With the continuous development of infrastructure in China, the demand for concrete has increased significantly. As a critical raw material in concrete production, sand has also seen a sustained rise in consumption. It is estimated that global annual sand usage ranges between 40 and 50 billion tons. Due to the non-renewable and non-recyclable nature of river sand, its excessive extraction has led to the rapid depletion of existing reserves. To enhance resource efficiency, the construction industry has gradually implemented a variety of mitigation strategies, such as incorporating recycled coarse aggregates into concrete and employing manufactured sand as a substitute for river sand. The use of manufactured sand not only alleviates the environmental degradation caused by overexploitation of river sand but also offers a practical solution to the growing scarcity of this natural resource. Manufactured sand is produced by mechanically crushing and screening rock particles. Compared with river sand, it is characterized by a rougher surface texture and a higher content of stone powder, which can significantly influence the long-term mechanical performance of concrete. Creep, defined as the time-dependent deformation of concrete under sustained loading, plays a crucial role in determining the structural stability and durability. However, existing creep models for concrete have proven inadequate in accurately predicting the creep behavior of manufactured sand concrete. In recent years, machine learning techniques have been increasingly applied to the prediction of creep behavior in recycled aggregate concrete, ultra-high-performance concrete, and concrete incorporating mineral admixtures, demonstrating excellent fitting accuracy and generalization capabilities. Compared with traditional predictive models, machine learning approaches are better suited to handling complex issues such as multi-factor coupling and nonlinear relationships. However, their limitations are also notable: these models often function as "black boxes," lacking physical interpretability, and are highly dependent on the quality and quantity of input data, making it difficult to uncover the underlying physical mechanisms of creep. To address these challenges, some researchers have introduced explainable algorithms to improve model transparency and robustness Simultaneously, other studies have explored explicit modeling techniques to directly generate predictive formulas. Nonetheless, the resulting expressions are often overly complex and lengthy, limiting their applicability in practical engineering scenarios.

To tackle these challenges, a novel modeling framework that integrates both implicit and explicit techniques is proposed. First, a dataset of manufactured sand concrete creep was constructed. Advanced implicit machine learning models, including the Feature Token Transformer, Physics-Informed Neural Networks (PINNs), TabNet, and XGBoost, were employed to predict the creep behavior of manufactured sand concrete. Feature importance analysis was subsequently conducted based on model outputs to identify the most influential variables. Next, AutoFeat was used to automatically generate and select polynomial features from the identified key variables, thereby constructing a high-order explicit model capable of achieving accurate creep prediction. Furthermore, neural symbolic regression (NSR) was introduced to automatically discover nonlinear relationships among variables through symbolic function composition, yielding concise, physically meaningful analytical expressions.

Existing concrete creep prediction models, when applied to manufactured sand concrete, exhibit substantial errors due to their inability to effectively capture the material-specific characteristics and time-dependent behavior. This highlights an urgent need for the development of more targeted and robust predictive models.

The constructed implicit machine learning models—TabNet, FT-Transformer, PINNs, and XGBoost—demonstrated excellent predictive performance. These models effectively capture the complex and nonlinear relationships between the creep behavior of manufactured sand concrete and various influencing factors.

Explicit prediction models for concrete creep compliance were developed based on two approaches: AutoFeat and Neural Symbolic Regression (NSR). Both methods successfully generated functional expressions that map input parameters to the target output, thereby enhancing model interpretability. The AutoFeat-based model combined SHAP-based feature importance analysis with Pearson correlation analysis during feature selection, ultimately yielding an interpretable model primarily composed of first-order linear terms, complemented by selected second- and third-order interaction terms. In contrast, the NSR method exhibited a superior capacity for capturing nonlinear dependencies. It automatically identified key variables and constructed complex mathematical formulations characterized by nested structures and nonlinear operator combinations, providing deeper insight into the underlying mechanisms governing creep development.

Under comparable levels of predictive accuracy, both AutoFeat and Neural Symbolic Regression are capable of generating explicit expressions with clear mathematical structures, significantly enhancing model interpretability. Despite differences in feature selection strategies and model architectures between implicit and explicit modeling approaches, both demonstrate strong fitting performance with respect to real-world data, thereby validating the feasibility of modeling concrete creep through multiple methodological pathways. The integrated application of both approaches facilitates an optimized balance between predictive accuracy and interpretability.

The distribution characteristics of microcapsules incorporated into matrix can influence the quantity of microcapsules required for embedding in the cementitious matrix, thereby determining the self-healing efficiency. Self-healing systems are essentially passive systems that autonomously initiate the healing process when cracks occur. The efficacy of such systems necessitates both adequate distribution and sufficient bonding of the capsules with the matrix. The recent preparation of microcapsules in self-healing composites presents significant challenges in achieving optimal self-healing effects. One of the key obstacles is the aggregation and uneven distribution of capsules within the matrix, which can significantly impair the healing efficiency of microcapsule self-healing cementitious materials. This nonuniform distribution diminishes the effectiveness of the capsules and adversely affects properties like early mechanical strength. Furthermore, the uneven distribution of capsules in the matrix will lead to the reduction of the number of capsules encountered by cracks. It is therefore essential to ensure the uniform distribution of capsules within the matrix in order to maximise the probability of microcapsule rupture and self-healing capacity.

In order to avoid agglomeration of microcapsules, a uniformed distribution model for microcapsules inspired by the structural form of a honeybee's hive is designed in self-healing cementitious materials and some embedded microcapsules are dispersed within the honeycomb. This honeycomb model ensures the uniform distribution of microcapsules inside the matrix and prevent the aggregation of microcapsules, thereby achieving a design strategy for the uniform distribution of microcapsules in self-healing cementitious materials.

