Influences of Different Solid Waste Powders on the Drying Shrinkage Characteristics of Metakaolin-Based Geopolymers

Abstract

This study investigates the effects of three solid waste powders—fly ash (FA), silica fume (SF), and phosphogypsum (P)—on the drying shrinkage behavior of metakaolin-based geopolymers. To systematically evaluate the performance and underlying mechanisms, a comprehensive experimental program was conducted, including compressive strength and elastic modulus testing, early-age and variable-humidity drying shrinkage monitoring, mercury intrusion porosimetry, and microcalorimetry analysis. Results demonstrate that all three materials effectively reduce drying shrinkage through distinct mechanisms. The incorporation of 30% FA optimized the capillary pore network and densified the matrix, achieving a peak compressive strength of 53.51 MPa and an elastic modulus of 9.23 GPa. SF exhibited a dose-dependent effect; at an optimal content of 7%, it enhanced compressive strength by 28.3% through its nucleation effect and micro-aggregate filling. However, excessive SF (9%) led to pore coarsening and increased shrinkage. Although P incorporation slightly reduced mechanical strength, it decreased cumulative porosity by up to 8% and formed needle-like Wairakite-Ca crystals that provided micro-structural support, resulting in a net shrinkage reduction of up to 137.83 µε. This study provides a scientific basis for designing low-shrinkage, low-carbon geopolymers by tailoring solid waste incorporation to engineer multiscale pore structures.

1. Introduction

Geopolymers, as low-carbon cementitious materials, exhibit significant advantages in mechanical properties and corrosion resistance, yet their high drying shrinkage remains a critical bottleneck limiting engineering applications [1,2,3,4]. Traditional shrinkage mitigation strategies fall into two primary categories: (1) Chemical admixtures (e.g., shrinkage reducers, air-entraining agents) that weaken capillary stress by reducing pore solution surface tension. Bakharev et al. [5] demonstrated that air-entraining agents and superplasticizers reduced shrinkage in alkali-activated slag concrete by altering surface tension. Shrinkage-reducing admixtures can halve pore solution surface tension, thereby decreasing the Kelvin radius at a given relative humidity [6,7,8]. (2) Internal curing/expansion agents (e.g., superabsorbent polymers, gypsum, reactive MgO) that compensate for shrinkage through gradual moisture release or chemical expansion [9,10]. However, external additives often entail trade-offs, where excessive air entrainment introduces harmful voids that reduce strength, while superfluous expansion agents may induce cracking.

In contrast, optimizing precursor composition and activator parameters (e.g., liquid-solid ratio, NaOH concentration, Na₂SiO₃/NaOH modulus) is recognized as a more fundamental approach to shrinkage control [11,12,13]. Extensive studies confirm that geopolymer shrinkage correlates closely with compositional parameters (NaOH concentration, liquid–solid ratio, Na₂SiO₃/NaOH ratio, calcium content) and curing conditions [14,15]. Precursor type, activators, and curing directly govern geopolymerization reaction kinetics. On the one hand, gel formation dictates the matrix’s mechanical and physicochemical properties. On the other hand, incomplete space-filling by gels leaves hardened paste with interconnected pores. Moisture loss through these channels drives shrinkage mechanisms. Thus, the core strategy lies in optimizing geopolymerization to engineer multiscale pore structures—specifically capillary pore distribution, density, and connectivity.

Solid waste powders, with their compositional diversity and reactive variability, emerge as key regulators for pore-structure engineering. Fly ash (FA), a coal combustion byproduct, slows early geopolymerization kinetics, prolonging geopolymerization reaction time to facilitate denser gel formation that impedes moisture egress and reduces shrinkage [16]. Silica fume (SF), a ferrosilicon smelting residue characterized by high specific surface area (>20 m²/g) and pozzolanic activity [17,18], provides nucleation sites that accelerate geopolymerization, boosting gel formation for enhanced strength [16]. Its micro-aggregate effect densifies the matrix and refines pore structure. Critically, optimal SF content (typically ≤ 10%) is essential; excess disrupts geopolymerization reaction kinetics and compromises strength [19,20,21]. Phosphogypsum (P), a phosphate fertilizer waste composed primarily of CaSO₄·2H₂O with trace impurities, poses environmental challenges due to underutilization [22]. Studies show its incorporation in geopolymers not only offers a waste solution but also enhances compressive/tensile strength at ~5% content [23,24,25]. In metakaolin-based systems, P inhibits microcracking, increases density, strengthens the geopolymer network, and suppresses heavy metal leaching [26,27].

