Abstract
Delayed ettringite formation (DEF) is a form of internal sulfate attack that compromises concrete durability, particularly when early-age curing temperatures exceed critical limits. Once overlooked and considered a minor phenomenon, DEF gained prominence after causing cracking in precast railway sleepers and later in major infrastructure such as bridges, dams and nuclear facilities. The problem originates internally: elevated temperatures dissolve early-age ettringite which subsequently reprecipitates and expands, reducing mechanical performance and threatening structural integrity. As a result, DEF represents a significant durability concern for long-service life structures including nuclear waste containment systems.
Despite increased attention, several fundamental aspects of DEF remain insufficiently understood, including the spatial precipitation patterns, the underlying reaction mechanisms, and the combined influence of mix design and curing conditions. The multi-scale interplay between chemistry, temperature, pore structure and aggregate characteristics has not yet been comprehensively resolved, and current models lack the ability to link microscale processes to macroscale damage evolution. This study, therefore, provides an integrated, multi-scale investigation aimed at clarifying DEF mechanisms, assessing structural risk, and informing prevention strategies.
Thermodynamic modelling was used to explore stability domains and reaction pathways of sulfateand aluminate-bearing hydrates across a wide range of cementitious systems. DEF-governing reaction paths were identified and validated against a benchmarked chemical model. Coupling the chemical model with a hydration heat model enabled sensitivity analyses and simulations of DEF development in cemented waste drums. The effects of SCM incorporation, waste-sludge content, early-age temperature profiles, heating duration, and alkali, sulfate, and aluminate concentrations were quantified. Global sensitivity analyses established the relative influence of individual parameters and their interactions, yielding a mechanistic understanding of DEF from chemical origins to structural-scale implications.
At the micro- and meso-scale, an extensive experimental program examined how aggregate volume and size affect precipitation behavior, expansion kinetics, crack morphology, and mechanical degradation under accelerated DEF conditions. Mortar and paste samples were characterized using XRD/QXRD, N₂ adsorption, MIP, SEM, and optical microscopy. Machine learning techniques were employed for high-resolution crack segmentation, and compressive strength testing assessed mechanical consequences.
Thermodynamic results show that ettringite is stable at ambient temperature but destabilizes at elevated temperatures in mixtures with high aluminum or low calcium content. Its decomposition is Al-dependent and forms phases such as monosulfoaluminate, katoite, CASH, and aluminum hydroxide. Two principal and one secondary DEF pathways were identified, determined by CaO/SiO2 and Al2O3 ratios, and found to be consistent with the benchmarked chemical model. The coupled chemo-thermal model demonstrated that DEF severity depends strongly on drum size, waste loading, SCM content, and the thermal history of hydration. High temperatures or prolonged heating do not necessarily induce severe DEF. Thermal effects dominate at low alkali contents, whereas high alkali contents may suppress DEF even under otherwise favorable conditions.
Microscale analysis revealed a strong dependence of DEF mechanisms on sand content. Delayed ettringite preferentially formed in larger pores (250-1000 nm) in sand-rich mortars, while in paste and low-sand mortars precipitation also occurred in smaller pores (70-250 nm). Increasing sand content reduced interparticle spacing and accelerated precipitation. Crack segmentation showed that high sand content produced extensive internal cracking and large surface cracks (>50 µm), while low sand content led to finer, more distributed cracks (<50 µm). Only cracks >50 µm consistently reduced compressive strength, indicating that expansion alone is not a reliable indicator of DEF-related mechanical degradation.
This multi-scale investigation advances the mechanistic understanding of DEF across diverse material compositions and environmental conditions. The findings provide guidance for refining predictive models and improving DEF assessment in civil engineering and nuclear waste applications.
Despite increased attention, several fundamental aspects of DEF remain insufficiently understood, including the spatial precipitation patterns, the underlying reaction mechanisms, and the combined influence of mix design and curing conditions. The multi-scale interplay between chemistry, temperature, pore structure and aggregate characteristics has not yet been comprehensively resolved, and current models lack the ability to link microscale processes to macroscale damage evolution. This study, therefore, provides an integrated, multi-scale investigation aimed at clarifying DEF mechanisms, assessing structural risk, and informing prevention strategies.
Thermodynamic modelling was used to explore stability domains and reaction pathways of sulfateand aluminate-bearing hydrates across a wide range of cementitious systems. DEF-governing reaction paths were identified and validated against a benchmarked chemical model. Coupling the chemical model with a hydration heat model enabled sensitivity analyses and simulations of DEF development in cemented waste drums. The effects of SCM incorporation, waste-sludge content, early-age temperature profiles, heating duration, and alkali, sulfate, and aluminate concentrations were quantified. Global sensitivity analyses established the relative influence of individual parameters and their interactions, yielding a mechanistic understanding of DEF from chemical origins to structural-scale implications.
At the micro- and meso-scale, an extensive experimental program examined how aggregate volume and size affect precipitation behavior, expansion kinetics, crack morphology, and mechanical degradation under accelerated DEF conditions. Mortar and paste samples were characterized using XRD/QXRD, N₂ adsorption, MIP, SEM, and optical microscopy. Machine learning techniques were employed for high-resolution crack segmentation, and compressive strength testing assessed mechanical consequences.
Thermodynamic results show that ettringite is stable at ambient temperature but destabilizes at elevated temperatures in mixtures with high aluminum or low calcium content. Its decomposition is Al-dependent and forms phases such as monosulfoaluminate, katoite, CASH, and aluminum hydroxide. Two principal and one secondary DEF pathways were identified, determined by CaO/SiO2 and Al2O3 ratios, and found to be consistent with the benchmarked chemical model. The coupled chemo-thermal model demonstrated that DEF severity depends strongly on drum size, waste loading, SCM content, and the thermal history of hydration. High temperatures or prolonged heating do not necessarily induce severe DEF. Thermal effects dominate at low alkali contents, whereas high alkali contents may suppress DEF even under otherwise favorable conditions.
Microscale analysis revealed a strong dependence of DEF mechanisms on sand content. Delayed ettringite preferentially formed in larger pores (250-1000 nm) in sand-rich mortars, while in paste and low-sand mortars precipitation also occurred in smaller pores (70-250 nm). Increasing sand content reduced interparticle spacing and accelerated precipitation. Crack segmentation showed that high sand content produced extensive internal cracking and large surface cracks (>50 µm), while low sand content led to finer, more distributed cracks (<50 µm). Only cracks >50 µm consistently reduced compressive strength, indicating that expansion alone is not a reliable indicator of DEF-related mechanical degradation.
This multi-scale investigation advances the mechanistic understanding of DEF across diverse material compositions and environmental conditions. The findings provide guidance for refining predictive models and improving DEF assessment in civil engineering and nuclear waste applications.
| Original language | English |
|---|---|
| Qualification | Doctor of Science |
| Awarding Institution |
|
| Supervisors/Advisors |
|
| Date of Award | 19 May 2026 |
| Publisher | |
| State | Published - 19 May 2026 |
Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver