Foundation backfill prepared from inferior desulfurization gypsum and fly ash: laboratory evaluation and field application

1 Introduction

Nowadays, the method of Wet Limestone-gypsum Flue Gas Desulphurization is the most widely used in the world for coal-fired power plants (Nie et al., 2019). With the continuous optimization of China’s energy structure and the deepening implementation of environmental policies, the large-scale application of flue gas desulfurization (FGD) technology in coal-fired power plants—particularly the limestone-gypsum wet desulfurization process—has led to a steady increase in the production of FGD gypsum and fly ash. The waste produced by the method is flue gas desulfurization gypsum in the form of calcium sulfate (Koralegedara et al., 2019; Ma et al., 2020). China’s annual production of desulfurization gypsum has exceeded 70 million tons since 2012 (Fu et al., 2019; Wu et al., 2019), and has exceeded 100 million tons by 2022 (Xiao et al., 2022).

As a gigascale material, because of its bulk mineralogy and chemistry, fly ash can serve as a source of materials for large-volume, low tech applications. In recent years, the low activity fly ash is used as a base material more and more (Hu et al., 2018; Koshy et al., 2019). However, the desulfurization gypsum mineral composition is complex, in addition to containing gypsum, but also contains other substances. Actually, the nature is unstable, and cannot meet the processing and use requirements of industrial materials or building materials, which has not been fully utilized and has been discarded in landfills (Bai et al., 2023). They not only occupy a lot of land, but also may cause soil acidification and secondary pollution of water sources. Coal fly ash and flue gas desulfurization gypsum are worthy of consideration because they are utilized to a still unsatisfactory extent (Balsamo et al., 2012; Montagnaro and Santoro, 2019).

Many scholars have made fruitful achievements in the basic research of polymer cement and fly-ash cement system (Davidovits, 1994; He et al., 2023). Duxson et al. (2007) elaborated the inorganic polymer green concrete technology. Guo et al. (2013) formed flue gas desulphurization geopolymer (FGDG) using high calcium fly ash and desulfurization gypsum as raw materials (Telesca et al., 2013; Bai et al., 2024). Bai et al. (2024) summarized beneficial FGDG applications that have been deemed to pose minimal environmental concern, emphasizing their principles, research gaps, and potential developments, with the aim of increasing the reuse rate of FGDG.

Hansen and Sadeghian (2020) studied the strength characteristics of cementitious materials after cement is mixed with different ratios of desulfurization gypsum and fly ash. Jang et al. (2024a) utilized glass powder and steel slag as a solid precursor material along with fly ash as a key ingredient to produce geopolymer composite. Geopolymer composites exhibit excellent resistance to high temperature; their compressive strength remains significant, and their toughness to high temperature exposure is demonstrated (Janga et al., 2024b; Bai et al., 2026). Wang et al. (2023) optimized the performance of GGBS-FA binary geopolymer; a ternary geopolymer was designed by the addition of desulfurization gypsum (DG). The compressive strength of GGBS-FA-DG ternary geopolymer and the influence of content change on strength were tested. Zhou et al. (2021) discussed the feasibility of preparation of MPPC by incorporation of phosphogypsum, determined the early properties of MPPC slurry, compressive strength and microstructure properties after hardening, and discussed the hydration mechanism. Recent studies have further highlighted the broader sustainability potential of upcycling diverse waste streams into circular construction materials. For instance, fully recycled concrete and brick powders, autoclaved aerated concrete waste, sand washing waste, and accelerated carbonation bio-composites have been successfully engineered into cement-free geopolymers, yielding materials with highly optimized mechanical, rheological, and structural attributes (Moodi et al., 2025; Mollaei et al., 2023; Hanafi et al., 2025; Hanafi et al., 2026a; Rahmani et al., 2026; Hanafi et al., 2026b). Despite these advances, the existing literature is predominantly descriptive and focuses heavily on high-purity, well-crystallized FGD gypsum used as a minor additive for structural building materials. These studies do not adequately address the unique challenges posed by long-term stockpiled, low-quality industrial by-products.

Geopolymer is a kind of three-dimensional reticular gel formed by the joint of silica tetrahedron and aluminoxy tetrahedron by shared oxygen atom (Nie et al., 2015; Bai et al., 2025). It possesses excellent mechanical properties, high temperature resistance, corrosion resistance, and solid sealing ability against heavy metal ions. However, most existing research has primarily focused on high-purity, well-crystallized, and relatively stable FGD gypsum (Nie et al., 2016). For inferior, low-activity, and poor-stability materials stored for a long time, such as inferior desulfurization gypsum, current application research is mainly limited to chemical activation or utilizing them merely as minor admixtures in other building materials. There is a distinct lack of systematic studies investigating their physicochemical and material properties when utilized alone as a primary backfill material (Marinković et al., 2017; Iribarne et al., 2001). Consequently, combining inferior FGD gypsum with fly ash under alkaline activation presents an optimal technical route: the highly alkaline environment accelerates the dissolution of aluminosilicate phases in fly ash, generating a geopolymeric network capable of chemically encapsulating and stabilizing the deleterious components inherent in the inferior gypsum. Compared to conventional Controlled Low-Strength Materials (CLSM) or traditional cement-stabilized soils, utilizing 100% solid waste-derived geopolymers for backfilling offers a promising pathway for large-scale industrial waste utilization and reduces the consumption of natural quarried aggregates. However, fully validating this transition requires a multi-scale perspective. Therefore, combining microstructural analysis with macroscopic shear strength testing and large-scale field validation is strictly necessary to bridge the gap between chemical stabilization mechanisms and practical geotechnical reliability.

This study aims to investigate how alkaline activation effectively stabilizes the volumetrically unstable phases of inferior FGD gypsum when combined with fly ash, and to critically evaluate whether the resulting geopolymer satisfies the rigorous load-bearing and deformation criteria required for factory foundation backfilling. Through an integrated approach of laboratory evaluation and large-scale field monitoring, this study establishes a systematic methodology with two primary objectives: investigating the potential microstructural stabilization of volumetrically unstable inferior FGD gypsum, and validating the macroscopic engineering feasibility of this solid waste-derived formulation as a large-scale foundation backfill.

2 Microstructure characteristics of desulfurization gypsum and fly ash

2.1 Components of desulfurization gypsum and fly ash

The elemental composition of FGD gypsum is relatively simple (Table 1), primarily consisting of sulfur (S), calcium (Ca), and oxygen (O), with SO₃ and CaO as its main chemical components. The mineral composition is predominantly gypsum (CaSO₄·2H₂O) and calcium sulfite hemihydrate (CaSO₃·0.5H₂O), with gypsum accounting for 78.8%, which is lower than the 85% requirement specified in the Flue Gas Desulfurization Gypsum Standard (GB/T 37785–2019, Table 2). Additionally, calcium sulfite hemihydrate accounts for 17.2%, exceeding the standard’s 0.5% threshold. This material is classified as inferior FGD gypsum and does not meet the grading criteria. The results of the Energy Dispersive X-ray Spectroscopy (EDX) and X-ray Diffraction (XRD) analyses of FGD gypsum are shown in Figure 1.

TABLE 1. The elemental composition of desulfurization gypsum.

TABLE 2. Technical requirements for desulfurization gypsum.

From the XRD analysis of FGD gypsum, peak values appear in the range of 10°–30° (2θ), with narrow peak widths, indicating high crystallinity and relatively large crystal grains. The diffraction peaks are primarily associated with gypsum and calcium sulfate compounds. The presence of Ca²⁺ and SO₄ ^2- provides the necessary conditions for the formation of ettringite (3CaO·Al₂O₃·3CaSO₄·32H₂O), which enhances the feasibility of geopolymer formation when combined with fly ash.

The XRD analyses of fly ash are presented in Figure 2, and the elemental and mineral compositions are summarized in Tables 3, 4. From Table 4, the elemental composition of fly ash is relatively complex, typically containing aluminum (Al), silicon (Si), calcium (Ca), iron (Fe), oxygen (O), and amorphous carbon (C). Its main chemical components include Al₂O₃, SiO₂, and FeO, which together account for approximately 96.72%. These components provide essential reactants such as Al³⁺, SiO₄ ^4-, CO₃ ^2-, and Fe³⁺ for the geopolymerization reaction with inferior FGD gypsum under alkaline conditions.

TABLE 3. The mineral composition of desulfurization gypsum.

TABLE 4. The elemental composition of fly ash.

From Figure 2, the XRD analysis of fly ash reveals peak values within the 15°–40° (2θ) range, indicating an overall amorphous nature with some crystalline phases. The primary mineral components include mullite (Al₆Si₂O₁₃), gypsum (CaSO₄·2H₂O), and quartz (SiO₂). Mullite accounts for 78.3%, while quartz comprises 17.1%. Mullite (Al₆Si₂O₁₃): A silicoaluminate mineral characterized by a high melting point, excellent thermal stability, and strong mechanical properties, which contribute to enhancing the mechanical strength, thermal stability, and chemical resistance of geopolymers. Quartz (SiO₂): In a highly alkaline environment, the crystalline structure of quartz partially dissolves, releasing reactive SiO₂ that participates in the geopolymerization reaction. Quartz not only serves as a filler but also acts as a supplementary silicon source, influencing the final properties of the geopolymerized material.

To provide a more comprehensive chemical characterization of the raw materials, their bulk chemical compositions were quantitatively evaluated. As summarized in Table 5, X-ray fluorescence (XRF) analysis reveals the precise oxide compositions of the precursors. Additionally, the loss on ignition (LOI) was determined to be 19.59% for the inferior FGD gypsum and 10.65% for the fly ash. These precise chemical characteristics provide a solid foundational understanding of the reactive components available for the subsequent geopolymerization process.

TABLE 5. Chemical composition of the desulfurized gypsum and fly ash (wt%).

