Natural Clay in Geopolymer Concrete: A Sustainable Alternative Pozzolanic Material for Future Green Construction—A Comprehensive Review

Natural Clay in Geopolymer Concrete: A Sustainable Alternative Pozzolanic Material for Future Green Construction – A Comprehensive Review

Manuscript ID: Sustainability-3946454

We thank the editor and reviewers for their thorough review and thoughtful comments and suggestions, which have further helped us improve our manuscript. We have carefully addressed all comments and carried out the necessary revisions. In this document, our responses are highlighted in blue, while the corresponding changes made in the manuscript are highlighted in red. The revised manuscript is submitted in Word format alongside this document.

Thank you for being so considerate.

Kind Regards

Bidur Kafle and coauthors

Reviewer #1 Comments: This review article (sustainability-3946454) addresses the need for sustainable alternatives to Ordinary Portland Cement, which contributes over 5% of global CO₂ emissions. It highlights geopolymer concrete as an eco-friendly option and explores the potential of locally available clays such as Kaolin, Halloysite, Illite and Bentonite to reduce reliance on declining industrial by-products like fly ash.  After reviewing the manuscript, I have the following minor comments:

Abstract

Clearly define all technical abbreviations ANN, MEP, XGB, BR upon first use.

The technical abbreviations ANN, MEP, XGB, and BR have been added to lines 37-38. We also ensured that other abbreviations were defined on first use.

Emphasise why natural clays represent a strategic alternative, especially in light of the decreasing availability of industrial by-products.

Thank you for your valuable comment. New information has been added to the abstract, lines 17-22, to address this.

The impact of clay on workability is negative, briefly outlining mitigation strategies used in literature

The mitigation strategies of negative impact on workability have been added in lines 31 to 32.

Keywords: I would recommend changing "Sustainable" into "Sustainable construction materials".

Thank you for this recommendation. We have changed it as per your recommendation.

Introduction

Enhance the explanation of geo-polymerisation chemistry by including a brief summary of its key reaction mechanisms.

Thank you for this suggestion. We have explained the geopolymerization chemistry and mechanisms in lines 57-67.

Please add a paragraph comparing the mechanical properties and durability of geopolymer concrete with those of traditional OPC.

A new paragraph has been added in lines 74-95 comparing the mechanical and durability properties of GPC and OPC.

Please explain the criteria for selecting natural clays as SCMs by detailing the mineralogical properties relevant to geopolymer reactivity.

Thank you for your valuable comment. However, the detailed criteria and requirements for a material to qualify as SCMs are already listed in Table 2.

Additionally, we have added the selection criteria for a clay to be an SCM, as defined by ASTM. We have added the following information in lines 109 to 113:

However, there are some criteria to choose a clay as an SCM. According to ASTM, the total amount of SiO₂, Al₂O₃, and Fe₂O₃ should be at least 70% by weight, ≤ 34% should be retained on a 45 µm sieve, the moisture content should be ≤ 3.0%, and the Strength Activity Index (SAI) should be at least ≥ 75% than that of the control mix [117].

Please add sub-title to point out potential challenges like variability in composition and activation complexity.

Thank you for this significant suggestion. We have added a sub-title below the Introduction part, where the potential challenges involved in incorporating natural clays as SCM have been discussed in lines 139 to 153. We have added the following discussion:

Potential Challenges in Utilising Clay as an SCM:

Clay-based SCMs offer promising sustainability benefits; however, their practical implementation encounters various challenges. A primary challenge lies in the variability of mineral composition, as natural clays exhibit considerable differences in reactive minerals such as kaolinite, illite, and montmorillonite. This variability influences pozzolanic reactivity and complicates quality control processes, primarily when clays originate from diverse geological sources [35]. Furthermore, the complexity of activation presents a significant challenge: clays generally require thermal treatment (calcination) at precisely controlled temperatures—typically between 600 and 950 °C- to achieve proper reactivity. Incorrect calcination, whether too high or too low, can negatively impact performance. Moreover, the energy-intensive nature of this process raises concerns about sustainability [36,37]. Furthermore, impurities such as quartz, carbonates, or organic matter can disrupt hydration reactions and diminish the effectiveness of the SCM process [38]. These factors underscore the need for rigorous characterisation and tailored activation protocols when incorporating clay into cementitious systems.

