Optimization and Mechanistic Investigation of Coal Gangue–Blast Furnace Slag Composite Geopolymers

Abstract

Coal gangue (CG), a major solid waste generated during coal development, presents critical environmental challenges due to its large-scale accumulation and associated ecological impacts, thereby necessitating the development of efficient utilization strategies. This investigation developed a composite geopolymer system through the alkali-activated co-utilization of uncalcined CG and blast furnace slag (BFS), demonstrating an environmentally sustainable approach for industrial byproduct value addition. The effects of key parameters, including BFS content, liquid-to-solid ratio, alkali activator dosage, waterglass modulus, and curing regime, on the strength development were first investigated through single-factor experiments. Based on these results, response surface methodology was applied to optimize the preparation parameters and develop a quadratic regression model describing the relationship between compressive strength and the influencing factors. The optimal conditions (a waterglass modulus of 1.06, an alkali activator dosage of 13.81%, and an initial 24 h curing temperature of 30 °C) were determined to maximize compressive strength. The reaction mechanisms were further explored using XRD and SEM-EDS, which confirmed the existence of calcium silicate hydrate, calcium aluminum silicate hydrate, and geopolymer gel in the composite geopolymer matrix.

1. Introduction

Coal remains a cornerstone of China’s national economy, despite a decline in annual raw coal production. By the end of 2023, it accounted for over 66% of the primary energy supply [1], maintaining its irreplaceable role in the short term. However, the expansion of the coal mining industry has exacerbated environmental pressures. The accumulation of coal gangue (CG), a predominant solid waste generated during coal extraction and processing, poses significant environmental challenges through land occupation and ecological deterioration in mining regions [2,3,4], necessitating the development of sustainable valuable strategies to mitigate its environmental footprint while advancing green mining initiatives.

CG, primarily composed of aluminosilicate compounds [5], can be recycled for geopolymer production, promoting waste reutilization. Geopolymers are innovative inorganic materials with cementitious properties similar to those of ordinary Portland cement. Geopolymers represent a technologically viable substitute for conventional cement-based materials as a result of their environmentally benign synthesis process [6,7], which generates substantially lower CO₂ emissions compared to conventional Portland cement production [8], coupled with exceptional durability characteristics including enhanced mechanical strength [9], reduced permeability [10], and remarkable stability under aggressive service conditions such as marine environments [11] and elevated temperatures [12]. Geopolymerization is a complex chemical process in which aluminosilicate precursors undergo alkaline dissolution, followed by polycondensation into oligomeric species, and the subsequent formation of a three-dimensional gel network. Aluminosilicate-rich precursors, derived from both natural minerals and industrial byproducts, are classified as low-calcium or high-calcium systems based on their CaO content. The evolution of the gel phase in geopolymerization systems is influenced by calcium content, with low-calcium precursors such as metakaolin [13], fly ash [14], and CG [15] forming sodium aluminosilicate hydrate (N-A-S-H) networks, while high-calcium sources such as blast furnace slag (BFS) predominantly yield calcium aluminosilicate hydrate (C-A-S-H) matrices [16]. Additionally, when water is present, hydration reactions occur, leading to the formation of hydrated calcium silicate (C-S-H) gel. The synergistic use of low- and high-calcium precursors in geopolymer formulations results in a composite gel system comprising both C-S-H and (N, C)-A-S-H phases, whose relative proportions and structural characteristics fundamentally determine the mechanical performance and long-term durability of the final products.

Raw CG has a stable crystal structure and low reactivity, resulting in poor geopolymer performance, which significantly limits its resource utilization. Current research primarily focuses on enhancing the reactivity of CG through high-temperature calcination to improve the mechanical properties of geopolymers [17,18,19]. However, this process requires substantial energy, similar to cement production. In recent years, the combined use of various raw materials in composite geopolymer synthesis has gained increasing attention [12,20,21,22]. The incorporation of high-calcium components into a low-calcium geopolymer formulations introduces calcium oxide, which actively engages in the geopolymerization process, leading to the generation of C-S-H, C-A-S-H, and N-A-S-H gels, thereby contributing to the superior mechanical strength [23]. BFS, a metallurgical residue generated during iron smelting, demonstrates significant global availability owing to its high production volumes. Utilizing CG and BFS, both abundant solid waste materials, as raw materials for composite geopolymer production offers significant economic and environmental benefits.