It should be noted that the inferior FGD gypsum utilized in this laboratory study was sourced from a single representative batch. This approach was intentionally selected to strictly control material variables and establish the fundamental baseline mechanisms of alkaline activation and mixture optimization. However, in practical engineering scenarios, inferior FGD gypsum exhibits inherent heterogeneity due to variations in storage conditions, source compositions, and aging histories. To mitigate the impact of this raw-material variability on large-scale reproducibility and practical scalability, the proposed methodology relies on robust alkaline activation (which chemically buffers fluctuations in precursor reactivity) and dynamic moisture adjustment based on standard compaction tests (which accommodates the highly variable natural moisture content of stockpiled gypsum). These strategies collectively ensure that the proposed backfill material can maintain stable engineering performance despite the inherent heterogeneity of industrial solid wastes.

2.2 Microstructure of desulfurization gypsum and fly ash

The microstructures of FGD gypsum and fly ash are shown in Figures 3, 4. The microscopic morphology of FGD gypsum exhibits well-developed plate-columnar crystals composed of Ca(SO₄) (H₂O)₂ (gypsum). These crystals maintain a specific length-to-diameter ratio, with relatively rough surfaces and an irregular stacking pattern. The arrangement of the crystals lacks a distinct order, showing interwoven and overlapping formations, which results in the presence of pores and voids. Additionally, a certain amount of smaller needle-like or columnar crystals, identified as CaSO₃·0.5H₂O (calcium sulfite hemihydrate), are interspersed within the structure. These appear as relatively thin, layered structures. Due to its redox properties, calcium sulfite hemihydrate undergoes volume changes and crystallographic transformations during phase transitions, which in turn affect the overall chemical and physical stability of the material.

Fly ash primarily consists of spherical particles of varying sizes, ranging from a few micrometers to several hundred micrometers in diameter. These particles exhibit relatively smooth surfaces and a generally homogeneous amorphous structure. Some of the spherical particles contain internal voids, forming hollow structures, which result from the expansion and escape of volatile substances during combustion. Additionally, various surface deposits, including fine crystalline structures, mineral fragments, and residual unburned carbon particles, are observed. These surface attachments can influence the chemical composition and physical properties of fly ash, thereby affecting its performance in geopolymerization reactions.

3 Low-strength geopolymer using inferior FGD gypsum and fly ash

3.1 Experimental design

The feasibility of polymer formation under alkaline activation conditions was analyzed. The selected DG:FA ratios (70:30, 50:50, and 30:70) were designed to cover a broad range of precursor proportions to systematically evaluate the transition from gypsum-dominated to fly ash-dominated behaviors. The final optimal mixture was determined through a multi-criteria evaluation approach, integrating macroscopic mechanical performance (compressive modulus and shear strength), environmental durability (water stability and volumetric expansion), and practical field requirements (compactability). NaOH and sodium silicate were chosen as the most conventional and representative alkaline activators to assess the efficiency of the geopolymerization process. Three different composite material mixtures were prepared with FGD gypsum to fly ash mass ratios of 70:30, 50:50, and 30:70. The samples were tested under three activation conditions: without an activator (pure water solution), with sodium hydroxide (NaOH) solution, and with sodium silicate (water glass) solution. The material proportions are based on mass ratios, and the specific experimental mixture designs are shown in Table 6. Here, desulfurization gypsum is abbreviated DG, and fly ash is abbreviated FA. The NaOH solution added is the percentage of the total mass of the mixed material accounted for, and the concentration of NaOH solution is 30%. For clarity, the activator dosages discussed in the text (e.g., 1% and 3% NaOH) correspond strictly to the mass percentages of the dry solid precursors, which explicitly translates to the absolute mass values listed in Table 6.

TABLE 6. Mixing ratio design and hydration parameters (%) (Curing conditions: 20 °C ± 2 °C, RH ≥ 95%).

Because this mixture is designed as a compacted geotechnical foundation backfill rather than a flowable geopolymer slurry, a fixed liquid-to-solid ratio is not applicable. The water introduced by the activator is strictly reproducible based on the fixed 30% concentration of the NaOH, solution. Any additional moisture required for specimen preparation was adjusted dynamically to achieve the optimal moisture content (OMC) and maximum dry density via standard compaction tests, which effectively accommodates the natural moisture fluctuations inherent in stockpiled inferior FGD, gypsum.

To ensure experimental reproducibility, a standardized preparation and curing procedure was explicitly defined. First, the specified masses of dry precursors (DG and FA, as detailed in Table 6) were mechanically dry-mixed to achieve homogeneity. Subsequently, the alkaline activator was introduced. The activator dosage (e.g., 1% or 3%) strictly represents the mass ratio of the dry active solid (NaOH) to the total dry precursors. The solid NaOH was pre-dissolved in water to form a solution with a fixed concentration of 30% before being uniformly mixed into the dry powder. Unlike flowable geopolymer slurries that utilize a fixed liquid-to-solid ratio, this geotechnical backfill material requires mechanical compaction. Therefore, any supplementary water necessary for specimen preparation was calculated and added dynamically to ensure the final mixture reached the optimal moisture content (approximately 22%–25%) determined via standard compaction tests. The well-mixed moist materials were then statically compacted into standard ring cutters (Φ61.8 mm × 20 mm) at their maximum dry density. Following compaction, the sealed specimens were transferred to a standard curing room and cured at a temperature of 20 °C ± 2 °C and a relative humidity of ≥95% until reaching their designated testing ages.