Section 2

Please discuss the challenges in surface modification and activation methods specific to each clay type.

The challenges in surface modification and activation methods have been added for specific types of Kaolin (lines 198 to 203), Illite (lines 225 to 236), Bentonite (lines 254 to 261), Halloysite (lines 286 to 296), Palygorskite (lines 311 to 318), Sepiolite (lines 339 to 350).

Please provide a more detailed discussion on how minor chemical components, such as Fe₂O₃, influence geopolymerization and mechanical performance.

A more detailed discussion on how minor chemical components influence geopolymerisation and mechanical performance has been added for Kaolin (lines 187 to 197), Illite (lines 216 to 224), Bentonite (lines 248 to 253), Halloysite (lines 273 to 285), Palygorskite (lines 308 to 310), Sepiolite (lines 327 to 338).

Section 4

Please include a clearer synthesis of the contradictory findings on raw versus calcined Bentonite.

Thank you for your valuable comment. We have provided a clearer synthesis of the contradictory findings on raw versus calcined Bentonite, described in lines 632-635 and 643-653.

In lines 632 to 635, we have added the following information:

This phenomenon is consistent with typical findings in clay activation, in which thermal treatment can cause interlayer spaces to collapse and reduce the number of active sites, resulting in a less absorptive or workable material in fresh concrete mixes [64].

In lines 643 to 653, we have added the following information:

The difference in slump behaviour between calcined Bentonite observed by Waqas et al. [146]  and Reddy et al. [149]  can be explained by the nature and extent of calcination and its impact on Bentonite's physical and chemical properties. Moderate calcination, as reported by Waqas et al. [146], reduces surface area due to partial dehydroxylation and collapse of interlayer spaces, thereby lowering water demand and improving workability, resulting in a slight increase in slump. In contrast, intense calcination at higher temperatures (700–800°C), as studied by Reddy et al.  [149], can cause structural disruption and increased internal porosity, creating more reactive sites that absorb water more aggressively, thereby reducing slump. Furthermore, the replacement level of Bentonite and the interaction with other mix components, such as fly ash and polypropylene fibres, significantly influence the rheological behaviour of the concrete, making direct comparisons between studies complex and context-dependent.

Section 5

As a review article, the tone in this section should remain analytical, neutral, and well-synthesised, avoiding overly descriptive language.

Thank you for this valuable suggestion. We have revised Section 5, making it neutral and reducing overly descriptive language. From lines 727 to 742 and 778 to 803, we have reduced the over-description, remaining analytical and well-synthesised.

Summarise optimal replacement levels for each clay type in a comparative table.

Thank you for this recommendation. However, instead of drawing a new table for the optimal replacement, we have added it to Table 5. Since this table summarises different studies on different types of clay, we believe a summary of optimal replacement in the same table would be helpful to readers. Therefore, we have changed the column title from "Remarks" to "Summary and optimal replacement" and added the information.

Please maintain consistency in terminology in the whole manuscript, use either "treated" or "calcined", and ensure clear distinctions between clay types and treatment methods.

Thank you for this suggestion. We have tried our best to maintain consistency in terminology. However, "treated" and "calcined" were used differently in the text. "Treated" means the clay might have undergone either heat treatment, also called the calcination process, or mechanical activation or other methods, such as chemical activation.

Section 6

Please specify the advantages and limitations of each model, with particular focus on those applied to clay-based systems.

The advantages and limitations have been discussed in lines 984-1010. We have added the following information to the manuscript:

In the context of clay-based geopolymer concrete systems, machine learning models have shown significant promise in predicting mechanical properties. However, there are some limitations too. This section summarises the advantages and shortcomings of the models discussed above. Machine learning models, particularly ANN, effectively forecast properties such as compressive strength. ANN surpasses alternative models in accuracy and demonstrates proficiency in managing complex relationships between mix components and curing factors, as exemplified by its successful application to nano-silica modifications.