Geopolymer performance is governed by a complex interplay of the type and proportion of aluminosilicate materials, curing conditions, water-to-binder ratio and activator formulation [24,25]. Kul et al. [26] found that the mechanical properties of construction waste-based geopolymers are significantly affected by chemical composition, crystalline structure, and fineness. Kucukyildirim et al. [27] demonstrated that a lower Na₂SiO₃/NaOH ratio of 2.5 accelerated early-stage hydration kinetics but yielded the lowest cumulative heat release. In contrast, a higher ratio of 3.5 promoted the formation of a denser microstructure of the pumice-based geopolymer, consequently achieving the highest compressive strength (36.3 MPa at 28 days). Bai et al. [28] revealed a non-monotonic relationship between alkali activator concentration and compressive strength, with an initial increase followed by a decline, indicating the existence of an optimal concentration threshold. Furthermore, this critical concentration was observed to shift toward higher values with increasing curing temperature, suggesting a thermally dependent activation mechanism in the geopolymer system. Xu et al. [29] demonstrated that optimizing the SiO₂ modulus to 1.5 yielded fly ash-based geopolymers with superior mechanical performance and minimal microstructural porosity, indicating an ideal balance between dissolution and polycondensation reactions. Liu et al. [30] systematically evaluated silica fume’s role in modifying the rheological properties of a slag–fly ash-based geopolymer, demonstrating that its addition decreased yield stress and plastic viscosity in sodium silicate-activated systems but increased these parameters when sodium hydroxide served as the primary activator. Therefore, optimizing synthesis parameters is essential for developing practical and efficient geopolymer products.

In this study, composite geopolymers were synthesized using uncalcined CG and BFS as raw materials. Initially, single-factor experiments were conducted to determine optimal settings for key parameters, including the BFS content, waterglass modulus, alkali activator dosage, liquid-to-solid ratio, and curing temperature. Based on these results, response surface methodology (RSM) was employed to design experiments and develop a multiple regression equation relating the waterglass modulus, alkali activator dosage, curing temperature, and compressive strength. RSM has emerged as a robust statistical technique for constructing mathematical models to predict the performance characteristics of cementitious materials. The methodology facilitates a comprehensive evaluation of parameter interactions while enabling the simultaneous optimization of multiple variables [31,32]. Furthermore, the reaction mechanisms of the composite geopolymer were characterized through X-ray diffraction (XRD) and scanning electron microscopy with energy dispersive spectroscopy (SEM-EDS).

2. Experimental Program

2.1. Materials

The CG and BFS utilized in this study were collected from industrial sites in Shanxi and Chongqing, China, respectively. The loss on ignition of CG and BFS was 12.66% and 0.53%, respectively. And X-ray fluorescence spectroscopy analysis quantified their chemical compositions (

Table 1. Elemental constituents of raw materials (wt%).

	SiO2	CaO	MgO	Al2O3	K2O	Na2O	Fe2O3	P2O5	TiO2	SO3
CG	59.52	0.26	0.18	36.57	0.75	-	1.24	0.05	1.14	0.16
BFS	35.62	36.31	7.50	14.93	0.39	0.22	0.60	-	1.28	2.29

), with particle size distributions demonstrating similar median diameters of 22.01 μm for CG and 21.35 μm for BFS, as illustrated in Figure 1. The alkaline activator formulation consisted of analytical-grade sodium hydroxide flakes combined with industrial waterglass (SiO₂/Na₂O = 3.0) in deionized aqueous medium.

