times, and the average results were recorded along with error analysis.

2.7. Molecular Dynamics Simulation

To further elucidate the mechanism by which CMC-co-AMPS inhibits coal spontaneous combustion, molecular dynamics simulations were performed using Materials Studio software to model its influence on the interaction and diffusion of water on coal. Initially, molecular structure models for coking coal, water, and CMC-co-AMPS were constructed and geometrically optimized. Subsequently, a simulation system was built by placing five coking coal molecules, 210 water molecules, and two CMC-co-AMPS molecules within a rectangular cell (21 Å × 21 Å × 73 Å) featuring three-dimensional periodic boundary conditions. A dynamic relaxation was first conducted on the optimized model using the NPT ensemble, with a temperature of 298 K, a pressure of 0.1 MPa, a simulation duration of 300 ps, and a time step of 1.0 fs. Subsequently, the production simulations were conducted using the COMPASS II force field and the Nosé–Hoover thermostat under the NVT ensemble. The temperature was maintained at 298 K, with a time step of 1.0 fs and a total simulation duration of 300 ps.

The governing equations of the Force Field account for Coulomb interactions, van der Waals forces, and bond interactions. Specifically, the total potential energy is expressed as

$$E_{\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}}=E_{\mathrm{C}\mathrm{o}\mathrm{u}\mathrm{l}\mathrm{o}\mathrm{m}\mathrm{b}\mathrm{i}\mathrm{c}}+E_{Vdw}+E_{\mathrm{b}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{s}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{t}\mathrm{c}\mathrm{h}}+E_{\mathrm{a}\mathrm{n}\mathrm{g}\mathrm{l}\mathrm{e}\mathrm{b}\mathrm{e}\mathrm{n}\mathrm{d}}$$ 2

$$E_{\mathrm{C}\mathrm{o}\mathrm{u}\mathrm{l}\mathrm{o}\mathrm{m}\mathrm{b}\mathrm{i}\mathrm{c}}=\frac{e^2}{4\pi {\epsilon }_0}\sum _{i\ne j}\frac{q_i{q}_j}{r_{ij}}$$ 3

$$E_{Vdw}=\sum _{i\ne j}4{\epsilon }_{ij}[{\left(\frac{{\sigma }_{ij}}{r_{ij}}\right)}^{12}-{\left(\frac{{\sigma }_{ij}}{r_{ij}}\right)}^6]$$ 4

Combining rules

$${\sigma }_{ij}=\frac{{\sigma }_i+{\sigma }_j}{2}$$ 5

$${\epsilon }_{ij}={({\epsilon }_i{\epsilon }_j)}^{1/2}$$ 6

$$E_{\mathrm{b}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{s}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{t}\mathrm{c}\mathrm{h}}=\sum _{\mathrm{b}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{s}}{k}_1{(r_{ij}-r_0)}^2$$ 7

$$E_{\mathrm{a}\mathrm{n}\mathrm{g}\mathrm{l}\mathrm{e}\mathrm{b}\mathrm{e}\mathrm{n}\mathrm{d}}=\sum _{\mathrm{a}\mathrm{n}\mathrm{g}\mathrm{l}\mathrm{e}\mathrm{s}}{k}_2{({\theta }_{ij}-{\theta }_0)}^2$$ 8

where ε₀ is the vacuum permittivity, q _i and q _j are atomic charges, r _ij is the distance between atoms, ε _ij and σ _ij are Lennard-Jones parameters.

The migration ability of water molecules is quantitatively described by the mean square displacement (MSD), which is defined as follows

$$\mathrm{M}\mathrm{S}\mathrm{D}(\tau )=⟨{|r_i(t+\tau )-r_i(t)|}^2⟩$$ 9

in which r _i (t) denotes the position vector of the i atom or molecule at time t, τ is the time interval.

The diffusion coefficient D can be obtained from the linear slope of the mean square displacement curve based on the Einstein relation.

$$D=\frac{1}{6N}\underset{\tau \to \infty }{\lim }\frac{d}{d\tau }{[ri(t+\tau )-ri(t)]}^2=\frac{1}{6}\underset{\tau \to \infty }{\lim }\frac{d}{d\tau }\mathrm{M}\mathrm{S}\mathrm{D}(\tau )$$ 10

The Radial Distribution Function (RDF) is used to analyze the density distribution of particles within a specified distance range.