Nevertheless, ANN requires extensive datasets and is frequently regarded as a "black box' owing to its limited interpretability [218]. MEP offers explicit mathematical formulations, enhancing interpretability and facilitating integration within engineering contexts. However, it may lack the accuracy of more sophisticated models such as XGB or ANN. XGB is proficient at handling high-dimensional data, capturing intricate patterns, and often achieving superior accuracy compared to other ensemble techniques. Nevertheless, XGB models are computationally intensive and less transparent, which may limit their use in engineering applications that require interpretability [219,220]. BR provides probabilistic forecasts that are valuable for uncertainty quantification; however, it may encounter challenges when applied to nonlinear data in geopolymer systems. LSTM networks, primarily designed for time-series analysis, can model dependencies in the curing process but are limited in predicting static properties such as compressive strength and necessitate extensive data [220,221]. Support Vector Regression (SVR) demonstrates strong performance, particularly with limited datasets, and exhibits effective generalisation. A study indicated that SVR marginally surpassed ANN in predicting the compressive strength of geopolymer fibre-reinforced concrete. Nevertheless, the outcomes of SVR depend on kernel selection and parameter tuning, which can pose challenges. Each model possesses distinct advantages, and the optimal choice is influenced by dataset characteristics, interpretability requirements, and resource availability [222].

Please discuss the limited data and modelling efforts for non-traditional clays such as Halloysite and Palygorskite to highlight research gaps and encourage further study.

The limited data and modelling efforts for non-traditional clays, such as Halloysite and Palygorskite, have been highlighted to identify research gaps and encourage further study in lines 1015 to 1024. In the manuscript, we have added the following discussion:

However, research on non-traditional clays, such as halloysite and palygorskite, in geopolymer concrete is notably scarce, indicating a significant gap in the existing literature. These clays, characterised by their distinctive tubular and fibrous morphologies, possess the potential to enhance mechanical and durability properties. However, their intricate structures and diverse compositions pose challenges for modelling efforts. Consequently, there are limited applications of machine learning techniques, such as predicting properties like compressive strength and durability, for geopolymers based on halloysite and palygorskite. This paucity of data and modelling frameworks underscores the necessity for further experimental and computational investigations to harness their potential and develop dependable predictive tools for these understudied materials.

Section 7

Please match the recommendations to real-world construction applications to enhance their practical relevance.

Thank you for this recommendation. We have added a new subsection (8.2 now) to describe the real-world construction application and to enhance the practice of different types of clays, presented in lines 1121 to 1134. In the manuscript, we have added the following discussion:

8.2 Recommendations for real-world applications

Different clay minerals possess distinct properties that can be utilised in various construction applications. For example, Kaolin, with its high purity and reactivity, is well-suited for the manufacturing of high-strength structural components and precast elements. Illite, with moderate reactivity, is suitable for general-purpose concrete used in low-load-bearing structures. Bentonite, recognised for its swelling and sealing properties, is advantageous in underground construction and serves as a self-healing agent in water-retaining structures. Halloysite, characterised by its nanotubular structure, demonstrates potential for use in lightweight panels and high-performance composites due to its capacity to enhance mechanical strength and durability. Palygorskite, characterised by its fibrous structure and thermal stability, is ideally suitable for fire-resistant panels and infrastructure in demanding environments. Sepiolite, characterised by its high porosity and water-retention capabilities, can be utilised in permeable concrete and in environmentally sustainable construction materials.

Section 8

Stress the need to overcome challenges in AI modelling by proposing avenues for further investigation.

Thank you for your valuable comment. We have added the following discussions in lines 1188 to 1195 to address your comment:

To effectively utilise AI in modelling clay-based geopolymer concrete, targeted research needs to overcome existing challenges. A significant issue is the scarcity of high-quality, diverse datasets, particularly for clays, which limits the training and validation of machine learning models. Future research should focus on creating hybrid models that combine the high accuracy of deep learning with the interpretability of symbolic or rule-based systems. These efforts will help bridge the gap between AI capabilities and engineering requirements, leading to more reliable and explainable predictive tools for sustainable construction.

Including a summary table to illustrate ideal treatment methods and performance results across various clay types would be beneficial.

Thank you for your suggestion. We have added a new table (Table 9) in the conclusion section to summarise the ideal treatment method and performance results across various clay types.