The probabilistic characteristics of a planar crack that appears inside a composite matrix and intersects with embedded microcapsules is investigated from the viewpoint of geometrical probability. The probability of a randomly occurring planar crack in the matrix when it encounters a spherical or spherocylindrical microcapsule is derived, and an intersection probability formula is obtained. Furthermore, the influences of capsule dimensions and quantities on the intersection probability is examined in capsule-based self-healing composites. In addition, comparative analysis is conducted on microencapsulated composite matrix models with varying shapes. Finally, the results of the theoretical model are validated using Monte Carlo simulation.

The honeycomb-like composite structure is a three-dimensional structure composed of several small cubes (cells), with a certain number of spherical or spherocylindrical microcapsules randomly dispersed within the small cubes, ensuring the uniform distribution of the capsules and avoiding the aggregation of microcapsules. When a random planar crack appears within the composite, the crack will meet some microcapsules and causes the ones to rupture releasing a healing agent to heal the crack, which is necessary for self-healing to work. Three-dimensional probabilistic analytical models are developed for the spherical microcapsules and spherocylindrical microcapsules in order to represent the intersection of a planar crack with randomly dispersed microcapsules, respectively.

The findings indicate that uniformly dispersed microcapsules can increase the probability of capsule rupture and improve the self-healing efficiency. The developed three-dimensional geometric probability model of a planar crack intersecting with randomly distributed microcapsules in the matrix cells is close to a binomial probability distribution. The use of spherocylindrical microcapsules can increase the healing potential of the same volume fraction of spherical microcapsules if the microcapsules are correctly oriented. The theoretical probability values are verified by comparison with the results obtained independently by computer programming simulations.

The proposed theoretical formulas quantitatively express the effect of the size and dosage of different-shaped microcapsules within the matrix on the probability of a planar crack hitting microcapsules and to predict the number of microcapsules that are most likely to intersect. If the volume fraction (or size) of microcapsules is given and a planar crack requires a fixed number of microcapsules to heal, the maximum intersection probability can be calculated from the theoretical equation to determine the optimum size (or volume fraction) of microcapsules to be mixed within the model. The results of this research will help to provide a theoretical basis for the quantitative design of self-healing materials based on microcapsules of different shapes. Furthermore, employing 3D printing technology could be a feasible approach to realize the design strategy for the uniform distribution of microcapsules within self-healing cementitious materials.

The free lime (f-CaO) content in cement clinker is a key indicator for evaluating calcination quality and process stability, directly influencing cement soundness and mechanical performance. Efficient and accurate real-time prediction of f-CaO content is essential for achieving stable cement quality control. In actual production, process parameter data are sampled at approximately 1-second intervals, while f-CaO content is typically measured once per hour, resulting in a severe scarcity of labeled data. This asynchronous sampling and data heterogeneity constrain both the scale and diversity of data available for supervised learning models, thereby reducing prediction accuracy and generalization capability. To address this challenge, this study proposes a method based on a semi-supervised generative adversarial network (SSGAN) to enhance the diversity and representativeness of the training dataset. Through evaluating its impact on the downstream f-CaO prediction task and conducting ablation studies, the research explores the feasibility of applying GANs to address the issue of asynchronous sampling between process parameters and offline measurement data.

This study proposes a data augmentation model based on a semi-supervised generative adversarial network (SSGAN), which generates virtual data resembling the actual data distribution to expand the training dataset and effectively enhance the model's learning capability under conditions of limited labeled data. The model consists of three main components: an encoder, a generator network, and a discriminator network. The encoder first extracts features from a small amount of process parameter data with real labels, which are then used as conditional inputs. The generator learns the latent features of the true label distribution through adversarial training and generates corresponding label values based on unlabeled process parameter data. The discriminator distinguishes between real and generated labels, guiding the generator to continuously optimize its generation strategy. As adversarial training progresses, the generator produces synthetic label data that closely resemble actual production data, while the discriminator continuously improves its understanding of the true data distribution and label characteristics.

When the SSGAN model is trained to approximately 10 000 iterations, both the discriminator loss and the generator loss converge to near zero and remain stable, indicating that a balance has been achieved between the generator and discriminator, and the model has effectively learned the latent distribution features within the process parameters. Further analysis of the distribution of generated label values shows that they exhibit a distribution trend generally similar to that of the real labels. The model demonstrates strong learning and fitting capability within the primary range of label values, and the generated data are reliable. After introducing SSGAN for data augmentation, the MAE, MSE, and RMSE of the downstream GRU (Gate Recurrent Unit) model for f-CaO content prediction decreased from 0.24, 0.08, and 0.29 to 0.19, 0.06, and 0.24, respectively, indicating an improvement in the model's ability to predict outliers and extreme samples. Ablation studies revealed that removing the encoder, the GRU module, or introducing random noise led to significantly higher prediction errors, highlighting the critical role of temporal information and deep features embedded in the process parameters for generating high-quality label data. In particular, the encoder and its internal GRU module are essential for capturing long-range temporal dependencies and dynamic variation patterns.