Despite the existing research on geopolymer composition, there is a lack of systematic comparative data on how different industrial solid wastes—specifically fly ash (FA), silica fume (SF), and phosphogypsum (P)—differently influence drying shrinkage under variable humidity conditions. While traditional additives often prioritize a single performance metric, the combined impact of these wastes on reaction kinetics and pore structure development remains insufficiently quantified. The practical innovation of this study lies in the systematic correlation established between the reaction heat, pore structure evolution, and macroscopic deformation of metakaolin-based geopolymers (MKBGs) modified by these three distinct wastes. By evaluating the specific thresholds for each additive (e.g., 30% for FA, 3% for SF, and >3% for P), this study provides a direct experimental basis for optimizing geopolymer stability while maximizing the utilization of industrial byproducts.

2. Test Program

2.1. Test Materials

(1) Metakaolin Raw Material

The metakaolin used in this study was K-1300 high-reactivity metakaolin from the KAOPOZZ series of Chaopai Brand, produced in Hohhot, Inner Mongolia, China. Its particle size passed through a 4000-mesh sieve, its density was 2700 kg/m³, and its specific surface area was approximately 1.112 m²/g. The main chemical compositions are presented in

Table 1. Chemical composition of metakaolin.

Oxides/%	SiO2	Al2O3	CaO	SO3	TiO2	K2O	L.O.I
Metakaolin	52.53	45.42	0.26	0.04	0.97	0.18	0.6

. The XRD pattern of metakaolin is shown in Figure 1.

(2) Solid Waste Powders

The solid waste materials utilized in this study, namely fly ash (FA), silica fume (SF), and phosphogypsum (P), exhibit a synergistic relationship characterized by distinct gradients in chemical composition, mineral structure, and particle size distribution. In terms of chemical evolution, the active SiO₂ derived from the amorphous spherical particles of SF facilitates early-stage nucleation, while the Al₂O₃ provided by FA serves as a primary precursor for the formation of the N-A-S-H framework. Simultaneously, the CaO introduced by P leads to the development of a Wairakite-Ca supporting structure, leveraging the high degree of crystallinity dominated by gypsum (calcium sulfate dihydrate) to enhance the structural integrity of the matrix.

From a physical perspective, these raw materials establish a well-defined particle size gradient, the specifics of which are summarized alongside their chemical compositions and XRD patterns in

Table 2. Chemical composition of solid waste powders (%).

Oxides/%	SiO2	Al2O3	CaO	MgO	SO3	TiO2	K2O	Fe2O3	Na2O	P2O5	L.O.I
Fly ash	56.9	31.24	3.05	0	0	1.34	2.06	3.8	0	0	1.61
Silica fume	97.08	0.34	0.07	0.06	0.02	0.04	0.04	0.37	0.06	0	1.92
Phosphogypsum	7.53	1.15	38.18	0.18	50.36	0.14	0.39	0.58	0.10	1.12	0.27

, Figure 1 and Figure 2. As illustrated by these data, SF represents the finest fraction with a distribution concentrated between 0.1 μm and 1 μm, whereas FA—an industrial waste composed of an amorphous glass phase and crystalline mullite—exhibits a broader distribution peaking in the 20–30 μm region. P displays a distribution peak around 40–60 μm as the coarsest component. This physical stratification, combined with the diverse diffraction characteristics of the amorphous and crystalline phases, ensures dense particle packing and a robust chemical environment for the resulting geopolymer.

(3) Alkaline Activator

The alkaline activator used in this study was prepared by mixing a sodium silicate solution (Na₂O: 8.35%, SiO₂: 26.54%, H₂O: 65.11%), flake sodium hydroxide solids (food-grade, purity ≥ 99.9%), and deionized water. After mixing, the alkaline activator was immediately sealed with plastic film to prevent moisture evaporation. It was then magnetically stirred for at least 6 h, followed by standing to ensure complete homogenization, and finally cooled to room temperature prior to use.