2.2. Experimental Methods and Testing

2.2.1. Sample Preparation Process

Figure 2 presents a schematic representation of the sequential steps in the specimen fabrication process. Sodium hydroxide, industrial-grade waterglass, and deionized water were uniformly blended, naturally cooled to ambient temperature, and rested for 18 h to prepare the alkali activator. Dried CG and BFS were ball-milled and sieved through a 200-mesh screen before being mixed in a predetermined ratio. The raw materials were homogenized with the alkali activator through continuous mechanical stirring to achieve a uniform composite mixture. The resulting slurry was cast into 20 mm cubic molds and subjected to 5 min vibration to minimize entrapped air voids. Finally, the specimens underwent a two-phase curing regimen, beginning with a 24 h thermal treatment in an oven, followed by ambient curing after demolding.

2.2.2. Mixing Ratio Design

In the single-factor experiment, the effects of the BFS content (10 wt%, 30 wt%, 50 wt%, 70 wt%, and 90 wt%), waterglass modulus (0.6, 0.8, 1.0, 1.2, and 1.4), alkali activator dosage (5%, 7.5%, 10%, 12.5%, and 15%), liquid-to-solid ratio (0.21, 0.23, 0.25, 0.27, and 0.29), and initial 24 h curing temperature (30 °C, 40 °C, 50 °C, 60 °C, and 70 °C) on the compressive strength of the samples were analyzed. Twenty-five sample groups were fabricated in triplicate to ensure statistical reliability. The BFS content level was calculated as the mass percentage of BFS replacing CG in the solid precursors, while the alkali activator dosage represented the weight ratio of alkali activator to the aluminosilicate materials. Waterglass modulus variations were achieved by controlled NaOH titration to modify the SiO₂/Na₂O molar ratio in the activator solution.

Building upon the single-factor experiments, a RSM experiment using central composite design (CCD) was conducted for parametric optimization and experimental effort reduction. This experimental framework facilitated the construction of second-order polynomial models to quantify factor–factor interactions.

Table 2. Factors and levels in RSM.

Variables	Factors	Coded Levels of Variables
		−1.682	−1	0	1	1.682
Waterglass module	X1	0.2	0.4	0.7	1	1.2
Alkali activator dosage/%	X2	2.5	5.54	10	14.46	17.5
Curing temperature/°C	X3	30	38	50	62	70

presents the coded values and corresponding levels for each factor, including the modulus of waterglass, the dosage of alkali activator, and the curing temperature. The CCD incorporating three independent variables necessitates a total of 2ⁿ + 2n + m experimental runs, as illustrated in Figure 3, with n representing the variable count under investigation. Specifically, the 2ⁿ cube points correspond to combinations of coded coordinates (x = ±1) at the vertices, while the 2n axial points are located at ±α radial distances from the origin along each factor axis. The center point (m) is where all coded levels are set to 0 at the cube’s center. Since this study considers three variables, a total of 23 experimental runs were conducted, comprising 8 cube points (2ⁿ), 6 axial points (2n), and 9 center points (m) (

Table 3. Experimental scheme in CCD.

Number	X1	X2	X3	Number	X1	X2	X3
1	1	1	−1	13	−1	−1	1
2	−1.682	0	0	14	0	0	0
3	0	−1.682	0	15	0	0	1.682
4	0	0	0	16	−1	1	−1
5	−1	1	1	17	1	1	1
6	0	0	0	18	0	0	0
7	0	0	0	19	0	0	0
8	1.682	0	0	20	0	0	0
9	0	0	−1.682	21	−1	−1	−1
10	1	−1	1	22	0	0	0
11	1	−1	−1	23	0	0	0
12	0	1.682	0

). To ensure both rotatability and orthogonality in a three-factor CCD, the axial distance (α) and center point replicates (m) were mathematically determined as 1.682 and 9, respectively. This configuration satisfies the rotatability condition (α = 2^3/4) while maintaining design orthogonality through balanced center point repetitions. Furthermore, a quadratic response surface model was developed to quantify the parametric influences on strength development.