To address the issue of label scarcity caused by the mismatch in sampling intervals between process parameters and f-CaO content measurements, this study proposes a data augmentation method based on SSGAN. By using an encoder to extract high-dimensional features and a generator to learn the latent distribution of true labels, the approach constructs a more diverse and representative training dataset. Ablation experiments were conducted to evaluate the effectiveness of each key module. The main conclusions are as follows: The introduction of the SSGAN-based data augmentation strategy effectively expanded the training dataset. The distribution of the generated label values closely matched that of the true labels, verifying the reliability of the synthetic data. The application of the SSGAN method significantly improved the accuracy of the downstream f-CaO content prediction task, with MAE on the test set decreasing from 0.24 to 0.19, MSE from 0.08 to 0.06, and RMSE from 0.29 to 0.24. Ablation experiments further revealed that introducing random noise, removing the encoder, or omitting the GRU module led to significantly increased prediction errors. This highlights the critical role of temporal information and deep features embedded in process parameters for generating high-quality label data. In particular, the encoder and its internal GRU module are essential for capturing long-range temporal dependencies and dynamic variation patterns.

Alkali-activated slag (AAS) is a cementitious material prepared by using blast furnace slag as the precursor, triggered by alkaline activators to undergo dissolution, polycondensation, and solidification reactions. Compared to traditional cement-based materials, inherent defects of AAS, such as rapid setting and hardening, significant volumetric shrinkage, and susceptibility to cracking, severely limit its engineering applications. The hydration of magnesium oxide (MgO) generates magnesium hydroxide, with a solid-phase volume expansion of approximately 118%, which effectively compensates for the volumetric shrinkage caused by AAS hydration. However, existing studies are predominantly confined to single-dimensional analyses, such as the impact of MgO reactivity and content, fixed activator parameters, or specific curing ages on isolated performance, lacking systematic investigation into the coupling effects of multiple factors, microstructural evolution, and macro-performance development. This paper systematically investigates the regulatory mechanisms of MgO as a modifier on the hydration kinetics, hardening behavior, and microstructure evolution of AAS.

The precursor used in this experiment was slag with a specific surface area of 426 m²/kg. The alkaline activator was liquid sodium silicate which adjusted to a modulus of 1.5 using sodium hydroxide and deionized water. The sand employed was ISO standard sand. The additives included high-reactivity HR-MgO (reactivity time: ~20 s) and medium-reactivity MR-MgO (reactivity time: 70~90 s). The experimental parameters were set as follows: alkali content (Na₂O as a mass percentage of slag) = 6%, modulus (Ms = SiO₂/Na₂O molar ratio) = 1.5, water-to-solid ratio (including water in the activator and mixing water) = 0.4, and MgO dosage (as a mass percentage of slag) = 0%, 4%, 8%, and 12%. The specimens were molded at room temperature, demolded after resting for 6 h, and then cured under sealed conditions at a temperature of (20 ± 2) ℃ and relative humidity ≥ 95% until the specified curing age. The flowability, setting time, hydration heat, compressive strength, TG, XRD, FTIR, and SEM tests were conducted.

During hydration, MgO consumes partial free water, significantly reducing the fluidity of AAS mortar, with the fluidity loss intensifying as MgO reactivity and dosage increase. Concurrently, MgO incorporation prolongs the setting time of AAS paste, particularly with HR-MgO exhibiting stronger retardation effects. This phenomenon primarily arises from the formation of a passivation layer—composed of Mg(OH)₂ and M-S-H gel generated via MgO hydration—on slag particle surfaces, which inhibits further slag dissolution and thus delays setting. Moderate MgO content enhances the early hydration exothermic rate of AAS, shortening the duration of the induction period, with MR-MgO exhibiting superior enhancement effects compared to HR-MgO. While MgO minimally impacts the peak exothermic rate during the acceleration stage, it significantly increases cumulative heat release over 72 h. For instance, the HR-12 group achieved 178.84 kJ/kg, representing a 13.8% improvement over the control group. This indicates that MgO synergistically promotes hydration by accelerating slag dissolution and ion exchange processes, though excessive dosage diminishes early activation efficiency. MgO incorporation improves the early compressive strength of AAS. The HR-8 group attained a maximum 8 h compressive strength of 51.76 MPa, representing a 38.82% increase compared to the control group. However, strength growth in MgO-doped samples lags behind the control group at later ages, with 28 d compressive strength decreasing as MgO reactivity and dosage increase. MgO incorporation reduces the geometric crystal growth index (n) during the nucleation and growth (NG) stage, shifting hydration from homogeneous to heterogeneous nucleation. This accelerates crystal nucleation/growth rates, enabling earlier transition to the phase boundary reaction (I) stage. During Stage I, the synergistic interaction between MgO's secondary reactions and slag particle phase boundary reaction dominates hydration kinetics, causing actual exothermic curves to deviate upward from fitted curves and resulting in reduced curve fitting accuracy. As hydration progresses, the hydration product layer gradually covers slag particles, inhibiting ion diffusion and thereby reducing reaction rates during the diffusion-controlled (D) stage. Additionally, MgO alters AAS hydration product composition: M-S-H gel (with lower hydration heat than C-A-S-H gel) further decreases the D-stage reaction rate constant, impacting long-term hydration.