2.2. Specimen Mix Proportioning and Preparation

The control MKBG without solid waste additives (designated MKBG-S0) was prepared with an activator modulus of 1.2, a liquid-to-solid ratio of 1.2, and an activator concentration of 35%. These baseline parameters were established through preliminary investigations to ensure an optimal balance between mechanical performance and fluidity. The specific mix ratio is shown in

Table 3. Content levels and proportions of powder materials.

Material	Replacement Ratio (% by Mass)	Group Designation
Fly ash	10%/20%/30%/40%	10%FA/20%FA/30%FA/40%FA
Silica fume	1%/3%/5%/7%	1%SF/3%SF/5%SF/7%SF
Phosphogypsum	0.5%/1%/3%/5%	0.5%P/1%P/3%P/5%P

. The dosages of solid waste powders were subsequently determined through pre-tests to evaluate their specific impacts.

Pre-weighed metakaolin was mixed with the alkaline activator (cooled to room temperature). The paste was stirred at low speed (140 ± 5 rpm) for 1 min, followed by high-speed stirring (285 ± 5 rpm) for 2 min to ensure homogeneity. The uniform slurry was poured into molds and compacted on a vibration table for 3 min to eliminate air bubbles. Specimens were transferred to a standard curing chamber (20 °C, RH ≥ 90%) for 24 h before demolding. Demolded specimens were returned to the curing chamber until reaching the required testing age.

2.3. Testing Methods

(1) Compressive Strength

Testing followed the JGJ/T70-2009 [28]. Cubic specimens (70.7 mm × 70.7 mm × 70.7 mm) were cast in triple PVC molds. For each mix proportion, three specimens were prepared, and the average compressive strength was reported.

(2) Elastic Modulus

Testing complied with the GB50081-2002 [29]. Prismatic specimens (40 mm × 40 mm × 160 mm) were molded in triple PVC molds. Six specimens were cast per mix: three were tested for failure load, and the maximum load for elastic modulus testing was calculated as one-third of the average failure load. The remaining three specimens underwent elastic modulus measurement, with the average value recorded.

(3) Early-Age Shrinkage

To decouple autogenous/chemical shrinkage from drying shrinkage, early-age deformation was monitored per ASTM C1698-09 [30]. Fresh paste was poured into sealed corrugated PVC tubes (Ø29 mm × L420 mm), vibrated to eliminate air pockets, and cured at 20 ± 2 °C and RH ≥ 95%. Linear shrinkage was measured continuously for 7 days, starting at 24 h. For each mix, three replicate specimens were tested simultaneously, and the average value was used for analysis. The transition point to stabilized autogenous shrinkage defined the onset for drying shrinkage tests.

(4) Drying Shrinkage under Variable Humidity

Testing adhered to the JC/T 603-2004 [31]. Prismatic specimens (25 mm × 25 mm × 280 mm) were cast in triple cast-iron molds and cured at 20 ± 2 °C and RH ≥ 95% for 24 h. After demolding and initial curing to the shrinkage onset time determined in (3), specimens were unwrapped and exposed to stepwise humidity reductions from 90% RH to 30% RH. Three specimens were tested for each environmental condition, and the average values for length change and mass loss were recorded every 5 days. Humidity was decreased by 20% RH when sequential shrinkage increments dropped below 100 µm/m (100 µε).

(5) Geopolymerization Reaction Heat Analysis

Hydration heat was monitored using a TAM Air microcalorimeter (TA Instruments, New Castle, Delaware, USA; operational range: 20–90 °C). A mixture of aluminosilicate powder (1 g) and alkaline activator (mass calculated per liquid-solid ratio) was loaded, and heat flow was recorded automatically for ≥24 h. Triplicate measurements were conducted for each group to ensure consistency, and the most representative heat evolution profile was selected for analysis.

(6) Pore Structure Characterization

Mercury intrusion porosimetry (MIP) was performed using a PoreMaster 33GT analyzer (Quantachrome Instruments, Boynton Beach, FL, USA; max pressure: 33,000 psi; measurable pore range: 5 nm–1080 µm). To ensure reproducibility, each group was tested three times, and a representative dataset was chosen for detailed characterization.