$$g_{\mathrm{A}\mathrm{B}}(r)=\frac{d{N}_{AB}(r)}{4\pi r^{2}dr}·\frac{V}{N_{\mathrm{A}}{N}_B}$$ 11

where g _AB(r) is the density distribution of particles in the range of r distance, V is the system volume, N _A and N _B represent the number of two types of particles. ,

3. Results and Discussion

3.1. FTIR

As shown in Figure , broad O–H stretching vibration bands in the range of 3200–3600 cm^–1 are observed for all three substances. For CMC, this band originates from its abundant hydroxyl groups, while for AMPS, it is attributed to the –OH of the sulfonic acid group. The presence of a distinct absorption band in this region for the CMC-co-AMPS product confirms the existence of hydroxyl groups, which may stem from unreacted hydroxyls in CMC and/or the sulfonic acid groups from AMPS. Additionally, AMPS exhibits a band near 3250 cm^–1 corresponding to the N–H stretching vibration of the secondary amide group. ,

1.

FTIR spectra of CMC, AMPS and CMC-co-AMPS.

As for the characteristic absorption peaks of the reactant molecule structure, specifically, in the spectrum of CMC, the absorption peaks near 1610 cm^–1 and 1410 cm^–1 are ascribed to the asymmetric and symmetric stretching vibrations of the COO^– group, respectively. The absorption band between 1000 and 1100 cm^–1 arises from C–O and C–O–C stretching vibrations in the sugar rings, glycosidic bonds, and secondary alcohol groups. The FTIR spectrum of AMPS shows a characteristic peak at 1670 cm^–1 for the CO stretching vibration of the secondary amide, a band at 1550 cm^–1 associated with C–N out-of-plane bending and N–H deformation vibrations, and a peak at 1610 cm^–1 for the CC stretching vibration. The absorption band in the 1050–1250 cm^–1 region is characteristic of the sulfonic acid group. ,

In the spectrum of the product CMC-co-AMPS, the absorption band near 1650 cm^–1 is likely a superposition of the asymmetric COO^– stretching vibration from CMC and the CO stretching vibration from the amide group of AMPS. Characteristic peaks of the secondary amide group are present at 1550 cm^–1, and the characteristic bands for the sulfonic acid group and the sugar rings/glycosidic bonds are observed in the 1050–1250 cm^–1 and 1000–1100 cm^–1 regions, respectively. Crucially, the absorption peak at 1610 cm^–1, corresponding to the CC bond, is absent. These findings collectively indicate that the molecular structure of the CMC-co-AMPS product incorporates characteristic functional groups from both CMC and AMPS. The disappearance of the CC absorption peak provides strong evidence for the successful grafting of AMPS onto the polysaccharide chain of CMC.

3.2. NMR

Figure shows the ¹H NMR spectra. For CMC, the characteristic signals between 3.5 and 4.0 ppm are assigned to the protons on the pyranose ring and the –CH₂COO^– group. The broad signal near 4.1 ppm indicates a β-configuration in the CMC backbone. In the spectrum of AMPS, the most intense signal at 1.34 ppm corresponds to the equivalent methyl protons of the tert-butyl group, while the signal at 3.26 ppm originates from the methylene group connected to the amide nitrogen. The doublets at 5.53 and 5.98 ppm are characteristic of the vinyl group protons.

2.

¹H NMR spectra of CMC, AMPS and CMC-co-AMPS.

In the spectrum of the polymerization product CMC-co-AMPS, some characteristic signals from the reactants exhibit an upfield shift, which can be attributed to the formation of hydrogen bonds and changes in the chemical environment upon the formation of the polymer chains. Critically, the characteristic doublets of the vinyl group at 5.53 and 5.98 ppm disappear, confirming the consumption of the CC double bonds in AMPS during the reaction. Furthermore, a broad, weak signal emerges in the region of 1.0–2.5 ppm, which is ascribed to the methylene protons within the polymer backbone. Additionally, a new set of signals appears in the 6.9–8.0 ppm region, likely corresponding to the amide N–H protons of the polymer.

3.3. XRD

Figure presents the XRD patterns of the raw materials and the resulting product. The broad diffraction peak near 20° observed for the raw material CMC indicates its amorphous nature. In contrast, the AMPS pattern displays multiple sharp diffraction peaks, characteristic of a typical crystalline structure with numerous diffraction planes, suggesting a complex crystal lattice. The three strongest peaks, located at 11.5°, 15.3°, and 23.1°, correspond to diffraction from the primary crystal planes. The XRD pattern of the CMC-co-AMPS product is distinctly different from those of the starting materials, exhibiting both a broad hump and several sharp peaks, which indicates a biphasic structure of partial crystallinity. The most intense peak at 22.2° is likely retained from the crystal plane corresponding to the 21.9° peak in AMPS. Furthermore, diffraction planes in the range of 24.3° to 27.7°, present in the AMPS structure, are also preserved in the product. Additionally, several weak crystalline peaks appear at 33.9° and higher angles, possibly resulting from the formation of fine microcrystalline structures or minor molecular impurities during the polymerization process.