For all mechanical and physical tests (including compressive modulus, direct shear, and volumetric strain), each data point represents the arithmetic mean of at least three independent parallel specimens. To evaluate experimental uncertainty, standard deviations were calculated for all datasets. The coefficient of variation (CV) across all tested groups was strictly maintained below 5%, demonstrating high experimental reproducibility. Because of this uniformly low variance and the necessity to preserve visual clarity in the multi-series time-dependent graphs (which contain numerous overlapping trajectories), error bars are not explicitly plotted. However, this statistically negligible dispersion confirms that the comparative differences observed between the different formulations are scientifically reliable.

3.2 Feasibility analysis of low-strength geopolymer formation

To investigate the heat release characteristics of the polymerization reaction, a thermal analyzer was used to monitor the exothermic behavior of the reaction. The experiment involved mixing inferior FGD gypsum and fly ash at a 50:50 mass ratio and adding NaOH or sodium silicate as activators. The direct experimental observations indicate that the exothermic reaction in NaOH solution is significantly more pronounced than in sodium silicate or pure water, with heat release levels being 3.5 times and 7 times higher, respectively (Figure 5). This indicates that NaOH activation promotes a more intensive initial reaction, with the heat release rate gradually stabilizing over time (0–2 h). Additionally, the microscopic morphology (Figure 6) presents an integrated matrix where original particles are embedded within newly formed phases.

Regarding the mechanistic interpretation of these direct observations, it can be inferred from established reaction mechanisms in similar alkaline-activated systems (e.g., Wang et al., 2023; Zhou et al., 2021), it is hypothesized that under NaOH activation, the glassy silicon-aluminum phase in fly ash dissolves, releasing active components that form amorphous, non-crystalline calcium silicate hydrate (C-S-H) and calcium aluminate hydrate (C-A-H) gels. Morphologically, the dihydrate gypsum crystals in FGD gypsum appear to react with some of the C-A-H gel to form the observed needle-like ettringite. In an alkaline environment, further breakdown of C-S-H and C-A-H gels results in Si-O-Si and Al-O-Al bond cleavage, leading to polymerization reactions between silicon-oxygen and aluminum-oxygen tetrahedra. This continuous theoretical geopolymerization process aligns with the sustained exothermic behavior observed in the calorimetry results (Figure 5). While these microstructural and calorimetric observations provide strong supporting evidence, they suggest rather than definitively prove the exact phase assemblage. As advanced spectroscopic characterizations (e.g., FT-IR, XPS, or SEM-EDS mapping) were not conducted in this study, the proposed stabilization mechanisms represent plausible interpretations rather than definitively confirmed processes. A hypothesized mechanism is the formation of polymeric structures with ettringite interlocking and fine aggregates of FGD gypsum and fly ash embedded within, forming a cohesive and dense geopolymer.

3.3 Strength characteristics of the geopolymer mixture

The compressive modulus is a key indicator of the deformation resistance of fill materials under constrained conditions. Higher compressive modulus values indicate reduced settlement and improved structural stability. The experimental results (Figure 7a) show that, without an activator, the compressive modulus increases over time and stabilizes after 90 days. The final compressive modulus for mixtures with 70:30, 50:50, and 30:70 ratios reached 14.26 MPa, 18.11 MPa, and 16.32 MPa, respectively, with the 50:50 mixture achieving the highest value. Although the numerical differences between the 50:50 ratio and the others are relatively modest, this formulation optimally balances maximum solid waste utilization with mechanical reliability. While compressive modulus alone is insufficient to dictate the final mix design, these observed values explicitly align with the intended load-bearing service function of the fill.

With NaOH activation (Figure 7b), the 70:30 mixture stabilized after 60 days, with a 14% increase in compressive modulus compared to the unactivated sample. Increasing the NaOH dosage from 1% to 3% led to a further 9% increase in compressive modulus due to enhanced geopolymerization and improved aggregate bonding. However, activation with sodium silicate alone resulted in a 5%–15% reduction in compressive modulus, indicating that sodium silicate alone is insufficient to form a geopolymer structure.

Direct shear tests were conducted on the mixed material specimens with different amounts of activators and different ages to determine the change characteristics of the shear strength index of the materials with time. The shear strength index of the mixed material specimens with ratios, DG: FA = 50:50 (Figure 8) indicates obvious stage characteristics in the shear strength of the mixed material changes with time. The change of the strength parameter is divided into two stages, and the strength is increasing first and then tends to be stable. 1n the first stage, within 60 d, the strength parameter of the composite material adding activators changes significantly with time. It indicated that the geopolymerization reaction has different degrees of influence on the strength of the material. The second stage is after 60 d; the strength change tends to stabilize. It indicated that the geopolymerization reaction reaches a stable state.