3. Results and Analysis

3.1. Effect of Fly Ash Incorporation on Drying Shrinkage

Figure 3 shows the XRD patterns of MKBGs with varying fly ash (FA) contents. The incorporation of FA did not generate new crystalline phases; the matrix remained predominantly amorphous aluminosilicate. The observed mullite peaks originated from the crystalline phase present in the raw FA.

Figure 4 illustrates the early-age shrinkage of geopolymers with different FA contents. Shrinkage at Day 1 decreased progressively with higher FA content. However, shrinkage at Day 7 did not follow this trend. The control sample stabilized after Day 2, exhibiting a daily shrinkage increment < 50 µε. In contrast, samples with 10–40% FA continued shrinking until Day 4, when their daily increment dropped below 50 µε, indicating prolonged chemical shrinkage kinetics. Early-age shrinkage stabilized by Day 8, establishing this time point as the baseline for drying shrinkage tests.

Figure 5 demonstrates drying shrinkage behavior under stepped humidity reduction (90% → 50% RH). FA incorporation significantly reduced drying shrinkage, with higher FA contents yielding progressively greater reduction—though the magnitude of improvement diminished beyond 30% FA. This confirms that FA mitigates shrinkage within an optimal range, beyond which efficacy plateaus. In the 70–50% RH range, control samples completed most shrinkage within 5 days, whereas FA-modified samples required >10 days to stabilize, suggesting FA reduces moisture sensitivity by decelerating capillary stress development.

Figure 6 and Figure 7 respectively show the pore size distribution and cumulative porosity of MKBGs with different FA contents. The incorporation of FA shifted the pore size distribution towards smaller sizes and progressively decreased cumulative porosity. When the FA content was 10–20%, the pore size distribution retained a unimodal characteristic, indicating that metakaolin remained the primary reactant consumed in geopolymerization. However, FA incorporation promoted the geopolymerization reaction of metakaolin, resulting in material densification. Microcalorimetry results in Figure 8 revealed that with increasing FA content, the early-stage exothermic peak decreased significantly, but sustained heat release after 12 h exceeded that of the FA-free MKBG. Fly ash exhibits lower alkali-activation reactivity than metakaolin, leading to slower dissolution rates in alkaline activators at ambient temperatures. Increased FA content reduced the early-stage geopolymerization reaction rate and total heat release. This slower geopolymerization reaction rate prolongs the polymerization stage, promoting densification of the gel before hardening and facilitating the formation of geopolymer paste with smaller pore sizes and lower porosity [16]. When the FA content reached 30%, the pore structure exhibited coarsening and the pore size distribution range widened. At a 40% content, the pore size distribution displayed a distinct bimodal characteristic typical of FA-based geopolymers. The secondary peak originated primarily from cavities formed by incomplete dissolution of larger FA glassy particles, confirming substantial participation of FA in geopolymerization.

Figure 9 shows the 28-day compressive strength and elastic modulus of MKBGs with varying FA contents. Both properties initially increased with FA content and then decreased. At 30% FA content, peak values of 53.51 MPa (compressive strength) and 9.23 GPa (elastic modulus) were achieved. This enhancement stemmed from FA-induced gel densification during geopolymerization and its micro-aggregate effect filling existing pores in the matrix. At 40% FA content, reduced geopolymerization reaction extent caused decreased mechanical properties. Despite lower deformation resistance, drying shrinkage remained comparable to the 30% FA specimen due to stable overall porosity, coarsened pore structure reducing micro-capillary pores, and weakened capillary driving force.

3.2. Effect of Silica Fume Incorporation on Drying Shrinkage

Figure 10 shows the XRD patterns of MKBGs with varying SF contents. SF primarily consists of silica. As shown in Figure 10, with increasing SF content, the intensity of crystalline silica peaks increased moderately, but no new phases formed. The matrix remained predominantly amorphous.