2.2.3. Testing Methods

The elemental composition of CG and BFS was quantified via X-ray fluorescence spectroscopy (XRF-1800, Shimadzu, Japan). Particle size distribution was characterized using a laser particle size analyzer (Mastersizer 3000, Malvern Panalytical, UK). Specimens’ strength was determined via triplicate testing using a universal testing machine (AGN-250, Shimadzu, Japan) at a constant loading rate of 0.1 mm/min, with the results reported as the mean value to ensure measurement accuracy. Phase evolution during geopolymerization was characterized by X-ray diffraction (X’Pert PRO, PANalytical, Netherlands) comparing precursor and product spectra. Microstructural and elemental analysis was conducted via SEM-EDS (VEGA3, TESCAN, Czech Republic).

3. Results and Discussion

3.1. Single-Factor Experimental Study on Preparation of Composite Geopolymer

3.1.1. The Ratio of Liquid to Solid

Water is a crucial component in the preparation of geopolymers, acting as a medium that facilitates dissolution, polycondensation and gelation during the geopolymerization process. An insufficient water content could result in a poorly compacted matrix, leading to reduced strength [33]. Additionally, water serves as a reactant in the initial reaction phase. When the OH- concentration is sufficiently high, the presence of more water accelerates dissolution and hydrolysis. Consequently, as the liquid-to-solid ratio increases from 0.21 to 0.27, the strength of the geopolymer matrix rises from 38.93 MPa to 46.44 MPa (Figure 4a). At a liquid-to-solid ratio of 0.27, the slurry exhibits optimal viscosity and fluidity, achieving peak strength. However, further water addition adversely affects strength development. As the dominant reaction transitions from hydrolysis (which consumes water) to polycondensation, excess water could kinetically hinder the polycondensation process. Furthermore, the alkali activator concentration decreases with higher water content, which simultaneously enhances slurry fluidity. While increased fluidity may improve workability, excessive water can lead to over-dilution, promoting the formation of a porous and mechanically weakened matrix. This ultimately impairs the overall strength.

3.1.2. Alkali Activator Dosage

Figure 4b demonstrates that both insufficient alkali activator content and excessive alkali activator content negatively affect geopolymerization. Specifically, increasing the alkali activator dosage from 5% to 10% enhances the compressive strength of the geopolymer from 41.75 MPa to 45.79 MPa. However, further increasing the dosage to 15% reduces the strength to 33.48 MPa. The alkali activator plays a crucial role in facilitating the rapid dissolution of aluminosilicate materials [34]. In an alkaline medium, CG and BFS dissolve to produce silicate and aluminate tetrahedra, which subsequently promote the polycondensation of dissolved tetrahedral units. In reaction systems with low alkali activator content, the Na⁺ concentration is insufficient to fully support the polymerization reaction. Conversely, a highly alkaline environment is necessary for the breakdown of raw materials. When the OH⁻ concentration is too low to completely dissolve Si⁴⁺, Al³⁺, and Ca²⁺ from the precursor materials, the dissolution rate of CG and BFS decreases, leading to lower compressive strength. Increasing the alkali activator content enhances the mechanical properties of the matrix by supplying more OH⁻ and reactive silica, which accelerate the dissolution of raw material. However, exceeding the optimal alkali activator content could interfere with the formation of Si-O-Si bonds, ultimately reducing the compressive strength of the geopolymer matrix.

3.1.3. Waterglass Modulus

The waterglass modulus represents the molar ratio of SiO₂ to Na₂O in the alkali activator. As the waterglass modulus decreases, the compressive strength of the matrix gradually increases (Figure 4c). The optimal compressive strength (53.36 MPa) occurs at a modulus of 0.6. The waterglass modulus directly affects the Na₂O content in the alkali activator. A lower modulus increases the solution’s alkalinity, promoting the gradual depolymerization of silicate anion polymers [35]. During this process, sodium ion substitution induces Si-O-Si to Si-O-Na bond conversion. Additionally, highly alkaline media facilitate the breakdown of aluminosilicate structures, accelerating the release of silicate and aluminate monomers, which, in turn, facilitate polymerization and improve matrix strength [36]. However, a higher waterglass modulus corresponds to lower solution alkalinity. In such cases, raw materials with lower reactivity, such as coal gangue, require an alkali activator with a higher pH for effective activation. Furthermore, a high modulus signifies a greater concentration of Na₂SiO₃, which could hinder water evaporation and cause polymer chains to depolymerize into individual monomers, ultimately reducing the matrix strength.

