In a study published in Case Studies in Construction Materials, Aymane Elyazidi, Fatiha Berroug, Lahcen Boukhattem, and Rachid Bouferra tackled a stubborn gap in the literature: predicting the compressive strength of geopolymers made from fly ash and crushed waste glass. Fly ash, a byproduct of coal combustion, is the workhorse precursor of geopolymer chemistry, while glass powder — ground from discarded bottles and windows — is rich in the amorphous silica that fuels the geopolymerisation reaction. Prior experiments have shown that substituting a portion of fly ash with glass powder can yield compressive strengths exceeding 50 megapascals, but the interplay of chemical composition, particle size, activator concentration, and curing conditions is so complex that no single experiment can map it.

The researchers assembled a database of 429 experimental records drawn from eleven peer-reviewed studies, spanning mortar and paste formulations with wildly different mix designs. Fourteen input variables describe each record, including the quantities of fly ash, glass powder, sand, sodium silicate, and sodium or potassium hydroxide, the oxide composition of the precursors, the fineness of the glass particles, and the curing regime of temperature, duration, and age. Compressive strength values in the database range from nearly zero to 96 megapascals, with a mean of 17.38 megapascals — a spread that captures the full drama of geopolymer formulation, from failed mixes to high-performance binders.

Against this dataset, the team benchmarked seven machine learning algorithms: a multi-layer perceptron neural network, support vector regression, random forest, AdaBoost, CatBoost, gradient boosting, and extreme gradient boosting, known as XGBoost. Each algorithm was first tuned with conventional manual settings, and then re-tuned through a hybrid scheme in which Differential Evolution, a population-based evolutionary optimizer, searched the hyperparameter space. Differential Evolution works by mutating candidate solutions with scaled vector differences between randomly chosen population members, then keeping whichever version performs better — a strategy the authors argue is ideally suited to the mixed, non-differentiable, and multimodal landscape of boosting hyperparameters, where tree depth, subsampling ratios, and regularization terms interact in ways a grid search cannot efficiently explore.

The results were dramatic. Among the conventionally tuned models, support vector regression led the pack with a coefficient of determination of just 0.8602 on the test set. After Differential Evolution optimization, every model improved, and the DE-XGBoost hybrid rose to the top with an R-squared of 0.9913, a root mean squared error of 1.613 megapascals, and a mean absolute error of 1.133 megapascals on unseen data. The DE-optimized gradient boosting model achieved a perfect training score, effectively interpolating its training points, while a Taylor diagram analysis confirmed that the hybrid models reproduced the observed variability of the data with correlation coefficients exceeding 0.97. The authors emphasize that the close alignment between training and test performance indicates stable generalization rather than overfitting.

Accuracy alone, however, is not enough for a field that needs physical insight, so the team turned to SHAP analysis, a technique from cooperative game theory that decomposes each prediction into the marginal contributions of every input variable. The verdict was striking: the molarity of the sodium hydroxide activator emerged as the single most decisive factor governing predicted strength, followed by the silica content of the precursor blend. High activator molarity consistently pushed predictions upward, consistent with its chemical role in dissolving the aluminosilicate precursors and releasing silicate and aluminate ions — the first step of geopolymerisation. Age and calcium oxide content also contributed positively, reflecting the slow maturation of the geopolymeric network and the co-precipitation of calcium-rich hydrates in ambient-cured systems.

Perhaps the most counterintuitive finding concerned silica itself. Although silica is the principal network former of the geopolymer gel, high silica content correlated with lower predicted strength. The authors offer a mechanistic explanation: in a fly ash–glass powder system, high total silica is effectively the signature of a glass-rich blend, since soda-lime glass contains 70 to 73 percent silica compared with 45 to 55 percent for most fly ashes. Excess silica raises the silicon-to-aluminum ratio beyond the window in which fly ash geopolymers develop their highest strengths, leaving unreacted siliceous phases that delay setting and leave a more porous matrix. The strong positive contribution of calcium oxide reinforces this picture, showing that strength in much of the database is carried by the coexistence of two hydrate families rather than by silica alone.

Three-dimensional partial dependence plots extended this analysis into the realm of practical mix design. The surfaces revealed that the optimal formulation keeps glass powder replacement below roughly 20 percent of the fly ash, maintains the alkaline activator at around 12 moles per liter, and holds the total silica content of the precursor blend within a median range of 35 to 45 percent. Curing temperature showed its strongest effect between 60 and 80 degrees Celsius, with a thermal curing duration of about 24 hours. The researchers are careful to note that these recommendations describe the densely populated region of their database rather than a universal chemical threshold, since well-optimized high-replacement mixtures remain scarce in the compiled literature.