The value of cohesion of the mixed materials is fluctuating. The NaOH activator improves the cohesion and internal friction angle of the materials, but with time, the cohesion is weakened. The friction angle of the materials shows a gradual increase within 60 d, and then tends to stabilize. The results show that, in the condition of no activator, the geopolymer structure with strength has not been formed between the mixed materials. The cohesion and the internal friction angle have certain fluctuations due to the impact of the compacting work. The long-term strength of the material without an activator is smaller than that of the geopolymer structure formed with the activator.

From a mechanistic perspective, the above test results suggest that under the action of NaOH activator, the mixed material particles formed a geopolymer structure, and the fly ash geopolymer particles wrapped around the inferior FGD gypsum to form a joint structure, which improved the inter-particle friction. However, there was no obvious enhancement to the macroscopic cohesion, which was only manifested as a certain increase in the internal friction angle. The initially observed fluctuation and gradual weakening of cohesion over time likely reflect the complex physical competition between ongoing geopolymerization and the localized micro-softening of unreacted gypsum. Nevertheless, the stabilized long-term experimental strength parameters firmly demonstrate that the material remains structurally acceptable under continuous wet conditions.

The strength development and stabilization observed in this study align with the fundamental alkaline activation mechanisms reported in recent literature (e.g., Janga et al., 2024a; Wang et al., 2023), yet they highlight the unique kinetic behavior of inferior precursors. Unlike geopolymers synthesized from high-purity FGD gypsum, which typically exhibit rapid early strength acquisition, our 50:50 mixture demonstrates a more gradual strength gain that stabilizes after 60 d. Mechanistically, this delayed but sustained enhancement is governed by the slow dissolution of the aluminosilicate glassy phases in fly ash. In the highly alkaline environment provided by NaOH, the continuous cleavage of Si-O and Al-O bonds supplies monomers that slowly polycondense. Concurrently, the deleterious and unstable calcium sulfite hemihydrate inherent in the inferior FGD gypsum is gradually encapsulated by the newly formed C-A-H and C-S-H gels. Furthermore, the precipitation of ettringite acts as an interlocking micro-reinforcement within the geopolymeric network. Therefore, while the ultimate compressive modulus is relatively lower than that of conventional high-strength structural geopolymers, this chemically stable and interlocked microstructure effectively arrests micro-crack propagation, satisfying the specific mechanical criteria for bulk foundation backfilling.

3.4 Water stability test of inferior FGD gypsum and fly ash geopolymer mixtures

A direct shear test was conducted to evaluate the cohesion of geopolymer mixtures under water immersion. The test results for the 50:50 mixture are shown in Figure 9. Under water immersion, the cohesion of activated mixtures decreased over time (0–60 d). The rate of decrease was more pronounced with lower NaOH content (3%–1%).

At 60 d, the cohesion of the 50:50 mixture with 3% NaOH was approximately twice that of the 1% NaOH mixture, after which values stabilized. Unactivated mixtures exhibited consistently lower cohesion than activated ones, demonstrating that alkaline activators significantly enhance cohesion. For the 50:50 mixture, the final cohesion values for 1% and 3% NaOH activation were 24.6 kPa and 34.6 kPa, respectively. Compared to dry conditions, these values decreased by 42% and 33%, respectively.

Experimentally, the macroscopic reduction in cohesion under water immersion indicates a physical degradation process. We hypothesize that water molecules penetrate the porous structure, causing a localized softening effect primarily on the unreacted FGD gypsum microaggregates that are not fully encapsulated by the fly ash matrix. This presumed internal lubrication leads to the observed macroscopic loss in cohesion. The friction angle of activated mixtures gradually increased over the first 60 days and stabilized at 90 days. The final friction angle for NaOH-activated mixtures was 8.9%–9.2% higher than that of unactivated mixtures. Unactivated mixtures were significantly more water-sensitive, leading to shear strength reduction, whereas activated mixtures maintained their structural integrity. These findings suggest that NaOH activation promotes geopolymerization, forming a stable structure resistant to water-induced degradation.

The compressive modulus variation of geopolymer mixtures under dry and water-immersed conditions at different curing ages is shown in Figure 10. Under the condition of non-immersion, when the NaOH activator is added to the mixed material, the compressive modulus obviously tends to increase. The compression modulus of 180d increases from 14.9 MPa to 16.27 MPa with the activator addition ratio from 1% to 3%, which represents an increase in compression modulus of about 9.2%.

Under the condition of water immersion, the compression modulus of the mixed material specimens at each age condition had different degrees of reduction, in which the modulus of the mixed material specimens with 3% NaOH activator addition decreased the most, close to 10%. The test results show that the compression modulus of mixed materials with time change is affected by the dosage of activator, and the stabilization time and change trend of compression modulus show different patterns. Without an activator, the compression modulus of the mixture was affected by the impact of the time effect of compaction, manifested as the growth of time first decreased and then increased, and its compression modulus tends to stabilize after 90 d.