3.1.4. BFS Content

The mechanical strength of geopolymers largely depends on the amorphous phase content in the raw materials. Figure 4d illustrates that the compressive strength of the geopolymer increases monotonically with the addition of BFS. The incorporation of BFS enhances the mixture’s reactivity by introducing a greater number of reactive components. BFS has a relatively high CaO content, which dissolves more readily than SiO₂ and Al₂O₃ in an alkaline solution due to OH- polarization. The released calcium ions rapidly interact with silicates in the solution to form C-S-H gel [37]. The initial formation of calcium-containing hydration products depletes calcium ions from the solution, accelerating BFS dissolution and promoting continued reaction progression. Additionally, the matrix derived from BFS consists of calcium-rich dense gels, such as C-S-H and C-A-S-H, which coexist with geopolymer gels formed through the alkali activation of CG. This combination results in a more uniform and compact microstructure, where the well-reacted matrix integrates with unreacted particles, ultimately enhancing strength development.

3.1.5. Curing Temperature

Figure 4e presents the effect of initial 24 h curing temperature on the mechanical property of the geopolymer. The compressive strength decreases linearly with increasing curing temperature. Specifically, when the curing temperature rises from 30 °C to 70 °C, the strength declines from 77.76 MPa to 37.27 MPa, demonstrating that higher curing temperatures negatively impact strength development. This occurs because optimal curing temperatures facilitate the formation of a well-structured matrix, whereas exceeding this threshold could adversely affect the geopolymer’s pore structure. While higher curing temperatures enhance the release of silicon and aluminum species from the precursor materials, accelerating polycondensation and matrix strengthening, excessive heat exposure leads to strength degradation [38,39]. The rapid formation of C-S-H and geopolymer gels on the surface of unreacted particles may hinder the further dissolution of the aluminosilicate precursor. Additionally, excessively high temperatures accelerate pore solution evaporation, resulting in incomplete reactions and reduced bulk density, ultimately lowering strength.

3.1.6. One-Way ANOVA Test

A one-way ANOVA at a 95% confidence level was performed to evaluate the impact of the liquid-to-solid ratio, alkali activator dosage, waterglass modulus, BFS content, and initial 24 h curing temperature on the mechanical properties of the geopolymer matrix. As presented in

Table 4. One-way ANOVA test results.

Factor	Source	Sum of Square	d.f.	Mean Square	F	p-Value
Liquid-to-solid ratio	Between	164.46	4	41.11	2.87	8.01 × 10−2
	Within	143.12	10	14.31
	Total	307.58	14
Alkali activator dosage	Between	328.53	4	82.13	15.81	2.52 × 10−4
	Within	51.96	10	5.20
	Total	380.49	14
Waterglass modulus	Between	480.11	4	120.03	35.06	7.40 × 10−6
	Within	34.24	10	3.42
	Total	514.35	14
BFS content	Between	3038.50	4	759.62	32.73	1.02 × 10−5
	Within	232.12	10	23.21
	Total	3270.62	14
Curing temperature	Between	3245.97	4	811.49	44.55	2.43 × 10−6
	Within	182.16	10	18.22
	Total	3428.13	14

, the alkali activator content, waterglass modulus, BFS proportion, and curing temperature significantly influenced the development of compressive strength, while the liquid-to-solid ratio had no significant effect on strength development. To optimize the utilization of CG, the waterglass modulus, alkali activator content, and initial 24 h curing temperature were identified as the primary factors for the subsequent RSM experiment, with compressive strength as the response variable. Other components were treated as secondary factors, with their quantities held constant (a liquid-to-solid ratio of 0.27 and a BFS content of 50%) while the effects of the primary factors were investigated.

3.2. RSM Experimental Study on the Preparation of Composite Geopolymer

A CCD within the RSM framework was employed to model the correlation between process parameters (waterglass module, alkali activator dosage, and curing temperature) and the dependent variable/response (compressive strengths).