MgO markedly reduces the fluidity of fresh paste through free water absorption. During the initial hydration phase, the low solubility of Mg (OH)₂ under high alkalinity enables the formation of a passivation layer on slag particles, effectively delaying slag dissolution and prolonging setting time. As hydration progresses, pH reduction enhances Mg(OH)₂ solubility, releasing Mg²⁺ and OH^- ions that exhibit synergistic effects: 1) Elevated alkalinity accelerates slag dissolution; 2) Mg²⁺ reacts with dissolved [SiO₄]^4- and [Al(OH)₄]^- to generate secondary hydration products (M-S-H gel and hydrotalcite, Ht), substantially improving early compressive strength (8 h strength increased by up to 38.82% with MgO content of 8%). However, excessive Mg²⁺ competitively consumes [SiO₄]^4- and [Al(OH)₄]^- with Ca²⁺, suppressing the crystallinity of C-(A)-S-H gels and reducing their aluminum substitution, ultimately leading to structural loosening and compromised long-term strength development (28 d strength reduction up to 15.99%). Krstulovic-Dabic hydration kinetic modeling demonstrates that MgO incorporation enhances the nucleation and crystal growth (NG) rate but hinders the diffusion-controlled (D) stage by forming a shielding product layer. Notably, the total hydration heat exhibits positive correlation with MgO reactivity and dosage, attributed to continuous formation of magnesium-bearing phases (Mg (OH)₂, Ht, M-S-H). This work provides theoretical insights for optimizing AAS formulations to balance setting regulation, early strength enhancement, and long-term strength sustainability.

Magnesium oxide (MgO) content in Portland cement clinker critically influences clinkering and long-term soundness. Limited MgO(≤2%, mass fraction) optimizes mineral formation by reducing liquid-phase viscosity during sintering, but excessive MgO forms low-reactivity periclase. Delayed hydration of periclase to brucite (Mg(OH)₂) causes destructive expansion. Thermodynamically, Mg²⁺ can react with Al³⁺ from C₃A/C₄AF to form magnesium-aluminum layered double hydroxide (Mg-Al LDH), which offers superior anion binding capacity. However, kinetic competition with formation of AFt(ettringite)/AFm(monosulfate) phases in sulfate-rich systems often suppresses LDH formation. This study investigates the formation pathway of Mg-Al LDH and the role of temperature in MgO-doped C₃A hydration (without sulfate), revealing novel Ca-Mg-Al hydrates(X-phase) and microstructural evolution mechanisms.

MgO was obtained by calcining basic magnesium carbonate (900 ℃, 2 h). C₃A-MgO composites (40 g MgO to 100 g C₃A) were synthesized from CaCO₃/Al₂O₃ mixtures (Ca/Al=3∶2), sintered at 1450 ℃ for 3 h, and air-quenched. XRD-Rietveld analysis quantified phases: 69.70% cubic C₃A, 25.09% periclase, 3.58% C₁₂A₇, and 1.63% free CaO. Hydration suspensions (w/s=10∶1, under N₂ atmosphere) were cured at 20 ℃ and 50 ℃ for 1 h-120 d. Hydration was stopped with isopropanol; solids were filtered, vacuum-dried (40 ℃), and characterized. Phase evolution was analyzed via XRD (Rietveld refinement) and in-situ high-temperature XRD (25-200 ℃). Morphology and elemental distribution were examined using FIB-SEM (Helios 5 UC) and EDS (Oxford UltimMax65). Platy particles were cross-sectioned by FIB etching (500 nm-2 μm depths) for compositional profiling.

At 20 ℃, C₃AH₆ forms rapidly, constituting 50% of solids within 1 h, and C₃A is fully consumed by 3 d. Mg-Al LDH emerges after 6 h, grows gradually to 45% and with approximately 7% periclase residue at 56 d. XRD analysis reveals a novel X-phase at 28 d, characterized by distinct diffraction peaks at 6.70° and 13.38° 2θ. In contrast, at 50 ℃, accelerated dissolution of periclase enables complete hydrolysis of MgO by 28 d. Mg-Al LDH dominates the system after 7 d, reaching 54% at 56 d and replacing C₃AH₆ as the primary phase. The X-phase forms within just 1 h and consistently increases with prolonged curing.

The morphology and composition of Mg-Al LDH exhibit significant temperature dependence. At 20 ℃, LDH crystallizes as well-defined hexagonal sheet with Mg/Al ratios ranging from 2.77 to 3.16 (EDS data). SEM images and EDS confirmed C₃AH₆ dissolution is driven by Mg²⁺, and Al³⁺ originated from C₃AH₆ supports Al³⁺ for LDH growth. At 50 ℃, however, LDH shows irregularly shaped flake with a lower, more homogeneous Mg/Al ratio (2.21 ± 0.11). This decrease of Mg/Al ratio suggests enhanced Al³⁺ availability from C₃AH₆ at elevated temperatures.

The formation pathway of the X-phase involves a three-component reaction: C₃AH₆+Mg₄Al₂(OH)₁₄·3H₂O+CH→X. X-phase early appearance at 50 ℃ (1 h vs. 28 d at 20℃) confirms temperature-dependent kinetics. In situ high-temperature XRD indicates its structural reversibility: heating above 50 ℃ induces peak shifts (e.g., new peaks at 8.82° and 10.64° 2θ), yet cooling restores the original pattern, confirming thermally reversible structure. EDS of X-phase surface identifies composition of CaMgAl(OH)₇. FIB-EDS cross-sectional profiling exposes a linear Ca/Mg ratio within X-phase inner lamellae (slope = -0.46), implying continuous Mg²⁺ substitution for Ca²⁺ during maturation. Under curing at 50 ℃ for 120 d, platy X-phase particles evolve into stacked lamellar up to 20 μm width and 2 μm thickness. These particles encapsulate residual C₃AH₆, demonstrating the role of C₃AH₆ as reactants.

Elevating curing temperature from 20 ℃ to 50 ℃ accelerates periclase dissolution, enabling Mg-Al LDH to dominate C₃A hydration after 7 d and reducing its Mg/Al ratio from 2.77 to 2.21. Mg²⁺ etches C₃AH₆ surfaces, facilitating LDH growth by the release of Al³⁺. A novel Ca-Mg-Al hydrate (X-phase) forms through reactions of C₃AH₆, Mg-Al LDH, and Ca(OH)₂, exhibiting thermally reversible phase transitions and a composition of this phase surface near CaMgAl(OH)₇. Internal Ca/Mg ratio within X-phase lamellar particles suggests ion exchange during maturation.