When NaOH activator was added, the compression modulus increased with the increase of activator addition ratio. Under the condition of water immersion, when the ratio of mixed materials is 70:30, the compression modulus tends to be stabilized after 28d when the NaOH amount is 3%, and the final stabilized compression modulus is 14.65 MPa. The compression modulus tends to be stabilized after 60d when the NaOH amount is 1%, and the final stabilized compression modulus is 11.54 MPa. It indicates that under the condition of water immersion, when the amount of NaOH is 3%, the compression modulus of the material is higher than 1%. Based on this experimentally observed macroscopic stabilization of the compression modulus, it is hypothesized that the NaOH activation promotes the formation of calcium aluminate and calcium silicate hydrate gels. Although not directly monitored via continuous spectroscopy in this study, these geopolymeric products theoretically possess low solubility. It is presumed that they physically encapsulate the unreacted gypsum dihydrate particles, thereby mitigating water intrusion and explaining the observed enhancement in the water resistance of the mixture.

3.5 Expansion characteristics

FGD gypsum contains water-absorbing and swelling components. To evaluate volumetric stability, an expansion test was conducted using water immersion. Specimens with and without NaOH activation were tested for volumetric strain over time (Figure 11). Unactivated specimens exhibited a rapid linear increase in volumetric strain from 0.5% to 1% within the first 10 min, stabilizing at 1.1% thereafter. NaOH-activated specimens showed a slower, more gradual increase, stabilizing at 0.6% after 6–24 min. The NaOH-activated specimens demonstrated significantly improved volumetric stability compared to unactivated specimens.

Experimentally, NaOH-activated specimens demonstrated a significant reduction in macroscopic volumetric strain compared to unactivated specimens. To explain this observed volumetric stability, we hypothesize that the alkaline activation enhances the precipitation of ettringite and geopolymeric polycondensates. These presumed reaction products would theoretically form a protective encapsulation layer around the swelling dihydrate gypsum phases, mechanically constraining their expansion and reducing their susceptibility to water ingress. While this proposed micro-mechanical interlocking mechanism strongly aligns with the macroscopic evidence, it remains a hypothesis that warrants future direct spectroscopic validation. Furthermore, the stabilized volumetric strain of 0.6% falls well below the project-specific engineering acceptance threshold for this foundation backfill (which required a volumetric expansion rate of <1.5%), making it suitable for the investigated application conditions.

4 Field testing

4.1 Compaction characteristics of material

Following laboratory compaction tests, different mixture ratios were used in field compaction and rolling trials. The fill layer thicknesses were set at 30 cm and 50 cm, and compaction was performed using a 25-ton roller with 4, 6, and 8 passes. Compaction degree was measured after 7 days. The relationship between compaction degree and rolling passes for different mixtures is shown in Figure 12.

The results indicate that compaction degree increased with the number of rolling passes. Under the same compaction conditions, the 50:50 mixture exhibited higher compaction degrees than the 30:70 and 70:30 mixtures. Additionally, for identical conditions, a 30 cm fill thickness resulted in a higher compaction degree than a 50 cm fill thickness. The maximum compaction degree of 0.94 was achieved for the 50:50 mixture with an 8-pass rolling process and a 30 cm fill thickness.

These findings suggest that mixtures with a higher gypsum content are less compactable. Synthesizing the laboratory and field test results, the 50:50 DG:FA ratio activated with 3% NaOH was conclusively identified as the optimal formulation. This specific mixture was selected because it successfully maximized the solid waste utilization rate while simultaneously fulfilling rigorous multi-criteria engineering thresholds: achieving the highest compressive modulus (Section 3.3), maintaining superior cohesion under prolonged water immersion (Section 3.4), exhibiting negligible volumetric expansion below engineering limits (Section 3.5), and providing the optimal field compactability with a maximum compaction degree of 0.94.

4.2 Compressive modulus in field testing

Field samples were tested at 60, 90, and 180 days after compaction. The compressive modulus trends for the 50:50 mixture are shown in Figure 13. Results indicate that compressive modulus steadily increased with curing time. Thinner layers and higher rolling passes resulted in higher modulus values. At 180 days, the compressive modulus reached 10.74 MPa for a 30 cm fill thickness (8 passes) and 9.25 MPa for a 50 cm fill thickness (8 passes). These findings confirm that geopolymerization continues gradually over time, enhancing the cohesion of the structure and improving mechanical properties.

4.3 Dynamic penetration testing

To evaluate the effects of fill thickness and compaction passes on penetration resistance, light dynamic penetration tests were conducted at depths of 30 cm on 1, 3, and 7 days. The test results are shown in Figure 14.

Results demonstrate that the number of penetration blows increased with both curing time (0–7 days) and compaction passes (4–8 passes). For both 30 cm and 50 cm fill thicknesses, the 50:50 mixture achieved penetration resistance exceeding 70 blows at 7 days, indicating uniform compaction and high bearing capacity.