Reviewer #2 Comments: The paper is relevant, given the environmental concerns associated with Ordinary Portland Cement (OPC) and the uncertain supply of traditional industrial by-products like fly ash. The manuscript is well-structured, covers a wide range of clay types and treatment methods, and provides a valuable summary of existing literature, particularly in the areas of clay activation and mechanical properties. However, the manuscript, in its current form, has several significant scientific and structural defects that must be addressed before it can be considered for publication. The manuscript's structure and English language need to be reviewed and improved; some remarks may be significant.

1.     The authors often present data from various studies without providing a critical, synthesised conclusion, leading to confusion for the reader.

Thank you so much for your comment. We have revised the manuscript to enhance its writing quality, providing a more detailed and synthesised conclusion. We have added new information to improve clarity and draw a synthesised conclusion. For instance, in section 4, we have added the following information (Lines 632-635) & (Lines 643 to 654).

This phenomenon is consistent with typical findings in clay activation, in which thermal treatment can cause interlayer spaces to collapse and reduce the number of active sites, resulting in a less absorptive or workable material in fresh concrete mixes [64].

The difference in slump behaviour between calcined Bentonite observed by Waqas et al. [146]  and Reddy et al. [149]  can be explained by the nature and extent of calcination and its impact on Bentonite's physical and chemical properties. Moderate calcination, as reported by Waqas et al. [146], reduces surface area due to partial dehydroxylation and collapse of interlayer spaces, thereby lowering water demand and improving workability, resulting in a slight increase in slump. In contrast, intense calcination at higher temperatures (700–800°C), as studied by Reddy et al.  [149], can cause structural disruption and increased internal porosity, creating more reactive sites that absorb water more aggressively, thereby reducing slump. Furthermore, the replacement level of Bentonite and the interaction with other mix components, such as fly ash and polypropylene fibres, significantly influence the rheological behaviour of the concrete, making direct comparisons between studies complex and context-dependent.

In section 5.2, we have added the following discussion (Lines 836 to 842):

Collectively, the studies suggest that Bentonite clay can enhance the split tensile strength of concrete when used in moderate proportions (typically 15–30%). The improvements are attributed to the pozzolanic activity of silica and the filler effect. However, excessive replacement reduces strength, likely due to lower cement content and poor particle packing. Future research should focus on microstructural analysis, long-term durability, and the synergistic effects of Bentonite with other additives, such as PPF, to optimise performance.

2.     The text and Figure 4 exhibit significantly conflicting results regarding the impact of raw Bentonite on compressive strength. Research conducted by Karthikeyan [159] and Ahmad [160] indicates an increase in strength with the use of raw Bentonite, whereas Fode [118] finds a substantial decline. An adequate review should propose possible explanations for these variations. Possible factors might include.

Thank you so much for your valuable comment. We have added further discussion in lines 763 to 775, which contains a possible explanation and factors. The added discussions are as follows:

These variations in compressive strength across studies using raw and calcined Bentonite are attributable to several factors. Waqas et al. [146] documented enhancements in strength with both variants, particularly with the inclusion of PPF, likely owing to improved crack resistance and matrix densification. Conversely, Karthikeyan [193] observed inconsistent outcomes with raw Bentonite, noting a reduction in strength at a 25% replacement level and improvements at 30%, potentially due to optimal particle packing or filler effects. Ahmad et al. [194] and Fode et al. [148] demonstrated conflicting trends: Ahmad et al. [194] observed a 7.14% decrease at 20% replacement and recommended 15% as optimal, whereas Fode et al. [148] identified initial strength improvements at 5%, followed by declines, suggesting that raw Bentonite may lack pozzolanic activity unless subjected to calcination. These discrepancies may arise from differences in Bentonite source, calcination conditions, mix design, and testing methodologies, thereby highlighting the necessity for standardised evaluation procedures.

3.     What input variables are essential for developing robust predictive models for clay-GPC? How can the complexities of clay treatment be incorporated into these models?