Figure 11 shows early-age shrinkage of metakaolin-based geopolymers with different SF contents. Early-age shrinkage decreased significantly with higher SF content. MKBGs inherently exhibit lower autogenous shrinkage than cement or other alkali-activated materials. SF reduces early shrinkage primarily by enhancing paste strength and shrinkage resistance. Figure 12 presents drying shrinkage under stepped humidity reduction (90% to 50% RH). SF incorporation altered the drying shrinkage pattern. At ≤5% content, final shrinkage at 50% RH remained comparable to the control. At 9% SF content, shrinkage at 50% RH increased by 228.45 µε relative to control sample. SF amplified humidity sensitivity above 50% RH, with most shrinkage occurring at 70% RH. Minimal additional contraction occurred during further drying to 50% RH. This phenomenon—similarly observed in FA-based geopolymers—may arise from the spherical morphology of SF/FA particles, which creates smoother pore surfaces facilitating rapid moisture transport. Consequently, capillary water depletes predominantly at 70% RH.

Figure 13 and Figure 14 respectively show the pore size distribution and cumulative porosity of MKBGs with different SF contents. SF incorporation refined the pore structure of geopolymer. The most probable pore diameter decreased. Cumulative porosity decreased. The most probable pore diameter gradually decreased from 17.97 nm to 9.93 nm with increasing SF content. Porosity showed a different trend. At only 3% SF content, porosity decreased from 39.09% to 31.79%. With continued increase in content, porosity decreased by only 1.18%.

Microcalorimetry test results in Figure 15 showed that with increasing SF content, early heat release and exothermic peak intensity changed little. They first increased and then decreased. At 7% SF content, both values reached their maximum. The smaller particle size of SF provides a larger specific surface area. This allows greater contact area with the activator. It facilitates thorough dissolution of SF by the activator. The mixed liquid phase has strong alkalinity. In this environment, SF dissolution is higher than when used in cement [32]. Simultaneously, SF particles act as nucleation sites that promote the geopolymerization reaction to form a finer and denser matrix [18].

At 9% SF content, the early exothermic peak decreased, while geopolymer porosity and most probable pore diameter increased, indicating slight pore structure coarsening. This occurred because during activator dissolution of raw materials, the precursor released excessive silicate ion groups. These groups not only maintained the alkalinity of the mixed liquid phase but slowed metakaolin dissolution and disrupted subsequent polycondensation. Consequently, the final geopolymerization extent decreased, causing pore coarsening. From 3% SF content onward, reduced heat release during the 2–8 h geopolymerization reaction period was observed. This also resulted from excessively high silicate ion concentration in the mixed liquid phase during the mid-to-late geopolymerization reaction stages.

Figure 16 shows the variation in compressive strength and elastic modulus of MKBGs with different SF contents. Both properties increased initially with SF content and then decreased. At 7% SF content, compressive strength and elastic modulus reached peak values. Compared to the control sample, compressive strength increased by 28.3% and elastic modulus increased by 12.4%. At this content, the paste exhibited maximum resistance to shrinkage deformation. Optimal SF incorporation enhanced geopolymerization to densify the matrix while acting as micro-aggregate filler, thereby improving mechanical performance. This benefits geopolymer drying shrinkage control. Notably, SF incorporation reduces porosity in MKBGs but shifts pore size distribution toward finer diameters. This increases capillary stress during moisture loss from micro-capillary pores, potentially increasing drying shrinkage. These two aspects exhibit competing effects. At 3% SF content, shrinkage resistance improved compared to the control sample. Limited pore refinement occurred, causing marginal capillary stress increase. Consequently, final shrinkage reached its minimum value. At 9% SF content, shrinkage resistance decreased while pore structure maintained fine-scale distribution, resulting in maximum value within the test group.

3.3. Effect of Phosphogypsum Incorporation on Drying Shrinkage

Figure 17 shows the XRD patterns of MKBGs with varying P contents. P incorporation introduced new crystalline phases: Dihydrate calcium sulfate, Dicalcium silicate, Tricalcium silicate, and Wairakite-Ca. The raw P primarily contained CaSO₄·2H₂O and SiO₂, explaining the emergence of gypsum peaks. P introduced calcium and sulfur into the geopolymer system. Sulfur formed sodium/potassium sulfides, while calcium reacted with silicate groups under alkaline conditions to generate larnite and tricalcium silicate. These calcium silicates acted as final geopolymerization reaction products, enhancing the matrix through filler effects without further hydration to form C-S-H gel (unlike in cement systems). Geopolymerization remained dominant, with N-A-S-H gel encapsulating the calcium silicates and inhibiting hydration. This explains the absence of C-S-H gel despite the presence of calcium silicate minerals. Wairakite-Ca, a needle-like/fibrous calcium silicate mineral (structurally similar to ettringite), may induce expansion in cementitious systems. Its acicular crystals exhibit high surface area for water binding (increasing crystal volume) and form a micro-supporting framework that resists shrinkage.