3.

XRD spectra of CMC, AMPS and CMC-co-AMPS.

Based on the collective analysis from the FTIR, NMR, and XRD, it is confirmed that CMC and AMPS have combined via the opening of the carbon–carbon double bond, with the molecular structure of the product retaining the characteristic functional groups and partial crystalline features of the raw materials. The proposed reaction mechanism is schematically illustrated in Figure . It is hypothesized that the hydroxyl groups on the CMC molecular backbone are activated by the initiator, subsequently initiating a chain growth reaction by opening the double bond and ultimately forming a three-dimensional network molecular structure with AMPS via the cross-linker. The –SO₃H groups provide strong hydrophilicity and ionic character, which enhance water uptake capacity via hydrogen bonding and osmotic effects. Meanwhile, the −CONH- groups contribute to hydrogen-bonding networks for water retention. On the other hand, upon thermal decomposition, the –SO₃H groups promote char formation, which in turn improves flame retardancy.

4.

Reaction mechanism diagram.

3.4. Single-Factor Experiment Analysis

Figure presents the experimental results of the effects of the monomer AMPS, initiator KPS, and cross-linker MBA concentrations on the water absorbency of the product. At lower AMPS dosages, the product formed a loose cross-linked structure incapable of effectively retaining water molecules. The maximum water absorption rate of 52 was achieved at an AMPS dosage of 1.0 g. However, when the AMPS amount was further increased to 1.5 g, the water absorption ratio decreased, accompanied by an increase in fluidity. This phenomenon may be attributed to excess AMPS either self-polymerizing or remaining unreacted, with these small molecules disrupting the polymeric network structure.

5.

Single factor experiments curves.

The optimal initiator KPS concentration was found to be 1.0 mol %. An insufficient amount of initiator resulted in fewer activated hydroxyl sites on the CMC chains, leading to the formation of polymers with lower molecular weight that failed to develop an adequate three-dimensional network, thus yielding a lower absorption ratio. Conversely, an excessive amount of initiator could promote self-polymerization of the monomer and cross-linker, which is also detrimental to the formation of the desired molecular structure. ,

When the cross-linker MBA content was insufficient, the product tended to form branched molecular structures lacking a sufficient three-dimensional network to retain water. As the cross-linker dosage increased, the molecular network structure became more developed, thereby improving the water absorption ratio. However, when the cross-linker concentration exceeded 0.8 mol %, the excess cross-linker may have undergone self-polymerization or interfered with the chain propagation of the monomer, resulting in a reduction of the water absorption capacity. ,

A high water absorption rate corresponded to a small measurement error whose relative error was less than 10%, whereas a low water absorption rate resulted in a larger relative error. The higher instability in water absorption under excess reactant conditions may be attributed to the randomness of the material’s three-dimensional cross-linking structure.

3.5. Response Surface Analysis

Figure a presents the response surface illustrating the effects of CMC-co-AMPS and bentonite contents on the compressive strength of composite cement after 28 days of standard curing. When the CMC-co-AMPS content is low, the compressive strength increases with its content. This is primarily attributed to the moisture absorption and retention properties of CMC-co-AMPS, which provide the necessary moist conditions for cement hydration and thereby promote the hydration reaction during curing. However, when the CMC-co-AMPS content becomes excessively high, the compressive strength of the composite cement begins to decline. This decrease is due to the intercalation of excessive organic components between cement particles, which compromises the structural integrity. Similarly, the effect of bentonite content on the compressive strength exhibits a trend of first increasing and then decreasing. At low bentonite content, its water absorption and expansion behavior facilitate the formation of a dense cement structure. In contrast, when the bentonite content is too high, its water absorption effect hinders hydration reactions and disrupts the three-dimensional structure of the cement. ,

6.

Response surfaces of CMC-co-AMPS and bentonite dosages. (a) Compressive strength response; (b) Water absorption response.

As shown in Figure b, the water absorption rate of the composite cement is positively correlated with the amounts of CMC-co-AMPS and bentonite. This is attributed to the inherent hydrophilicity and water-absorbing nature of both materials.