Magnesium silicate hydrate (M-S-H) attracts significant attention as an alternative to traditional portland cement due to its unique properties, including reduced energy consumption during preparation. However, its synthesis typically requires MgO from calcined magnesite, which undermines the environmental benefits it provides. This study addresses these limitations by proposing a waste valorization strategy: recycling waste acid to dissolve naturally abundant magnesium silicate minerals (serpentine, olivine, and sepiolite) for M-S-H production. By combining acid waste management with mineral resource utilization, this approach not only avoids the energy-intensive MgO production process but also eases the ecological hazards caused by acid disposal.

The experimental involved three key stages: acid leaching, material preparation, and characterization. First, serpentine, olivine, and sepiolite were subjected to dissolution in a mixed waste acid solution (HCl: 6.20 mol/L, H₂SO₄: 7.77 mol/L) at 90 ℃ for 3 h under continuous mechanical agitation. The solid-to-liquid ratio was maintained at 1∶5 for acid leaching. After leaching, the insoluble residues (mainly amorphous silica) were separated with a centrifuge (Thermo Scientific ST 16R). The obtained residues were washed ultrasonically with deionized water 5 times to remove residual acid. Subsequently, the residue was dried in a vacuum drying oven at 50 ℃ for 24 h. The acid-leached filtrate was added into a 10 mol/L NaOH solution, with real-time pH monitoring until reaching pH 11.5. The obtained suspension was left to settle at room temperature for 24 h. The white flocculent precipitate was collected by vacuum filtration and dried at 100 ℃ for 24 h to obtain Mg(OH)₂-rich powder. The dried residue was mixed with Mg(OH)₂ powder at a ratio of n(SiO₂) ∶ n(MgO)=1.0∶1.4, and blended with water with a water-to-solid ratio of 1 ∶ 2 to form a paste. The paste was sealed in glass tubes and cured in water baths at 20, 50 ℃, and 80 ℃ for 3, 7, 14 d, and 28 d, followed by vacuum drying at 50 ℃ for 24 h prior to characterization. Mortar samples with the same residue-Mg(OH)₂ ratio were tested for compressive strength, fluidity, and drying shrinkage according to Chinese national standards.

The residues contain 89%-96% amorphous SiO₂, and Mg(OH)₂ precipitates exhibit 96% purity. Promoting curing temperature significantly accelerates M-S-H formation. At the same age, the diffraction peak of Mg(OH)₂ is lower in samples cured at higher temperature. TG curves reveal 3 distinct mass loss: 50-350 ℃ (loss of physically bound water, Δw₁), 350-450 ℃ (Mg(OH)₂ dehydroxylation, Δw₂), and 450-750 ℃ (Si-OH dehydroxylation, Δw₃). The Δm₂ content decreases with age and elevated temperatures, indicating accelerated Mg(OH)₂ reaction. A linear relationship between Δm₂ and (Δw₂+Δw₃) is established. A decline is observed in the slopes of fitting line, from 0.96 (20 ℃) to 0.80 (50 ℃) and 0.76 (80 ℃), which indicates that the contribution from unreacted Mg(OH)₂ decreases while the formation of M-S-H is promoted with increasing temperature. The broadening and intensification of the exothermic peak near 850 ℃ (M-S-H recrystallization) in the DSC curve indicate enhanced maturation of M-S-H with increasing curing temperature. ²⁹Si NMR spectra indicate elevated curing temperatures reduced the relative content of Q¹, confirming enhanced polymerization of silicon-oxygen tetrahedra in M-S-H phases, though infinite chain structures were not formed. The diminished Q³_c signal (corresponding to talc-type 2∶1 layered structures) indicated a gradual decline in "transitional" M-S-H (Mg/Si = 0.9-1.1), while increased Q³_a (serpentine-like 1∶1 layers) reflected accelerated alignment of Mg/Si ratios with the raw material's composition (1.4). M-S-H gels form via Mg²⁺/H₃SiO₄^- polymerization into Q²/Q³ networks with minor Q¹, accelerated by elevated temperatures to yield serpentine-like layer structure. The M-S-H mortars demonstrated remarkable early strength (12-18 MPa at 3 d). This was attributed to the high reactivity of amorphous SiO₂ and the rapid dissolution of Mg(OH)₂. After 28 d, compressive strength reached 40 MPa at 80 ℃. The mortar exhibited superior fluidity (157-181 mm flow diameter) without any superplasticizer, enabled by 28%-32% moisture in residues. However, excessive drying shrinkage was observed.

The mixed waste acid effectively dissolved serpentine, olivine, and sepiolite, yielding Mg(OH)₂ (≥96% purity) and SiO₂-rich residues (>89% SiO₂). The polymerization degree of silicon-oxygen tetrahedra in the resulting M-S-H phases increased with rising curing temperature (from 20 ℃ to 80 ℃) and extended curing time (from 3 d to 28 d), leading to the formation of serpentine-like layered structures. The mortar exhibited high early strength (12-18 MPa at 3 d) and reached 40 MPa at 28 d, which is attributed to the high reactivity of silica residues and rapid dissolution of Mg(OH)₂. Although the inherent moisture in the residues and the absence of water consumption by Mg(OH)₂ contributed to excellent fluidity without adding superplasticizers, this combination still causes significant shrinkage.