Table 5. Experimental results based on CCD.

Mix	Variables			Response
	Waterglass Module	Alkali Activator Dosage/%	Curing Temperature/°C	Compressive Strength/MPa
1	1	14.46	38	85.62
2	0.2	10	50	28.72
3	0.7	2.5	50	32.28
4	0.7	10	50	69.97
5	0.4	14.46	62	39.08
6	0.7	10	50	69.14
7	0.7	10	50	66.77
8	1.2	10	50	59.36
9	0.7	10	30	73.77
10	1	5.54	62	45.11
11	1	5.54	38	61.28
12	0.7	17.5	50	52.10
13	0.4	5.54	62	36.89
14	0.7	10	50	69.51
15	0.7	10	70	46.83
16	0.4	14.46	38	52.29
17	1	14.46	62	37.43
18	0.7	10	50	69.77
19	0.7	10	50	64.64
20	0.7	10	50	63.15
21	0.4	5.54	38	33.85
22	0.7	10	50	67.31
23	0.7	10	50	75.36

presents the experimental results of composite geopolymer with varying waterglass moduli, alkali activator contents, and initial 24 h curing temperatures. These data were employed to develop predictive models for estimating the strength development. Furthermore, the results served as the basis for conducting the ANOVA and subsequent optimization procedures.

3.2.1. Establish Function Model

A quadratic regression model was established via ordinary least squares to quantify the correlation between the response variable and multiple independent variables. An ANOVA was carried out to assess the model’s significance and accuracy. To ensure statistical validity and suitability for predictive purposes, the model needed to demonstrate statistical significance. The F-values and p-values of individual terms were calculated to determine their significance, as summarized in

Table 6. ANOVA summary for individual terms.

Source	Mean Square	F-Value	p-Value	Significant
X1	1034.58	90.43	<0.0001	Yes
X2	365.20	31.92	<0.0001	Yes
X3	1051.79	91.94	<0.0001	Yes
X1X2	1.96	0.17	0.6855	No
X1X3	367.23	32.10	<0.0001	Yes
X2X3	291.28	25.46	0.0002	Yes
X12	1107.04	96.77	<0.0001	Yes
X22	1287.28	112.52	<0.0001	Yes
X32	107.20	9.37	0.0091	Yes

. The statistical analysis employed a 95% confidence threshold (p < 0.05) for significance testing. When analyzing individual factors, variables X₁, X₂, and X₃ exhibited statistically significant effects on the model, as indicated by p < 0.0001 and exceptionally high F-values, confirming their strong influence on the mechanical properties of composite geopolymers.

Additionally, significant interaction effects were identified between X₁ and X₃ and between X₂ and X₃, with robust statistical evidence (p < 0.05 for all interactions). The interaction between X₁ and X₃ suggests that as the waterglass modulus increases, the curing temperature should be carefully adjusted to sustain an efficient geopolymerization process and prevent strength reduction in the matrix. Therefore, these factors should not be optimized independently as their synergistic effects should be considered.

The quadratic terms (X₁² and X₂²) exhibited statistically significant effects in the model (p < 0.0001 for both), indicating a nonlinear relationship between these factors and the compressive strength of composite geopolymers. Therefore, incorporating quadratic terms is essential for accurately capturing the complex behavior of these parameters. The nonlinear nature of these relationships suggests that the influence of these parameters on the mechanical property does not follow a fixed rate of change. While moderate elevations in these variables boost strength, surpassing a critical threshold reverses the beneficial impact. Accounting for nonlinear parameter interactions significantly increases the model’s predictive capability across diverse conditions. Moreover, accounting for these complexities allows for better adaptation to field-relevant conditions, improving predictive reliability.