Compared with previous machine learning studies on similar materials — including AdaBoost models for glass powder mortar at R-squared 0.94 and Sparrow Search Algorithm-optimized XGBoost for glass powder concrete at 0.9645 — the DE-XGBoost framework sets a new benchmark, though the authors caution that differing datasets and feature spaces make direct rankings imperfect. They also acknowledge the study’s limits: the oxide inputs represent bulk compositions rather than reactive amorphous fractions, no source-wise cross-validation was performed, and the evolutionary search used a fixed random seed. Looking ahead, the team plans to embed the predictor within multi-objective optimization loops that balance strength against cost and embodied carbon, to release an interactive web tool exposing the model and its SHAP decompositions, and to explore transfer learning so that the knowledge captured in these 429 records can extend to slag- and metakaolin-based geopolymer families. If they succeed, the mountains of waste glass piling up in landfills may find a second life as the backbone of a low-carbon built environment — with an algorithm pointing the way.

Subject of Research: Differential Evolution-optimized machine learning prediction of compressive strength in fly ash and glass powder geopolymers

Article Title: DE-optimized machine learning for compressive strength prediction of glass powder and fly ash geopolymers

Article References: Elyazidi, A., Berroug, F., Boukhattem, L., & Bouferra, R. (2026). DE-optimized machine learning for compressive strength prediction of glass powder and fly ash geopolymers. Case Studies in Construction Materials, 25, Article e06617. https://doi.org/10.1016/j.cscm.2026.e06617

Image Credits: AI Generated

DOI: 10.1016/j.cscm.2026.e06617

Keywords: geopolymers, glass powder, fly ash, machine learning, Differential Evolution, XGBoost, compressive strength, SHAP, sustainable construction, cement emissions, hyperparameter optimization, partial dependence plots

Geopolymers are aluminosilicate materials formed when industrial precursors such as coal fly ash are activated with alkaline solutions. Rather than burning energy in high-temperature kilns the way ordinary Portland cement does, geopolymers harden through a chemical dissolution-and-reassembly process, weaving amorphous networks of aluminate and silicate units. Because fly ash is an abundant byproduct of coal combustion, geopolymers built from it are cheap, and they have long been investigated both for waste encapsulation and as low-cost adsorbents for removing heavy metals and radionuclides from contaminated water.

The challenge addressed in the new study concerns trivalent actinides, a family of radioactive elements that includes americium-241, a component of spent nuclear fuel and legacy weapons-site waste. In aqueous environments, trivalent actinides behave chemically much like their lanthanide cousins, which is why the researchers used europium(III) as a nonradioactive chemical surrogate alongside trace amounts of americium-241 as a radiotracer. This pairing allowed the team to measure uptake quantitatively and to probe the chemistry of the bound metal ions with spectroscopic precision.

The researchers prepared two materials: a pristine fly ash geopolymer, designated FA-GP, and a modified version, FA-Al-GP, in which gamma-alumina was integrated into the geopolymer framework. Structural characterization confirmed that the γ-Al₂O₃ had genuinely become part of the matrix rather than merely sitting in it as an inert filler. Nitrogen adsorption measurements using the Brunauer–Emmett–Teller method revealed something unexpected: the modification cut the specific surface area from 41 square meters per gram in the pristine material down to just 16 square meters per gram. By the usual logic of adsorption science, that alone should have crippled its performance.

Solid-state nuclear magnetic resonance told a different story. Using ²⁷Al magic-angle spinning NMR, the team found that the modified geopolymer contained a higher proportion of octahedrally coordinated aluminum sites. These six-fold coordinated aluminum centers, along with associated aluminol groups, act as Lewis-basic anchors, chemically receptive to hard trivalent metal cations. In effect, the alumina modification traded sheer surface for a denser matrix whose surfaces are far richer in the specific chemical functionality that binds actinides and lanthanides.

Batch sorption experiments bore this out decisively. The modified geopolymer removed more than 97 percent of the target metal ions across a broad pH window from 4.0 to 8.0, a range relevant to many real waste streams and groundwater conditions. Its maximum adsorption capacity reached approximately 66 milligrams per gram at 328 kelvin, compared with roughly 39 milligrams per gram for the pristine fly ash geopolymer. In other words, the material with 60 percent less surface area captured about 70 percent more metal per unit mass.

Kinetic and thermodynamic analysis showed that uptake in both materials followed pseudo-second-order kinetics and fit Langmuir isotherms, the classic signatures of monolayer chemisorption. The process was found to be spontaneous and endothermic, meaning the metal ions form genuine chemical bonds with the surface rather than simply sticking through weak physical forces. That distinction matters enormously for nuclear waste management: chemically bound radionuclides resist being leached back out by changes in water chemistry, whereas physisorbed contaminants can be released when conditions shift.