In recent years, CO₂ curing has attracted increasing attention as it not only enhances the early age strength of cement-based materials but also captures and sequesters CO₂, thus it has been regarded as a new curing method with a broad application prospective for eco-friendly construction materials. Since CO₂ curing generally lasts several hours to several days, there are considerable residual cement clinkers in the cement matrix after CO₂ curing. These un-carbonated cement clinkers will rehydrate once in contact with water. The rehydration products of cement clinkers can recover the pH value of the pore solution, reduce the porosity, and refine the pore size distribution of the cement matrix. Thus, the subsequent hydration of CO₂-cured cementitious materials largely governs the long-term performance of cement-based materials, such as compressive strength and durability. Several researchers have studied the subsequent hydration of CO₂-cured cementitious materials, but their conclusions were contradictory. Some researchers reported that the carbonation products calcite promoted the subsequent hydration of un-carbonated cement clinkers. On the contrary, some researchers reported that CO₂ curing hindered the subsequent hydration of un-carbonated cement clinkers and thus lowered the late compressive strength of cement-based materials. Thus, more studies should be conducted to reveal the subsequent hydration mechanism and strength evolution of early CO₂-cured cement-based materials. In this investigation, the subsequent hydration and compressive strength evolution characteristics of early CO₂-cured cement paste were studied by macro-physical experiments and micro-characterization techniques.

Ordinary Portland cement P·I 42.5 complying with GB8076—2008, and de-ionized water was used to prepare cement paste samples. The water-cement ratios were 0.3, 0.4, and 0.5, respectively. Two groups of cement paste cubes measuring 20 mm were cast. One group of cement paste samples was subjected to 24 h of early CO₂ curing and subsequent standard curing in saturated lime water. The other group of cement paste cubes was subjected to standard curing in saturated lime water, which served as a reference.

After CO₂ curing, the carbonation depths of cement cubes were determined using phenolphthalein indicator, and CO₂ uptake was determined by thermal gravimetric (TG) analysis. The compressive strength of cement paste was measured at the curing ages of 2, 28 d, and 90 d, respectively. The pore solution was extracted with an expression device for pH value measurement. X-ray diffraction (XRD) and Fourier transform infrared spectroscopy (FTIR) were employed to characterize the phase evolution. The amounts of silicate and aluminate hydrates were assessed using thermal gravimetric and differential thermal gravimetric (TG-DTG) analysis. The microscopic characteristics and pore structure were examined by scanning electron microscopy (SEM) and mercury intrusion porometer (MIP) at the curing ages of 2, 28 d, and 90d, respectively.

Similar to the cement paste subjected to standard curing, the compressive strength of early CO₂-cured cement paste continues increasing as it is cured in lime water after CO₂ curing. Nevertheless, the strength improvement of the latter is lower than the former as the subsequent hydration of the former is less significant. In addition, the strength improvement of early CO₂-cured cement pastes in the lime water decreases with the increase of water-cement ratio. After curing to 28 d in lime water, the pH value of early CO₂-cured cement pastes with water-cement ratios of 0.3, 0.4, and 0.5 recovers to 13.14, 12.41, and 12.33, respectively, which is higher than the threshold of de-passivation.

XRD patterns reveal that the diffraction intensities of tricalcium silicate and dicalcium silicate decrease continuously when the early CO₂-cured cement pastes continue curing to 28 d and 90 d in lime water. However, this decrement is less significant than that of standard-cured cement paste. It indicates that CO₂ curing inhibits the subsequent hydration of un-carbonated cement clinkers. This can be attributed to the fact that the formation of calcium carbonate coating on the surface of cement particles. The FTIR spectra reveal that carbonation products silica gel is reactive towards Ca(OH)₂, and it is completely consumed before the age of 28 d. This is responsible for the absence of Ca(OH)₂ in the XRD patterns of early CO₂-cured cement paste with water-cement ratios of 0.4 and 0.5 at 28 d and 90 d, respectively. TG-DTG results suggest that less silicate and aluminate hydrates are produced in early CO₂-cured cement matrix than in standard-cured cement matrix during subsequent lime water curing, and it decreases with the increase of water-cement ratio as denser calcium carbonate coating on the surface of un-carbonated cement particles.

SEM results show that subsequent lime water curing improves the microstructure of early CO₂-cured cement paste due to precipitation of hydration products, while the improvement is less significant than that of standard-cured cement paste. Subsequent lime water curing decreases the porosity and refines the pore size distribution of early CO₂-cured cement paste. The pore diameter of early CO₂-cured cement paste mainly distributes in the range of 10 nm to 50 nm, both at 28 d and 90 d.

In the course of subsequent lime water curing, the main reactions that occurred in early CO₂-cured cement paste include the hydration of un-carbonated cement particles and the interactions between silica gel, calcium-modified silica and Ca(OH)₂. The subsequent hydration degree of early CO₂-cured cement paste is lower than that of cement paste subjected to standard curing, and it decreases with the increase of water-cement ratio. The compressive strength of CO₂-cured cement pastes continually increases with the prolonging of the subsequent curing period, and the growth rate of late compressive strength decreases with the increase of water-cement ratio.