According to the results of the significance test, an insignificant factor of the results was rejected and the developed quadratic model (Equation (1)) was obtained. The model exhibited an exceptionally low p-value (<0.0001), indicating strong statistical significance for predicting geopolymer properties. The adequacy, efficiency, and validity of a model were further assessed through lack-of-fit testing. For a well-fitted model with minimal noise, the p-value for the lack of fit should exceed 0.05, indicating that the lack of fit is not statistically significant. The developed models displayed p-values above 0.05 for the lack of fit, suggesting no significant deviation from pure error. Therefore, the model exhibits improved reliability and predictive capability with reduced error margins. Additionally, the scatter diagram in Figure 5 shows a strong agreement between the model-predicted values and the experimental measurements. This close correlation confirms the model’s predictive robustness and its faithful representation of experimental observations.

Y = −186.594 + 255.682X₁ + 15.868X₂ + 3.544X₃ − 1.899X₁X₃ − 0.114X₂X₃ − 94.013X₁² − 0.451X₂² − 0.018X₃²

(1)

3.2.2. Interaction Analysis

Figure 6a illustrates the significant influence of the waterglass modulus and the initial 24 h curing temperature on the compressive strength of composite geopolymers at a fixed alkali activator content of 10%. Increasing the waterglass modulus within an optimal range enhances the [SiO₄]⁴⁻ concentration in the alkaline solution, facilitating its reaction with Ca²⁺ dissolved from BFS to form calcium silicate hydrate (C-S-H) gel. This gel effectively fills pores and improves matrix densification. Simultaneously, the highly viscous sodium silicate solution promotes the formation of sodium aluminosilicate hydrate (N-A-S-H) gel, leading to a compact microstructure and enhanced mechanical performance, consistent with the findings of Çelikten et al. [40]. The strengthening effect exhibits a clear temperature dependence. When the waterglass modulus increases from 0.4 to 1.0, specimens cured at 30 °C experience a substantial strength increase from 30 MPa to 80 MPa, whereas those cured at 70 °C show a more moderate rise from 20 MPa to 50 MPa. This variation arises from distinct reaction mechanisms: at lower temperatures (30 °C), BFS hydration dominates, while CG exhibits limited geopolymerization activity; at elevated temperatures (70 °C), CG-BFS synergistic interactions become predominant. However, excessive temperatures have detrimental effects, accelerating moisture evaporation, leading to pore and crack formation, and inhibiting polycondensation reactions, ultimately reducing mechanical strength.

Figure 6b illustrates that the variation in compressive strength with alkali activator content and curing temperature follows a trend similar to that observed with waterglass modulus. Initially, compressive strength increases with rising alkali activator content, reaching a peak before subsequently declining. This behavior results from several interrelated mechanisms. In alkaline solution, CaO in BFS exhibits greater solubility than SiO₂ and Al₂O₃, and the continuous consumption of Ca²⁺ through hydration product formation further accelerates BFS dissolution kinetics [41]. As a result, higher alkalinity enhances BFS decomposition, stimulating the generation of C-S-H gels, which serve as nucleation sites to facilitate raw material dissolution. This process accelerates subsequent reaction stages, including rearrangement, gelation, and hardening. In essence, increased alkali activator content intensifies both geopolymerization and hydration reactions, while alkali metal cations contribute to structural stabilization through charge-balancing effects, collectively enhancing mechanical strength. However, excessive alkali activator content leads to strength deterioration due to carbonate formation. Notably, the effect of temperature is dependent on alkali activator content. At low alkali activator content levels (2.5%), increasing the curing temperature from 30 °C to 70 °C causes only minor strength fluctuations, whereas at high alkali activator content levels (17.5%), the same temperature increase results in a dramatic strength reduction from 89 MPa to 24 MPa. These findings highlight the complex interplay among the waterglass modulus, alkali activator dosage, and initial 24 h curing temperature in determining compressive strength evolution.

3.2.3. Optimal Scheme Prediction and Validation

The strength development of composite geopolymers is governed by the synergistic interactions of the waterglass modulus, alkali activator content, and initial 24 h curing temperature. Using target-driven optimization modeling, the 28-day compressive strength was systematically improved. The optimal processing parameters were identified as a waterglass modulus of 1.06, an alkali activator content of 13.81%, and an initial curing temperature of 30 °C, yielding a predicted strength of 94.41 MPa. To validate the model’s reliability, triplicate verification experiments were carried out under these optimized conditions. The measured average compressive strength of 91.13 MPa deviated by only 4.01% from the predicted value, demonstrating strong agreement and confirming the high predictive accuracy of the developed quadratic model.