To see exactly where and how the metals were binding, the team turned to time-resolved fluorescence spectroscopy and X-ray photoelectron spectroscopy. The fluorescence measurements, which exploit the sharp, environment-sensitive emission of europium(III), provided molecular-level evidence about the coordination environment of the sorbed ions, while XPS tracked shifts in binding energies that revealed the formation of strong inner-sphere complexes. Both techniques pointed to the same conclusion: the reactive aluminol and silanol groups on the geopolymer surface are the active sites that grip the trivalent metals directly, displacing their hydration shells.

The broader implication of the study is a design principle for next-generation sorbents. Rather than chasing ever-larger surface areas, materials engineers can boost performance by concentrating highly accessible, chemically reactive sites on the surfaces they already have. Because surface chemistry, not bulk area, dictates capacity, targeted modification strategies like alumina integration could improve a wide range of geopolymer and mineral sorbents. The approach also strengthens the case for geopolymers as dual-purpose nuclear materials, capable of both immobilizing waste in solid form and scrubbing actinides from aqueous streams, all while converting an industrial waste product, coal fly ash, into a functional environmental technology.

For the nuclear industry, where cleanup costs at legacy sites run into the billions and the safe management of actinides remains one of the field’s hardest problems, a sorbent made from power-plant ash that outperforms its unmodified counterpart is more than a laboratory curiosity. The work, led by Atanu Das and Aishwarya Soumitra Kar with colleagues at the Bhabha Atomic Research Centre, Homi Bhabha National Institute, and the Tata Institute of Fundamental Research, suggests that smarter surface engineering of abundant, inexpensive materials may offer a practical path toward more efficient actinide sequestration in contaminated waters around the world.

Subject of Research: Enhanced sorption of trivalent actinides by γ-Al₂O₃-modified fly ash-based geopolymers

Article Title: Structural evolution and enhanced sorption performance of γ-Al₂O₃-modified fly ash-based geopolymers for trivalent actinide sequestration

Article References: Das, A., Kar, A. S., Arunachalam, V., Tyagi, D., & Parayil, R. T. (2026). Structural evolution and enhanced sorption performance of γ-Al₂O₃-modified fly ash-based geopolymers for trivalent actinide sequestration. Environmental Science and Pollution Research, 33(28), 14505-14526. https://doi.org/10.1007/s11356-026-38180-2

Image Credits: AI Generated

DOI: 10.1007/s11356-026-38180-2

Keywords: geopolymers, fly ash, gamma-alumina, actinides, americium-241, europium, sorption, chemisorption, nuclear waste, radionuclide sequestration, XPS, fluorescence spectroscopy

The review, led by a team at Guangdong University of Technology with collaborators at Walailak University and Shenzhen University, synthesizes recent advances in multiscale characterization and computational modeling of geopolymer materials. Its central message is that the field is undergoing a structural transition, documented through bibliometric analysis, away from conventional phase and composition identification and toward precise atomic-scale analysis, dynamic in situ monitoring, and data-driven material design. In other words, geopolymer science is maturing from a descriptive craft into a quantitative, predictive discipline, and the tools driving that shift are the same ones that transformed semiconductor physics and structural biology: synchrotron radiation, neutron scattering, advanced electron microscopy, molecular simulation, and machine learning.

What makes geopolymers so difficult to characterize is their peculiar structural personality. Unlike crystalline ceramics, they are ordered over short and medium ranges but completely disordered over long ranges, forming three-dimensional aluminosilicate networks in which alkali cations balance the negative charge of aluminum tetrahedra. This amorphous nature defeats the workhorse technique of traditional materials science, X-ray diffraction, which relies on periodic lattice order to produce interpretable patterns. The situation is compounded by the growing trend of blending multiple industrial solid wastes into a single binder, which introduces substantial compositional complexity, heterogeneous gel formation, and dynamic structural reconstruction as the material cures and ages.

To pierce this amorphous fog, the review highlights the power of advanced spectroscopic techniques. Solid-state nuclear magnetic resonance spectroscopy, in particular, has become indispensable for resolving local atomic configurations, distinguishing silicon coordination environments, quantifying network connectivity, and identifying how heteroatoms such as iron, magnesium, and heavy metals are incorporated into the gel framework. Infrared spectroscopy, long used to track the progressive shift of silicon-oxygen-aluminum stretching bands during geopolymerization, now operates in time-resolved and spatially resolved modes that can watch gel nucleation unfold in real time. For iron-rich precursors such as red mud and volcanic ash, Mössbauer spectroscopy reveals the oxidation state and coordination of iron species, a critical variable because iron can either participate in the binder network or remain as inert inclusions that dictate strength and durability.