4.4 Load-bearing characteristics

A plate load test was conducted to evaluate the true engineering performance of the backfill material. The test area used the 50:50 mixture with an 8-pass rolling process and a 30 cm fill thickness. A 0.25 m² loading plate was used, applying loads in 40 kPa increments up to 520 kPa.

The results indicate a smooth settlement curve. No distinct inflection point was observed below 400 kPa, while rapid settlement occurred between 400 and 520 kPa. The total maximum settlement was 20.2 mm. The ultimate bearing capacity was determined to be 400 kPa, and the characteristic bearing capacity was 200 kPa, corresponding to a 4.0 mm settlement.

The results confirm that the material achieves a bearing capacity of 150–200 kPa, with a deformation modulus of 52.7–56.5 MPa, meeting the specific design criteria of the investigated factory foundation project (which mandated a minimum characteristic bearing capacity of 150 kPa).

5 Engineering application

5.1 Design plan for foundation backfilling

The selected industrial plant expansion project represents a highly suitable validation case due to its large-scale backfill volume requirements and the strict differential settlement limits associated with heavy industrial facilities. The site is located in a temperate monsoon climate zone, subjecting the shallow foundation to natural dry-wet seasonal variations. Geologically, the superficial soil layer consists of 1.5–2.0 m of silty clay, underlain by a stable sandy silt stratum. Groundwater was encountered at a depth of approximately 3.5 m below the existing ground surface, which sits below the maximum backfill depth, thereby avoiding continuous direct water immersion while still allowing for capillary moisture interaction.

Considering site topography, design elevations, and fill area requirements, the fill material was applied in three key zones: Composite Foundation Fill Zone: This zone consisted of a cement-soil mixing pile composite foundation. The fill material acted as the bearing layer, and its load-bearing and deformation properties were required to meet the project-specific engineering standards (i.e., a characteristic bearing capacity ≥150 kPa and strict differential settlement control). Natural Foundation Fill Zone: This zone had an original ground base where the fill material served as the bearing layer for structures, requiring it to meet these identical load-bearing and deformation performance criteria Building Floor Fill Zone: The fill material was used to fill gaps between foundations, providing a stable bearing layer for building floors. The material needed to meet general load and deformation requirements for normal use. For example, in the building floor fill zone, the design elevation was set at 31.6 m. This backfill area extended from the building foundation base to the ground floor elevation, primarily supporting vehicular and heavy industrial storage loads (typically ranging from 20 to 50 kPa). The backfill thickness was determined based on the foundation-to-floor elevation. The design specified a 300 mm layered fill thickness with a minimum compaction coefficient of 0.93.

5.2 Fill material preprocessing and mixing

To ensure material consistency, inferior FGD gypsum and fly ash underwent pre-processing before use. The procedure involved: Extracting raw materials from stockpiles in layers, removing surface debris and vegetation roots, and storing cleared materials separately. Sorting raw materials into designated storage areas and conducting moisture content testing. Mixing FGD gypsum and fly ash at a 50:50 mass ratio, strictly controlling the mix proportions. Adding 3% NaOH solution and blending the materials using a soil mixing machine for 5 min to ensure uniform hydration. Monitoring the moisture content of each batch to ensure compliance before use.

5.3 Fill material placement

Before backfilling, the ground surface was cleared of loose soil, roots, and debris. In the 1# Roasting Workshop, a building floor fill zone, the material was primarily used to fill foundation gaps and support the building floor load.

The fill placement process was as follows: Transportation and Spreading: The mixed material was transported to the site by dump trucks and spread using a paver in 30 cm layers. Compaction: A 20-ton vibratory roller compacted the fill with 8 passes, using a vibration frequency of 25–35 Hz and a travel speed of 3–5 km/h. Edge Finishing: Edges were manually compacted, and in narrow areas, hand-pushed carts were used for placement. The real-life scene of on-site operations is shown in Figure 15.

5.4 Evaluation of foundation treatment effectiveness

After fill placement, 900 tests were conducted to measure moisture content and dry density. Results showed a density of 1.26–1.28 g/cm³ and a moisture content of 22%–25%, meeting design requirements. The compaction acceptability criteria were strictly based on achieving a minimum compaction coefficient of 0.93 across all structural zones. Post-compaction tests revealed: Compaction degree: 0.94–0.96; Bearing capacity: 150–200 kPa; Deformation modulus: 52.7–56.5 MPa; These values confirmed compliance with the specific engineering acceptance criteria established for this project (minimum compaction degree of 0.93 and target bearing capacity ≥150 kPa). The settlement monitoring curve for the 1# Roasting Workshop is shown in Figure 16. To systematically evaluate the structural performance and ensure transferability of the results, the monitoring layout was established by affixing settlement observation points directly to the primary load-bearing columns and perimeter ring beams at intervals of 10–15 m across the foundation column grid. The engineering acceptance criteria for this industrial facility mandated a maximum allowable total settlement of 50 mm and an average settlement rate of less than 0.01 mm/day during the final stabilization phase. Over 18 months of monitoring, multiple structures using this fill material exhibited: 1. Maximum settlement: 35.44–44.07 mm; 2. Average settlement: 32.33–36.64 mm; 3. Average settlement rate (over two monitoring periods): <0.01 mm/day, indicating structural stability.