Thank you for this suggestion. We have added a discussion in lines 971 to 983 to address this comment as follows:

 To develop reliable predictive models for clay-based geopolymer concrete (clay-GPC), it is essential to consider key input variables such as the chemical composition of the clay (including SiO₂ and Al₂O₃), the nature and treatment of the clay (whether raw or calcined), calcination temperature and duration, the type and concentration of the alkali activator, the liquid-to-solid ratio, curing conditions, and the proportion of clay substitution [216]. Research conducted by Abdullah et al. [217] and Gupta et al. [216] demonstrated that factors such as slag content, Na/Al and Si/Al ratios, and curing temperature are critical determinants of compressive strength. To effectively capture the complexities of clay treatment, models should encode treatment parameters as separate features and employ techniques such as feature selection, normalisation, and SHAP analysis to identify nonlinear interactions. Advanced models, including Random Forest, Gradient Boosting, and neural networks, have demonstrated high accuracy in predicting strength outcomes when these factors are properly incorporated [216].

4.     Include a discussion on the importance of PSD and SSA for both the treatment efficiency (e.g., how grinding affects SSA) and the final concrete properties (workability, strength).

Thank you for this suggestion. We have added a new subsection (3.4) from lines 568 to 589 to address this comment. The newly added subsection is as follows:

3.4 Importance of PSD and SSD in clay treatment and concrete performance

Particle size distribution (PSD) and specific surface area (SSA) are critical parameters influencing the efficacy of clay treatment and the performance of clay-based geopolymer concrete. Calcination during processing alters the clay's mineral structure, thereby enhancing its pozzolanic reactivity. Additionally, ball milling increases SSA by reducing particle size and dispersing agglomerates. Bai et al. [180] demonstrated that extended grinding durations yield a greater abundance of ultrafine particles, thereby enhancing pozzolanic activity and mechanical strength. Elevated calcination temperatures augment reactivity; however, they may concurrently diminish fluidity due to particle agglomeration. SSA directly affects water demand and workability; an increase in SSA enlarges the reactive surface area but necessitates additional water, which can adversely impact workability if not adequately controlled. Muzenda et al. [181] noted that SSA exerted a greater influence on early hydration and compressive strength than thmetakaolin content, with SSA values ranging from 14 to 45 m²/g depending on clay mineralogy and treatment methods. Calcination, especially at optimal temperatures, alters the mineral phase and may induce agglomeration, thereby reducing the SSA unless counteracted through grinding. Clays such as Kaolin and halloysite respond effectively to thermal activation, whereas Bentonite, illite, palygorskite, and sepiolite may benefit more from mechanical or chemical activation [182]. Again, PSD influences packing density and rheology; finer particles enhance strength but diminish fluidity, whereas a broader PSD improves workability [183]. Therefore, it is essential to adjust PSD and SSA through appropriate treatment methods to optimise the reactivity and mechanical properties of clay-GPC systems.

5.     A new section titled Durability and Microstructural Characteristics must be added. This section should synthesise available literature.

Thank you so much for your suggestion. We have added a new section, "7. Durability and Microstructural Performance of Clay-based GPC," describing the durability and microstructural performance of the clay-based GPC, in lines 1028 to 1093. This new section covers both durability and microstructural aspects: 7.1 Durability performance and 7.2 Microstructural performance.

6.     Refrain from making absolute statements without proper citations. In Section 3.3, Chemical Activation, the statement However, due to the potential for corrosion, the authors pointed out that this activation method would not be appropriate for concrete applications with embedded steel represents a significant assertion that has not been examined. Is calcium chloride left in the pore solution in a free state? What is the process behind the suggested corrosion? This requires further explanation or a supporting reference.

Thank you so much for this important suggestion. We have added the reason the author found that the chemical action of clay with CaCl2 was not suitable for steel-reinforced concrete, as mentioned in lines 561 to 563. The corresponding statement and the supporting reference have now been revised and added as follows:

However, it was observed that a high concentration of free Cl- existed in the pore solution. Therefore, due to the possibility of corrosion, the authors noted that this activation technique would not be suitable for concrete applications with embedded steel  [130].

7.     Figures 3, 5, 6, 7: The data points in these figures are very dense and overlapping, making them difficult to interpret. Consider using larger, more distinct markers or splitting the data into multiple, clearer figures (e.g., one for 1:1 clays and one for 2:1 clays).

Thank you for this suggestion. We have improved these figures by using larger, more distinct markers.