The widespread use of cement in concrete leads to increased carbon emissions, so the demand for supplementary cementitious materials increases significantly. Limestone powder and steel slag powder are widely used as supplementary cementitious materials in modern concrete. However, for UHPC and self-compacting concrete, an extremely low water/binder (W/B) ratio is on one hand a key factor in achieving ultra-high strength and ultra-low porosity of the materials, on the other hand, also leads to the deterioration of the rheological properties of the fresh paste. Meanwhile, the existing researches focus on the influence of single limestone powder or steel slag powder on the mechanical properties or microstructure of cement-based materials. Little work on the influence of steel slag powder or limestone powder on the rheological properties of composite paste at an extremely low water/binder ratio has been investigated quantitatively. The mechanism of the effect of steel slag powder or limestone powder on the rheological properties of composite paste at extremely low water/binder ratios is still unclear. In this work, the effects of steel slag powder and limestone powder on the rheological properties of composite paste at different low water/binder ratios were analyzed via determining the flow diameter, setting time, marsh cone flow time, rheological parameters, and total organic carbon content.

A composite paste was prepared with P.I 42.5 ordinary Portland cement, steel slag powder, limestone powder, blast furnace slag and silica fume as raw materials in a certain proportion. To achieve the preparation of composite paste with extremely low water/binder ratios, a polycarboxylate superplasticizer with a water/reducing rate of 40% (Jiangsu Subot New Materials Co., Ltd., China) was used. The dosage of polycarboxylate superplasticizer for the composite paste with different water/binder ratios of 0.16 and 0.21 was 2% and 0.8%, respectively.

Composite pastes with different proportions of steel slag powder or limestone powder at water/binder ratios of 0.16 and 0.21 were prepared. The flow diameter (i.e., the larger the flow diameter, the better the fluidity), setting time (i.e., the time when the fluidity is lost) and marsh cone flow time (i.e., the shorter the flow time, the better the fluidity) were determined to analyze the fluidity of the composite pastes. the rheological properties of composite paste at a water/binder ratio of 0.16 for rheological properties tests were determined, and the rheological parameters were obtained by the H-B model. The adsorption performance of the polycarboxylate superplasticizer was analyzed by testing the TOC content.

When the ratio of water/binder is 0.16, both limestone powder and steel slag powder initially increase the flow diameter of the composite paste. However, the fluidity of the composite paste decreases over time, and the reduction is more pronounced with steel slag powder. This is because the nucleation and hydration promotion effect of limestone powder can reduce the loss rate of flow diameter, while the rough particles of steel slag powder increase the internal friction force, resulting in a decrease in the flow diameter of composite paste. Adding limestone powder and steel slag powder both shortens the setting time and marsh cone flow time of the composite paste. However, the steel slag powder addition of 30 % delays the final setting time due to its delaying effect. Also, limestone powder can enhance the thixotropy and reduce the yield stress and plastic viscosity, thereby improving the rheological properties. In contrast, steel slag powder can increase the yield stress and plastic viscosity, thereby weakening the rheological properties and thixotropy. Steel slag powder and limestone powder both can enhance the adsorption effect of polycarboxylate superplasticizer. Steel slag powder has a stronger adsorption effect. The composite paste containing limestone powder has a higher free water content. This is because the rough and porous surface of steel slag itself and the uneven particle shape lead to the physical adsorption of polycarboxylate superplasticizer molecules on the surface of steel slag particles, thereby reducing the effective content of the water reducer dispersion. Increasing the water/binder ratio to 0.21 results in a decrease in the flow diameter of the composite paste. Furthermore, the setting time and marsh cone flow time can prolong due to the reduction in the dosage of polycarboxylate superplasticizer.

The results showed that the fluidity loss rate of composite paste with limestone powder could be lower than that of composite paste with steel slag powder. Compared to steel slag powder, the addition of limestone powder reduced the setting time and marsh cone flow time of the composite paste. The addition of limestone powder could shorten the yield stress and plastic viscosity of the composite paste. Therefore, the composite paste with limestone powder had better rheological properties and stronger thixotropy rather than that with steel slag powder. The addition of limestone powder could improve the rheological properties of the composite paste. Compared to composite paste with limestone powder, a better adsorption effect of polycarboxylate superplasticizer on the surface of the composite binder with steel slag powder could be obtained. The free water content of the composite paste with limestone powder was still higher than that of the composite paste with steel slag powder. The fluidity and rheological properties of the composite paste with limestone powder could be better. The comprehensive analysis indicated that a positive effect of limestone powder on rheological properties of composite paste at an extremely low water/binder ratio could be more dominant than that of steel slag powder.

Inspired by the prestressing principle of tempered glass, this study proposes a novel strengthening method for mortar that establishes internal compressive stress without the use of steel reinforcement. This is achieved by restraining the expansion of an expansive surface layer through a shrinking substrate. The resulting pre-compressive stress significantly enhances the flexural performance of the mortar. Experimental results demonstrate that the flexural strength of unreinforced prestressed mortar (UPM) increases with the cross-sectional height ratio. When 6% expansive agent is incorporated into the surface layer and the height ratio reaches 4, the flexural strength improves to 11.24 MPa. As the expansion rate of the surface layer increases, the flexural strength of UPM continues to rise. For instance, with a 10% expansive agent, the 28 d flexural strength increases by up to 45%. Furthermore, by employing the casting method with time interval combined with surface roughening and the use of an interfacial adhesive, the flexural strength of UPM can reach 12.33 MPa. The maximum flexural-to-compressive strength ratio approaches 1∶3. These findings highlight the critical role of interfacial bonding quality in enhancing flexural performance—stronger bonding leads to greater strength gains. The existence of internal prestress is further verified through stress release monitoring during specimen cutting, where a maximum released stress of 1.68 MPa is recorded. However, it is important to note that the cutting method only releases a portion of the internal stress. As some prestress remains unreleased, the actual prestress present in UPM is likely higher than the measured value. Therefore, the released stress should not be regarded as the absolute value of the applied prestress. Although a correlation exists between stress relief and flexural strength improvement, the two are not equivalent. In practical applications, UPM should be evaluated and selected based on the overall enhancement in flexural strength rather than the measured released stress alone. Mechanistically, the expansion of the surface layer induces tensile stress in the substrate and compressive stress in the surface layer. When the surface layer is relatively thick, the compressive prestress decreases with depth along the normal direction of the interface. This stress gradient resembles that observed in tempered glass, where compressive stress is lower at the edges than at the center, rendering edge regions structurally more vulnerable. To better understand and optimize the internal stress distribution, stress redistribution, and failure mechanisms under external loading, finite element simulations are recommended. These analyses can offer theoretical insights that support the practical design and application of UPM in engineering practice. Overall, this study introduces a new approach for mortar toughening and strength enhancement, providing a foundation for the development and application of unreinforced prestressing technology in thin-walled structures and structural repair projects.