3.3. Microstructural Analysis

XRD analysis (Figure 7) indicates that no new crystalline phases are present in the composite geopolymer compared to the raw materials (CG and BFS). The low reactivity of uncalcined CG results in residual kaolinite and quartz phases persisting in the geopolymer matrix, partially constraining strength development. Notably, the significant reduction in kaolinite peak intensity confirms its partial dissolution under alkaline conditions. The broad hump observed at approximately 2θ = 29.5° suggests the development of amorphous gel phases, such as C-S-H, C-A-S-H and N-A-S-H, in the alkali-activated CG-BFS system. These gel phases exhibit continuous transitions with no distinct boundaries [42]. Additionally, a distinct leftward shift in the diffuse peak position, compared to the original BFS’s diffuse peak at 2θ = 31°, provides clear evidence of substantial structural reorganization during geopolymerization.

SEM micrographs of the raw materials (Figure 8) show that CG and BFS particles exhibit irregular morphologies. In contrast, the composite geopolymer matrix displays a smooth, dense structure, with its surface extensively coated by gel products (Figure 9a). EDS analysis confirms that the gel phase primarily comprises O, Na, Al, Si, and Ca elements (Figure 9b). In combination with the XRD analysis, these findings indicate that the gel system might consists of calcium silicate hydrate (C-S-H), calcium aluminum silicate hydrate (C-(A)-S-H), and geopolymer gel (N-A-S-H). Notably, the measured Ca/Si ratio is less than 1, significantly lower than that of conventional Portland cement hydration products (Ca/Si > 1.5). This suggests that the C-S-H gel in the composite geopolymer exhibits a distinct structural characteristic, where some Si⁴⁺ in the silicate tetrahedral chains may be substituted by Al³⁺, leading to the generation of the C-A-S-H gel phase. Consequently, the CG–BFS-based geopolymer system represents a complex matrix where the coexistence of C-S-H, C-A-S-H, and N-A-S-H gels has been qualitatively confirmed.

4. Conclusions

This study systematically examines the effects of the BFS content, liquid-to-solid ratio, alkali activator dosage, waterglass modulus, and initial 24 h curing temperature on the mechanical properties of the CG–BFS binary geopolymer system through single-factor experiments and RSM. Additionally, XRD and SEM-EDS analyses were conducted to investigate the reaction mechanisms. The key findings are as follows:

(1)

The results from the single-factor experiments indicate that compressive strength increases monotonically with BFS content from 10% to 90%, reaching its peak at 90%. Conversely, as the waterglass modulus increases from 0.6 to 1.4, compressive strength gradually decreases, with the highest strength observed at a modulus of 0.6. The liquid-to-solid ratio (0.21–0.29) initially enhances compressive strength, peaking at 0.27, before declining at higher ratios. Similarly, the alkali activator dosage (5–15%) follows a trend of increasing strength up to 10%, after which strength decreases. Furthermore, raising the curing temperature from 30 °C to 70 °C leads to a continuous decline in strength, with the highest value recorded at 30 °C.

(2)

Building on the single-factor experiments, a three-factor, three-level RSM optimization was conducted to refine the composite geopolymer’s preparation parameters. A second-order regression model was established to describe the relationship between the response and influencing factors, with the ANOVA confirming its validity and statistical significance. The optimal parameters were identified to be a waterglass modulus of 1.06, an alkali activator dosage of 13.81%, and an initial 24 h curing temperature of 30 °C, yielding a maximum compressive strength of 91.13 MPa. XRD and SEM-EDS analyses further revealed the existence of partially unreacted CG particles in the geopolymer matrix. The primary reaction products were qualitatively identified as an amorphous mixture of C-S-H, C-A-S-H, and N-A-S-H, indicating a complex gel-phase structure.