According to the response surface analysis, the optimal formulation, consisting of 6.00% CMC-co-AMPS and 5.02% bentonite, was predicted to yield a maximum compressive strength of approximately 18.8 MPa and a water absorption rate of 63.57% for the composite cement. This predicted ratio was subsequently validated experimentally. The measured compressive strength was about 19 MPa, and the measured water absorption rate was about 62%, resulting in prediction errors of 1.1% and −2.4%, respectively.

3.6. Microscopic Morphology Analysis of the Composite Cements

As observed in Figure a–c, the surface morphology of the ordinary cement appears relatively loose with numerous attached sand particles. This loose structure not only compromises its mechanical strength but also creates pores that allow oxygen permeation. In contrast, the composite cement with added CMC-co-AMPS exhibits a much denser surface morphology. This can be attributed to the CMC-co-AMPS intertwining within the cement structure, forming a compact coating. However, some micropores remain, likely resulting from the volume shrinkage of the polymer upon drying and dehydration. It is noteworthy that upon water absorption and swelling, the CMC-co-AMPS gel can effectively block these pores, thereby enhancing the material’s ability to block air. The surface morphology of the final CBBS product (Figure c) shows significantly reduced pore size and quantity, which is attributed to the pore-filling effect of nanosilica. Furthermore, bentonite, also possessing high water absorbency, can synergize with the polymer gel to absorb water, provide thermal insulation, and block air.

7.

SEM and AFM images. (a) SEM of cement; (b) SEM of CMC-co-AMPS cement; (c) SEM of CBBS cement; (d) AFM of cement; (e) AFM of CMC-co-AMPS cement; (f) AFM of CBBS cement.

Figure d–f present the AFM results, revealing that the ordinary cement surface features large granules with significant height variations. The stacking of such structures tends to form intergranular voids, adversely affecting the overall strength and air tightness. The surface of the CMC-co-AMPS modified cement becomes notably smoother, with a height difference confined within 10 nm. This smoothing effect is due to the fluid polymer gel, which exhibits structural adaptability and largely fills the original voids in the cement. With the further incorporation of boric acid, nanosilica, and bentonite (the CBBS composite), the surface regains some roughness due to the influence of these solid additives. Nevertheless, the surface topography remains more uniform, with reduced height variations compared to the ordinary cement.

3.7. Analysis of Thermal Stability Test

Figure a shows the TG-DSC curves of CMC-co-AMPS. As observed, the synthesized product exhibits relatively poor thermal stability, with continuous mass loss upon heating. Three distinct endothermic peaks are present in the DSC curve. The slight downward shift near 100 °C is attributed to the evaporation of bound water from the product. The endothermic peak starting around 170 °C corresponds to the decomposition of the polymerized AMPS structure within the product molecules, which catalyzes the dehydration of the main chain. The endothermic peak in the range of 350–400 °C is likely due to the scission of chemical bonds and decomposition of the polymer segments under high temperature. These two stages of thermal decomposition are dominant, accounting for approximately 50% of the mass loss. Finally, the absorption peak at 520 °C may be attributed to the further carbonization of the residual main chain, leaving about 40% carbon residue.

8.

TG-DSC curves of (a) CMC-co-AMPS; (b) CMC-co-AMPS and boric acid composite cement; (c) CMC-co-AMPS, boric acid and nanosilica composite cement; (d) CMC-co-AMPS, boric acid and bentonite composite cement;.

Figure b presents the TG-DSC curves of the composite cement with the CMC-co-AMPS and boric acid added. The DSC curve shows primary endothermic peaks near 100 °C, 120 °C, and 170 °C, accompanied by a mass loss of approximately 6%. These peaks are attributed to the evaporation of bound water in the product, the thermal dehydration of boric acid in the composite, and the decomposition of the AMPS structure within the polymer molecules.

From Figure c,d, it can be observed that since the cement, nanosilica, and bentonite in the composite material do not undergo thermal decomposition within the tested temperature range, the main mass loss occurs around 120 °C, corresponding to the dehydration of the polymer and boric acid in the composite material. Overall, the composite product demonstrates strong thermal stability. During heating, both the decomposition of the polymer’s AMPS structure and the reaction of boric acid release water. On one hand, the evaporation of released water molecules absorbs heat. On the other hand, the dehydrated products form dense carbon and boron oxide films, which provide sealing, thermal insulation, and oxygen barrier effects, thereby enhancing the flame-retardant performance of the composite.