All observed deformations complied with engineering standards. The roasting workshop, carbon block warehouse, and other structures at the carbon plant successfully utilized this foundation replacement technique, repurposing over 200,000 cubic meters of inferior FGD gypsum. This significantly reduced the use of natural sand and gravel, freed up land resources previously used for waste stockpiling, and demonstrated the practical engineering potential of upcycling solid waste on a macro scale. Based on building deformation monitoring and post-construction performance analysis, the strength and stiffness of the treated foundation met load-bearing and deformation requirements, ensuring smooth project operation. It should be noted that while the backfill material performed satisfactorily and met all design requirements for this investigated project, further site-specific evaluations under varying hydrogeological conditions are necessary before assuming universal suitability for all foundation fills.

While this study successfully demonstrates the macroscopic engineering feasibility of utilizing inferior FGD gypsum and fly ash as foundation backfill, the lack of a quantitative environmental evaluation remains a limitation of the current work. Based on extensive previous studies on solid waste-derived cementitious materials, the dense three-dimensional aluminosilicate network of geopolymers exhibits excellent immobilization and solid-sealing capabilities against heavy metal ions (e.g., Cu²⁺ and Cd²⁺) through strong physical encapsulation and ion exchange mechanisms within the gel matrix (Balsamo et al., 2012; Bai et al., 2024). Because the highly alkaline activation environment fundamentally promotes these stabilization mechanisms, large-scale leaching toxicity is widely considered to be effectively mitigated. Nevertheless, as the precise trace element content fluctuates depending on the origin of the inferior FGD gypsum and fly ash, specific leaching tests (e.g., Toxicity Characteristic Leaching Procedure, TCLP) were not conducted within the current scope to quantitatively verify this containment under complex hydrogeological conditions. Consequently, it must be explicitly acknowledged that the environmental safety of this specific material batch has not yet been experimentally verified.

Furthermore, compared to conventional foundation backfill materials such as cement-stabilized soils or Controlled Low-Strength Materials (CLSM), the proposed 50:50 DG:FA geopolymer presents distinct competitive advantages. Mechanically, its compressive modulus and shear strength comprehensively meet standard foundation backfill requirements, offering comparable or superior load-bearing capacity to typical CLSM, which generally targets strengths of 1–3 MPa to maintain future excavatability (Katz and Kovler, 2004). In terms of durability, conventional Portland cement-based materials are highly susceptible to destructive internal sulfate attack (resulting in delayed ettringite expansion) when exposed to high-sulfate industrial by-products like FGD gypsum (Rajasekaran, 2005). In contrast, the proposed geopolymer system beneficially consumes these sulfates to precipitate stable ettringite during the initial reaction phase, successfully transforming a durability threat into a structural advantage and ensuring exceptional volumetric stability. Environmentally and economically, this 100% solid-waste precursor formulation completely eliminates the reliance on Portland cement, thereby drastically reducing carbon dioxide emissions and material costs (Provis and Bernal, 2014) while achieving field compactability comparable to conventional aggregate backfills.

6 Conclusion

This study explores the engineering viability of utilizing long-term stockpiled, inferior FGD gypsum and fly ash as a primary geopolymer foundation backfill, demonstrating promising performance within the specific constraints of the investigated project. Under alkaline activation (3% NaOH), the 50:50 mixture forms a low-strength geopolymer where the inherently unstable calcium sulfite hemihydrate is effectively stabilized. Based on the microstructural and calorimetric evidence, this study postulates that these deleterious phases are likely encapsulated within the polymeric Si-O-Al backbone and interlocked by newly formed ettringite, contributing to the observed water and volume stability, although the exact chemical bonding and reaction products warrant further confirmation via advanced spectroscopic techniques. For a 50:50 mixture with a 30 cm fill thickness and 8 passes of compaction, the results showed a compaction degree of 0.94, a compressive modulus of 10.74 MPa, 7 d light dynamic penetration resistance of 70 blows, plate load test bearing capacity of 150–200 kPa, and the deformation modulus of 52.7–56.5 MPa. The use of inferior FGD gypsum and fly ash as fill materials in the specific industrial plant project resulted in a final building settlement of 35.44–44.07 mm and a settlement rate of <0.01 mm/day, which strictly meets the engineering design requirements of the investigated site.

While these findings robustly validate the geopolymer’s macroscopic mechanical reliability, it is explicitly acknowledged that the environmental safety of the material has not yet been experimentally verified. The absence of quantitative leaching evaluations under complex hydrogeological conditions (despite the strong theoretical basis for heavy metal encapsulation within the geopolymer matrix) remains a structural limitation of the current study.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Ramadhansyah Putra Jaya, Universiti Malaysia Pahang, Malaysia

Reviewed by: Hanane El Harouachi, Multidisciplinary Faculty of Nador, Morocco

Mohammad Reza Hanafi, Amirkabir University of Technology, Iran