ECC is characterized by high ductility and high toughness, and the generation of a large number of fine cracks gives the material a significant self-healing ability, thereby improving its safety and durability. In some northern regions of China, the soil pH value exceeds 9 in alkaline environments, sulfate content in marine environments can generally reach 1400-2700 mg/l, structures are typically subjected to dry-wet cycles in areas of water level fluctuations, and water is a necessary condition for the occurrence of self-healing. In order to comprehensively and systematically analyze the self-healing performance of ECC under different environments and preloading conditions, the tensile and flexural properties of ECC after self-healing under three environments (5% Na₂SO₄, saturated Ca(OH)₂, and water) were investigated in this paper, providing a basis for the further application of ECC materials in engineering practice.

Three environments were designed: 1) 5% Na₂SO₄ solution (by mass) under dry-wet cycles; 2) Saturated Ca(OH)₂ solution under dry-wet cycles; 3) Water environment which served as a control group. The preloaded specimens were placed in them individually.

A total of 9 batches of dog-bone specimens under three different environments were prepared. The group preloading to 1% and 2% strain and the intact specimens without preloading were placed in the self-healing environment. After reaching the corresponding age, tensile tests were conducted to obtain stress-strain curves under different conditions, and the recovery of tensile strength and ultimate deflection was analyzed.

Besides, 9 batches of four point bending thin plates under three different environments were prepared, the ultimate strength was obtained from the intact specimens, and 20% and 40% of the value were selected as the preloading levels for bending tests. The preloaded group and intact specimens without preloading were placed in the self-healing environments for the flexural tests to obtain the load-deflection curves, the recovery of properties such as flexural strength, ultimate deflection, and flexural toughness were investigated.

Preloading to 1% strain gained a significant advantage in restoring the tensile strength of specimens at different ages in three different environments. The tensile strength of ECC was improved when subjected to dry-wet cycle in both Na₂SO₄ and Ca(OH)₂ solutions, with a supreior recovery ability observed in the Ca(OH)₂ environment. It is also noted that under a 1% preload, the tensile strength increased by about 36.33% throughout the 90 d aging process.

Reloading after pre-tensioning led to a significant improvement on the tensile ductility of specimens at 0 d, with an increase in ultimate tensile strain of up to 138.4% with the preloading level increasing from 0% to 2%. The water environment had a significant effect on the ductile healing of intact specimens without preloading at long ages, with an increase in ultimate tensile strain of 137.52% with age from 0 d to 90 d, while the Na₂SO₄ and Ca(OH)₂ environments weakened the ductility of cracked ECC specimens, with the largest decreases of 52.78%-65.39% at 2% preloading, and the Na₂SO₄ environment had a better effect on the maintenance of ductility in the specimens. At 90d of self-healing, increasing preloading levels weakened the tensile ductility of the specimens, with the greatest decrease in water.

Under the unsoaked condition, 20% preloading increased the flexural strength of the bent plate, but weakened the flexural ductility. As the age of self-healing increased, the initial flexural strength and ultimate flexural strength of the specimens under 40% preloading were improved, and Na₂SO₄ environment was more effective, with an increase of 28.91% and 40.35%, respectively, compared with that at 0 d. Preloading by 20% significantly increased the ultimate deflection, and the best effect was achieved in water, with an increase of up to 202.83% compared with 0 d. Flexural ductility was positively correlated with multi-seam cracking pattern. Na₂SO₄ and Ca(OH)₂ environments were effective in restoring the flexural strength of the specimens at a short age, and had a positive effect on the healing of microcracks. After 90 d of self-healing, the initial flexural toughness ratio and residual flexural toughness ratio of the preloaded 20% thin plate recovered best. Ca(OH)₂ environment was favorable to enhance the pre-peak flexural toughness, and the water environment was favorable to enhance the post-peak flexural toughness.

The tensile and flexural properties of ECC after self-healing in three environments were investigated. The results showed good tensile and flexural self-healing properties. A 1% preloading level was clearly beneficial for tensile strength recovery at all ages in the three environments. Although Dry-wet cycling in Na₂SO₄ and Ca(OH)₂ enhanced the tensile strength of ECC, it reduced the tensile ductility of preloaded ECC specimens. The greatest decrease in ductility occured at 2% preloading, and the Na₂SO₄ environment was more effective in maintaining it. After 90 d of self-healing, tensile ductility of the specimens decreased with increasing preloading levels. A 40% preloading level had the most significant effect on flexural strength recovery after long-term aging, particularly in the Na₂SO₄ environment. The ultimate deflection recovery of the specimen was most significant at a 20% preloading level, with better performance in water. Furthermore, 20% preloading significantly improved the flexural toughness after 90 d of self-healing.