3.8. Analysis of Water Absorption Test

The water absorption test results are shown in Figure . The ordinary cement exhibited a relatively low water absorption rate, which stabilized at approximately 13% after 4 h of immersion. This is attributed to the large pores within the ordinary cement structure, which are ineffective at retaining water. In contrast, the CMC-co-AMPS cement showed a significant improvement, reaching a water absorption ratio of about 45%. This enhancement is due to the abundance of hydrophilic functional groups (e.g., hydroxyl, carboxyl, amino, and sulfonate groups) in the CMC-co-AMPS molecular structure, which adsorb a large number of water molecules via hydrogen bonding, leading to gel swelling. Furthermore, the incorporation of the gel reduced pore size in the composite cement, thus facilitated water uptake through capillary action. The swollen gel also trapped water within the pores, thereby improving the overall water retention capacity of the composite cement. With the additional inclusion of bentonite and nanosilica, the water absorption ratio further increased to approximately 62%, owing to the inherent hydrophilicity of nanosilica and the water absorbency of bentonite. The largest weighing error occurred after 2 h of soaking. This is attributed to differences in pore distribution among the different cement samples. In the early stage of soaking, water absorption is primarily confined to the surface and shallow pore structures, and has not yet penetrated into the interior. Consequently, a relatively large error at this stage is considered acceptable. As the soaking time increased, the relative error became limited to within 10%.

9.

Variation curves of water absorption rate with soaking time.

3.9. Analysis of Coal Oxidation Inhibition Test

Figure presents the experimental results of temperature variation and CO concentration generated during the heating of coal samples treated with different test materials. When heated to 120 min, the oven temperature reached approximately 150 °C. At this point, the coal itself began to release heat due to oxidative reactions, causing its temperature to exceed that of the oven and subsequently rise rapidly. By 140 min, the coal temperature surpassed 200 °C. All three tested flame-retardant components demonstrated some effect in delaying the oxidative heat release of coal. The temperature rise trends were relatively similar within the first 100 min of heating. The samples treated with CMC-co-AMPS and the composite gel showed lower temperatures, primarily due to the heat absorption from the evaporation of residual bound water in the gel. After 120 min of heating, the temperature of the sample treated with boric acid was slightly lower than that of the sample treated with CMC-co-AMPS. This is because the ambient temperature reached the decomposition temperature of boric acid, and the evaporation of released water molecules carried away a portion of the heat. The experimental group treated with the composite gel, which combined the inhibitory effects of both CMC-co-AMPS gel and boric acid, did not reach 200 °C until 200 min of heating. The relative error of temperature testing was generally around 5%, and the results were relatively stable. The causes of the error include differences in coal samples themselves, differences in sample preparation, and instrument accuracy issues.

10.

Results of coal oxidation inhibition tests.

In terms of CO release, boric acid provided a certain physical barrier effect but had a limited impact on CO release from the heated coal. The CMC-co-AMPS gel showed a more pronounced suppressive effect on CO release. According to TG-DSC analysis, CMC-co-AMPS undergoes side chain decomposition within the tested temperature range, which is a process that scavenges free radicals and thereby affects coal oxidation. The composite gel containing boric acid exhibited an even more significant inhibitory effect on CO release. This is attributed to the mutually promoting decomposition reactions of boric acid and CMC-co-AMPS, which enhance the suppression of coal oxidation. The inhibitory effect of CBBS cement on temperature rise and CO release is slightly weaker than that of the gel. This difference is likely because the cement partially coats the coal surface, thereby impeding its oxidation. The relative error of carbon monoxide content testing was generally within 5%, and the results were relatively stable. The causes of the error include differences in the coal sample itself, differences in sample preparation, and instrument accuracy issues.

3.10. Analysis of Compressive Strength Test

Figure shows the compressive strength test results for different cement types after various curing periods. The compressive strengths of ordinary cement after 3, 7, 14, and 28 days of curing were 8.1, 11.7, 12.8, and 13.7 MPa, respectively. The CMC-co-AMPS cement exhibited a slight improvement in maximum compressive strength, reaching approximately 15.9 MPa after 28 days of curing, which represents an increase of about 16.1%. This is likely attributable to the polymer chains of CMC-co-AMPS interpenetrating the layered structure of the cement, acting as bridges and thereby facilitating the curing reaction. Combined with the conclusions from microstructural analysis, the filling of cement voids by CMC-co-AMPS and the shrinkage stress induced by gel dehydration resulted in a more compact composite cement. With the further addition of nanosilica and bentonite, the maximum compressive strength after 28 days of curing reached 19.0 MPa, corresponding to an increase of approximately 38.7% compared to ordinary cement. This significant improvement can be attributed to two main factors: first, the incorporation of nanosilica and montmorillonite further densified the composite cement structure; second, the hydrophilic nature of nanosilica and the water-retaining capacity of montmorillonite help to maintain moisture, which is slowly released during the curing process, thereby providing a consistently humid environment that promotes the hydration of cement components. The relative error of compressive strength testing was generally within 5%, and the results were relatively stable. The error was due to small differences in the structure and curing process of the sample itself.