Figure 18 presents early-age shrinkage results. P incorporation slowed early shrinkage kinetics. On Day 2, shrinkage decreased progressively with higher P content, reaching a 214.34 µε reduction at 5% P content versus the control sample. Unlike the control sample, P-incorporated specimens exhibited continuous shrinkage until Day 7, indicating prolonged geopolymerization reaction. By Day 8, ≤3% P contents showed negligible impact on final shrinkage, while ≥3% P contents reduced shrinkage by up to 137.83 µε. Figure 19 displays drying shrinkage under humidity reduction. P did not alter the shrinkage pattern, with significant contraction at both 70% and 50% RH. However, shrinkage decreased consistently with higher P content across all RH levels, confirming its efficacy in reducing drying shrinkage.

Figure 20 and Figure 21 respectively show the pore size distribution and cumulative porosity of MKBGs with different P contents. The most probable pore diameter decreased with increasing P content but remained within the 10–20 nm range. P incorporation did not alter the pore size distribution pattern of MKBGs. This confirmed geopolymerization remained the dominant reaction in the mixture. Cumulative porosity decreased with higher P content, and the maximum reduction reached 8%. Under similar pore size distribution conditions, reduced porosity decreased capillary pore density. This lowered drying shrinkage.

Microcalorimetry test results in Figure 22 showed the early exothermic peak decreased with increasing P content. At 5% P content, the peak value was the lowest in the group. P contained residual incompletely decomposed phosphate rock and phosphoric acid. These acidic substances weakened the activation effect of alkaline activators on raw materials. This caused the dissolution exothermic peak to decrease. During subsequent sustained heat release, the heat release during polycondensation continuously decreased in the control sample. In contrast, the P-incorporated geopolymer exhibited longer and more stable heat release during this stage. This resulted from two factors. First, similar to cement systems, substantial Ca²⁺ was dissolved and released from P. Second, P reduced the alkalinity of activators. This prolonged the geopolymerization process and increased the sustained heat release duration.

Figure 23 shows the variation in compressive strength and elastic modulus of MKBGs with different P contents. Both properties decreased with increasing P content. Compressive strength declined from 43.20 MPa to 36.65 MPa. Elastic modulus decreased from 8.55 GPa to 7.27 GPa. P incorporation weakened the activation efficiency of alkaline activators. This reduced the geopolymerization extent and decreased N-A-S-H gel formation. These factors inevitably lowered mechanical performance. Notably, porosity decreased despite reduced mechanical properties. This likely resulted from filler effects of P and precipitated silicates. Increased Wairakite-Ca formation contributed to this effect. Its needle-like structure provided microstructural support beneficial for shrinkage inhibition.

3.4. Comparative Analysis of Solid Waste Powders on Drying Shrinkage

The comparative data reveals that while fly ash (FA), silica fume (SF), and phosphogypsum (P) all contribute to the densification of the metakaolin-based geopolymer (MKBG) matrix, they operate through distinct physical and chemical mechanisms, as comprehensively summarized in

Table 4. Comparative analysis of admixture impacts on MKBG performance.

Admixture Type	Structural Mechanism	Pore Distribution Effect	Impact on Mechanical Strength	Shrinkage Suppression Efficacy
Fly Ash	Micro-aggregate effect and N-A-S-H densification	Decreased connectivity with a shift to smaller diameters	Significant increase peaking at 30% dosage	High efficacy with reduced moisture sensitivity
Silica Fume	Nucleation sites and intense pozzolanic activity	Drastic refinement of the most probable pore diameter	Moderate increase peaking at 7% dosage	Optimal at 3% but limited at higher dosages
Phosphogypsum	Wairakite-Ca framework and physical filler effect	Reduced cumulative porosity with stable distribution	Progressive decrease due to weakened activation	Consistent reduction across all humidity levels

