Document Type : Research Paper
Expansive clay soils, particularly those dominated by montmorillonite, pose serious risks to infrastructure due to their high swell-shrink potential. In Iran, calcium bentonite deposits in the Kerman region are a typical example. Conventional stabilizers such as cement and lime are effective but environmentally costly. Geopolymers synthesized from waste materials offer a low-carbon alternative. Recycled glass powder (RGP), rich in amorphous silica, can serve as an aluminosilicate precursor. However, the behavior of calcium bentonite treated with an RGP-based geopolymer under ambient curing has not been systematically investigated. The key knowledge gap is whether increasing RGP content monotonically improves strength or if an optimum exists, and how microstructural changes correlate with mechanical response.
This study tests the following hypotheses: (1) The RGP-based geopolymer will increase unconfined compressive strength (UCS) of Kerman bentonite relative to untreated soil. (2) The strength gain will be time-dependent, with higher values at 28 days than at 7 days. (3) A non-monotonic relationship between RGP content and strength exists, with an intermediate drop followed by a sharp rise at high RGP content. (4) The strength improvement will be associated with formation of amorphous geopolymer gel, increased Si/Al ratio, and reduced porosity observable by XRD, SEM, and EDS.
The objectives are: (a) to determine the optimal RGP content for maximum UCS, (b) to quantify the effect of curing time, and (c) to characterize the microstructural evolution. This research is relevant for geotechnical engineers and waste management policymakers, and the pragmatic approach employs a mixed quantitative–qualitative (microstructural) design.
A completely experimental design was used. The soil was calcium bentonite from Kerman (XRF confirmed: SiO₂ 58.4%, Al₂O₃ 16.2%, CaO 4.8%, MgO 2.9%, LOI 12.5%). Recycled glass powder (RGP) obtained from municipal waste was ground to pass #200 sieve. Alkali activator was 5 M sodium hydroxide (NaOH) prepared from reagent-grade flakes.
Four RGP contents by dry weight of soil were tested: 0% (control), 15%, 20%, and 40%. For each condition, 3 replicate specimens were prepared for each curing time (7 and 28 days), totaling 4 × 2 × 3 = 24 specimens for UCS. Additional specimens were prepared for microstructural analyses (XRD, SEM, EDS) at 28 days for selected groups (0%, 20%, 40% RGP). The mixture proportion maintained a constant liquid-to-solid ratio (activator solution to total solids) of 0.35 by weight. Specimens were statically compacted in cylindrical molds (38 mm diameter, 76 mm height) to achieve dry density of 1.55 g/cm³. After demolding, they were sealed and cured at 23±2°C.
Primary outcome: unconfined compressive strength (UCS) measured at a strain rate of 1 mm/min using a universal testing machine. Secondary outcomes: strain at failure (toughness index), mineralogical phases (XRD with Cu-Kα radiation, 2θ range 5°–70°), morphology and porosity (SEM at 2000× magnification), and elemental Si/Al ratio (EDS spot analysis on three areas per sample). XRF was used for bulk oxide composition of raw materials.
No formal power analysis was conducted due to the exploratory nature. Mean UCS and standard deviation were calculated for each group. Effect size (Cohen’s d) was computed post-hoc for pairwise comparisons between 15% and 40% RGP groups at 28 days. All procedures followed institutional guidelines; no human or animal subjects were involved.
The control sample (0% RGP) at 28 days had UCS of 2.1 kg/cm² (SD=0.3). At 15% RGP, UCS increased to 23.19 kg/cm² (SD=1.2) at 28 days. At 20% RGP, strength dropped to 12.43 kg/cm² (SD=0.9). At 40% RGP, a dramatic rise occurred: UCS reached 63.22 kg/cm² (SD=2.4) at 28 days. The effect size between 15% and 40% groups was very large (Cohen’s d = 3.8). For all mixtures, 28-day values were 50–70% higher than corresponding 7-day values. Strain at failure was lowest at 20% RGP (2.3% at 28 days) and highest at 40% RGP (3.1%), indicating a transition from ductile to brittle behavior with increasing stiffness.