11.

Results of compressive strength tests.

3.11. Analysis of Molecular Dynamics Simulation

The influence of CMC-co-AMPS molecules on the adsorption and diffusion of water on coal was investigated using molecular dynamics simulation. From the molecular model changes before and after simulation in Figure , it can be observed that in the absence of CMC-co-AMPS, water molecules spread laterally after settling and contacting the coal molecules. In contrast, the CMC-co-AMPS molecular chains adopted a longitudinal arrangement after simulation, acting as a bridge between the coal molecules and water molecules, and the horizontal dispersion of water molecules also decreased. The relative concentration distribution of each component along the z-axis in the equilibrated system shows that without CMC-co-AMPS, the wetting of the coal surface by water is limited. The thickness of the adsorption layer (d₀) at the coal/water interface is approximately 7.0 Å (from 23.1 Å to 30.1 Å), and water molecules hardly penetrate into the coal matrix. In the presence of CMC-co-AMPS, the concentration profile changes significantly. Within the distribution region of coal molecules, a small number of water molecules appear in the range of 0.1 Å to 2.3 Å, and CMC-co-AMPS is distributed in the range of 14.3 Å to 16.3 Å, indicating that both water and CMC-co-AMPS molecules have penetrated into the layered structure of the coal. Furthermore, the adsorption layer thickness (d₃) at the CMC-co-AMPS/water interface is approximately 18.3 Å (from 39.2 Å to 57.5 Å), and both CMC-co-AMPS and water molecules exhibit high concentration distributions within this range. This demonstrates the high hydrophilicity of CMC-co-AMPS, which, through its hydrophilic groups, can enrich and draw water molecules into the coal matrix, thereby achieving wetting, cooling, and flame-retardant effects. ,

12.

Molecular dynamics simulation images.

4. Conclusion

This study developed a composite cement with integrated flame-retardant and support functionalities for backfilling underground coal mine goafs by modifying the natural polymer CMC with AMPS and incorporating boric acid, nanosilica, and bentonite as fillers. The molecular structural features and reaction pathway were analyzed using FTIR, NMR, and XRD. Key performance parameters, including thermal stability, flame retardancy, and compressive strength, were tested. The flame-retardant mechanism was elucidated through molecular dynamics simulation. The principal conclusions are as follows.

• (1)The natural polymer CMC was successfully modified via graft copolymerization with AMPS through free-radical initiation. This reaction resulted in a three-dimensional network structure that incorporated amide and sulfonate groups from AMPS while partially preserving its crystalline features.
• (2)Single factor experiments identified the reactant dosages for achieving the maximum water absorption rate. The product CMC-co-AMPS exhibited the highest water absorption rate, approximately 52, when the mass ratio of CMC to AMPS was 10:3, with KPS initiator at 1.0 mol % and MBA cross-linker at 0.8 mol %.
• (3)The composite cement exhibited significantly enhanced performance, with a 49% increase in water absorption capacity compared to ordinary cement. Upon exposure to heat, the CMC-co-AMPS hydrogel and boric acid dehydrate, providing a cooling effect and forming protective barriers that suppress coal oxidation and CO release. Molecular dynamics simulations confirmed that CMC-co-AMPS facilitates water penetration into the coal matrix. This property allows the preplaced backfill material to be activated via water injection upon detecting temperature anomalies, effectively preventing spontaneous combustion.
• (4)The modified composite cement demonstrated a compressive strength of 19.0 MPa, representing a 38.7% improvement over ordinary cement, which fulfills the mechanical requirements for supporting the overlying strata in a goaf.

The data that supports the findings of this study are available from the corresponding author upon reasonable request.

#.

Y.M. and J.S. contributed equally.

This work was financially supported by the National Natural Science Foundation of China (Grant No. 52404198, 52574228), the Natural Science Foundation of Jiangsu Province (Grant No. BK20250976), and the State Key Laboratory of Deep Earth Exploration and Imaging (Grant No. DEEI20252241).

The authors declare no competing financial interest.