. SF demonstrates the most aggressive pore refinement capability by significantly shifting the most probable pore diameter toward a finer range. This phenomenon is consistent with the findings of Appa Rao [33], who reported that SF addition significantly intensifies early-age drying shrinkage—potentially by 7–10 times—due to its high pozzolanic reactivity and radical pore size refinement. Furthermore, Zhang et al. [34] identified that SF optimizes the internal structure of geopolymers by refining detrimental capillary pores (ranging from 100 nm to 10 μm) and modifying the binding energy of the Si-O-T (T: Si or Al) framework. However, this creates a competing effect where the resulting increase in capillary stress can counteract the benefits of a denser matrix at higher dosages.

Conversely, FA reduces the connectivity of the capillary network and lowers moisture sensitivity without the extreme pore refinement seen in SF, leading to a more stable reduction in drying shrinkage. According to Yang et al. [16], the substitution of metakaolin with FA reduces the overall reactivity of the precursor system, which decelerates the reaction rate and effectively restricts water evaporation from the pore network. The spherical FA particles also serve as “micro-aggregates” that improve particle packing while maintaining lower moisture loss. In the case of P, the reduction in shrinkage is primarily attributed to the physical filler effect and the formation of a needle-like Wairakite-Ca framework, which provides internal structural support despite a measurable decrease in the overall geopolymerization extent.

The reaction kinetics and mechanical responses also distinguish these materials. FA and SF at optimal levels enhance both compressive strength and elastic modulus, thereby increasing the resistance of the hardened paste to shrinkage-induced deformation. Specifically, the incorporation of SF has been shown to enhance the density of the gel phase and improve the 28-day mechanical performance, whereas FA provides a long-term stabilization of the matrix despite an initial reduction in the geopolymerization rate. In contrast, the acidic nature of P weakens the alkaline activation efficiency, leading to a progressive decline in mechanical properties. However, P remains highly effective at suppressing shrinkage because the porosity reduction and mineral support outweigh the loss in deformation resistance. While SF-modified samples complete most of their contraction at high relative humidity due to rapid moisture transport facilitated by spherical particles, FA-modified samples exhibit a decelerated development of capillary stress, making them less sensitive to humidity fluctuations.

4. Conclusions

This study improved the drying shrinkage of MKBGs by enhancing matrix performance and optimizing pore structure characteristics. Fly ash (FA), silica fume (SF), and phosphogypsum (P) were incorporated individually, and their mitigation mechanisms were analyzed through strength, shrinkage, Mercury Intrusion Porosimetry (MIP), and microcalorimetry tests. The key findings are as follows:

(1)

FA incorporation significantly reduces drying shrinkage in MKBGs, serving as an effective shrinkage-suppressing additive. An optimal 30% FA content resulted in a peak compressive strength of 53.51 MPa and an elastic modulus of 9.23 GPa. Mechanistically, FA slows early-stage geopolymerization kinetics, allowing for a denser gel structure that reduces daily shrinkage increments to below 50 με by Day 4. Simultaneously, it reduces connectivity within the capillary pore network, thereby decreasing humidity sensitivity during the drying process.

(2)

SF exhibits a dual effect on shrinkage behavior. While a 7% dosage provides the highest mechanical boost—increasing compressive strength by 28.3% and elastic modulus by 12.4% compared to the control—it also refines the most probable pore diameter from 17.97 nm to 9.93 nm, which can increase capillary stress. Consequently, a 3% dosage is identified as the optimal balance for enhancing matrix resistance without detrimental pore refinement. Excessive SF hinders the geopolymerization reaction, weakening shrinkage resistance and ultimately increasing total shrinkage.

(3)

P incorporation effectively reduces drying shrinkage within the tested dosages. At a 5% content, it reduced early-age shrinkage by 214.34 με compared to the control and decreased total porosity by up to 8%. Although its acidity slightly weakened the activator—reducing compressive strength from 43.20 MPa to 36.65 MPa—the formation of needle-like Wairakite-Ca crystals provides a micro-supporting framework. This structural reinforcement, combined with the reduction in capillary driving forces, results in a significant net reduction in shrinkage.