Perhaps the most striking technical development the review documents is the rise of pair distribution function analysis, a total-scattering method that extracts structural information from the diffuse scattering that conventional diffraction discards. Applied with synchrotron X-rays and neutrons, pair distribution function analysis has allowed researchers to reconstruct the atomic structure of geopolymer gels directly, resolving debates about the role of charge-balancing extra-framework aluminum and tracking how local structure evolves during the earliest stages of gel formation. In situ neutron pair distribution function experiments have even captured the structural evolution of geopolymer gels as they form, while related work has mapped how local structure changes with temperature, carbonation, sulfate attack, and high-temperature exposure, connecting atomic-scale chemistry to the degradation mechanisms that matter in service.

At larger length scales, advanced microscopy is revealing the hierarchical architecture that atomic spectroscopy cannot see. Scanning and transmission electron microscopy, increasingly combined with focused ion beam sample preparation and cryogenic transfer, expose the nanoscale morphology of gel particles and the interfacial transition zones that govern composite behavior. Cryo-electron microscopy has delivered a genuinely viral result in the cement science community: direct observation of multistep nucleation and growth of aluminosilicate gel, showing that these amorphous binders assemble through intermediate phases rather than simple precipitation. Meanwhile, atomic force microscopy and nanoindentation map mechanical heterogeneity across gel phases and interfaces, and quantitative backscattered electron imaging paired with energy-dispersive spectroscopy now allows full-component characterization of interfacial zones in recycled and fiber-reinforced geopolymer concretes.

Pore structure, the hidden skeleton that controls transport, durability, and freeze-thaw resistance, has likewise come into sharper focus. X-ray and neutron tomography reconstruct three-dimensional pore networks non-destructively, from laboratory micro-computed tomography to hard X-ray nanotomography capable of resolving features tens of nanometers across. Low-field nuclear magnetic resonance relaxometry has emerged as a rapid, non-destructive probe of pore size distribution and water states, tracking setting, reaction kinetics, and even the influence of paramagnetic iron in waste-derived binders. Small-angle neutron scattering resolves bimodal pore evolution during early curing, and neutron radiography and tomography visualize capillary water absorption and moisture transport in real time, exploiting the exceptional sensitivity of neutrons to hydrogen.

The review also emphasizes that simulation and artificial intelligence are no longer peripheral add-ons but central pillars of the field. Molecular dynamics simulations of sodium aluminosilicate hydrate and calcium aluminosilicate hydrate gels now illuminate dissolution mechanisms, adsorption of heavy metals and radionuclides, ion migration in nanopores, and the interfacial bonding between gels and aggregates or fibers, often in direct dialogue with experimental scattering data. On the data side, machine learning models predict compressive strength, freeze-thaw degradation, bond strength, and multi-objective performance trade-offs involving cost and carbon dioxide emissions, with interpretable and transfer-learning frameworks extending predictions to ultra-high-performance geopolymer systems. Emerging machine learning interatomic potentials promise to bring quantum-level accuracy to simulations of cementitious systems at unprecedented scale.

Why does all this matter beyond the laboratory? Geopolymers sit at the intersection of two global imperatives: decarbonizing construction and managing mountains of industrial waste. They have demonstrated potential in marine structures, road construction, nuclear and heavy metal waste immobilization, fire-resistant panels, 3D-printed elements, and even neutron-shielding composites for radiation protection. But every one of those applications depends on reproducible performance, and reproducibility demands structure-property relationships grounded in measurement rather than recipe. The characterization toolkit assembled in this review is precisely what converts a variable waste stream into an engineering material whose atomic network, gel hierarchy, and pore architecture can be tuned deliberately.

The authors’ synthesis points toward a future in which synchrotron beamlines, neutron facilities, automated microscopy, and machine learning pipelines operate as an integrated design engine, closing the loop from composition to structure to property to prediction. Their bibliometric evidence shows the literature itself shifting toward dynamic monitoring and data-driven design, a signal that the community is already building that engine. For a material class once dismissed as too messy for fundamental science, geopolymers are becoming a showcase for how modern multiscale characterization can tame even the most disordered matter, and the payoff may be a generation of sustainable, high-performance binders designed on the computer before they are ever mixed in the lab.

Subject of Research: Multiscale structural characterization and structure-property relationships of geopolymer materials

Article Title: Review: advanced characterization of geopolymer materials-from multiscale structural understanding to structure-property relationships

Article References: He, X., Pantongsuk, T., Chen, W., Deng, R., Yu, T., & Zhang, B. (2026). Review: advanced characterization of geopolymer materials-from multiscale structural understanding to structure-property relationships. Journal of Materials Science. https://doi.org/10.1007/s10853-026-13826-1

Image Credits: AI Generated

DOI: 10.1007/s10853-026-13826-1

Keywords: geopolymers, aluminosilicate binders, low-carbon cement, solid-state NMR, pair distribution function, synchrotron radiation, neutron scattering, electron microscopy, molecular dynamics, machine learning, industrial solid waste, structure-property relationships