XRD patterns of untreated bentonite showed sharp peaks of montmorillonite (d001 ≈ 15 Å), quartz, calcite, and albite. After treatment with 40% RGP, the montmorillonite peak intensity decreased by ~60%, and a broad amorphous hump (20–35° 2θ) appeared, indicating geopolymer gel formation. SEM images revealed that untreated soil had a loose flaky texture with large pores (10–50 µm). In contrast, 40% RGP samples exhibited a dense, gel-coated surface with nearly closed pores (<5 µm). EDS analysis showed that the Si/Al ratio increased from 2.1 (control) to 3.8 (20% RGP) and to 5.2 (40% RGP) at 28 days, confirming incorporation of Si from recycled glass into the geopolymer network.
This study demonstrated that a recycled glass powder–based geopolymer (activated with 5 M NaOH) can substantially improve the strength of Kerman calcium bentonite under ambient curing. The most important finding is the non-monotonic response: strength peaks at 15% RGP (23 kg/cm²), drops at 20% (12 kg/cm²), and then soars to 63 kg/cm² at 40% RGP. This non-linearity suggests two competing mechanisms: at low to moderate RGP, reaction is limited by available activator; at 40% RGP, excess silica promotes extensive cross-linking. The microstructural evidence—amorphous XRD hump, reduced porosity in SEM, and increased Si/Al ratio—confirms the geopolymerization process.
Practical implications: For moderate strength requirements, 15% RGP suffices; for high-strength applications (e.g., pavement subbase), 40% RGP is recommended despite higher material cost. Environmentally, the use of recycled glass reduces landfill waste and avoids cement-related CO₂ emissions. A limitation is that only one activator concentration (5 M) and ambient curing were tested; future work should examine higher molarities and temperature effects. Additionally, durability under wet-dry cycles was not assessed. Nonetheless, the findings provide a robust basis for using waste glass geopolymer as a sustainable stabilizer for expansive clays in Iran.
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. The research expenses were covered personally by the authors.
Conceptualization, Seyed Morteza Marandi and Salaheddin Hamidi; methodology, Salaheddin Hamidi and Alisina Mobasherzadeh Mahani; software, Alisina Mobasherzadeh Mahani; validation, Salaheddin Hamidi; formal analysis, Alisina Mobasherzadeh Mahani; investigation, Alisina Mobasherzadeh Mahani; resources, Alisina Mobasherzadeh Mahani and Salaheddin Hamidi; data curation, Alisina Mobasherzadeh Mahani; writing—original draft preparation, Alisina Mobasherzadeh Mahani; writing—review and editing, Alisina Mobasherzadeh Mahani and Salaheddin Hamidi; visualization, Alisina Mobasherzadeh Mahani and Salaheddin Hamidi; supervision, Seyed Morteza Marandi and Salaheddin Hamidi.” All authors have read and agreed to the published version of the manuscript.
This manuscript is the experimental research and all of the results and data obtained from the different soil mechanics tests such as uniaxial tests, Atterberg limits, grain size distribution, etc.
This manuscript is the experimental research and has no associated data. Not applicable.
Acknowledgements
The authors would like to thank the staff and officials of the soil mechanics and pavement laboratory of Shahid Bahonar university of Kerman for their assistance and cooperation during the experimental work.
This study was an experimental laboratory study in the field of civil engineering and did not involve human participants, animals, human biological samples, questionnaires, interviews, or personal data. Therefore, no human or animal ethical approval was applicable to the experimental procedures. An application for ethical approval/confirmation was submitted to the relevant authority; however, no ethical code had been issued at the time of manuscript preparation.
The authors confirm that the research was conducted in accordance with principles of research integrity and that no data fabrication, data falsification, plagiarism, or other forms of research misconduct were involved.
The authors declare that they have no conflict of interest.
