Analysis of artificially structured soil in lightly cemented kaolin

$Cc*$

the intrinsic compression index

C_c

compression index

c

cement content

c₀

threshold of soil reaction to cement content

e

void ratio

e₀

initial void ratio

$e100*$

intrinsic void ratio of reconstituted soils at corresponding effective vertical stress of 100 kPa

$e1000*$

intrinsic void ratio of reconstituted soils at corresponding effective vertical stress of 1000 kPa

I_P

plasticity index

I_v

the void index as a normalising parameter

k

the strength coefficient

q_u

unconfined compressive strength of soil

w_L

liquid limit

w_P

plastic limit

ρ_S

soil particle density

σ′_v

effective vertical stress

σ′_y

pre-consolidation pressure

Many geotechnical laboratory studies rely on reconstituted soil specimens, which often fail to replicate the behaviour of naturally structured soils. The structural changes caused by remoulding are irreversible, making it difficult to reconstruct the original mechanical characteristics. To overcome this limitation, a distinctive reconstitution method has been developed using lightly cemented clay to create artificially structured soils, enabling investigations that closely mimic the structural and mechanical features of natural clays at both macroscopic and microscopic levels (Kasama et al., 2000; Khalid et al., 2019; Lei et al., 2022). These cement-treated soils are termed artificially structured soils due to the formation of structure bonding from cementation (Wang and Korkiala-Tanttu, 2016).

Despite the practical advantages of light cementation in geotechnical engineering laboratories, the fundamental understanding of how this treatment develops the internal structure of clayey soils remains less clear. While several studies have explored and reconstructed aspects of the mechanical behaviour of naturally and artificially structured clays (Bjerrum and Rosenqvist, 1956; Kasama et al., 2000; Khalid et al., 2019; Lei et al., 2022; Meijer and Dijkstra, 2013; Pusch and Arnold, 1969), fewer have focused on using a single dominant clay mineral such as kaolin in these investigations. The use of kaolin can be highly advantageous for demonstrating soil structuration due to its consistent properties. This controlled medium enables an in-depth investigation into how reconstituted clay soils can effectively exhibit and develop key characteristics observed in structured formations, offering a clearer understanding of the underlying principles.

This study, therefore, investigates a controlled reconstitution process utilising lightly cemented kaolin to explore the development of artificially structured soil. By systematically examining the impacts of initial water content, cement content, specimen orientation, and curing times, this research aims to provide a deeper understanding of the factors governing the formation and resulting properties of artificially structured soils, which are highly suitable for practical geotechnical engineering problems and laboratory-based geotechnical research.

This study used kaolin clay, which is particularly favourable due to its limited surface activity and lower chemical reactivity with pore water compared to minerals like montmorillonite. In addition, kaolin clay is relatively stable, with significantly less pronounced shrinkage and swelling tendencies than other clay minerals. The kaolin clay used was in a dry powder state. The physical and mechanical characteristics of the kaolin clay are detailed in Table 1. Murray and Lyons (1960) established a correlation between kaolinite crystallinity and specific surface area (SSA): well-crystallised kaolinite ranges from 8 to 12 m²/g, medium from 12 to 15 m²/g, and poorly crystalline from 16 to 26 m²/g. The kaolinite used in this study has an SSA of 9.68 m²/g, indicating good crystallinity.

The reconstitution process started with preparing the kaolin–cement slurry mixture, followed by its consolidation under effective vertical pressure. The initial water content ratio of the kaolin slurry was adjusted at 1.5 and 2.0 times its liquid limit (1.5 w_L and 2.0 w_L) to evaluate the effect of initial water content. Furthermore, as noted by Tsuchida (2001) and Udaka et al. (2013), the reason behind setting the initial water content of the reconstituted kaolin at 1.5–2.0 times the liquid limit was based on the research that the water content ratio of natural clay soil during deposition on the seabed typically falls within that range.

The cement mixture was mixed with a water-to-cement ratio (W/C) of 1.0 to enhance its ease of handling and was initially stirred for about a minute to eliminate clumps. After that, kaolin slurry and cement mixtures were then mixed for an additional 10 min for homogeneity (Miura et al., 2001) using the mixer, and then transferred into the consolidation tube as shown in Figure 1. The cement used was produced by Ube-Mitsubishi Cement Co., Ltd in Japan, usually called as US-10. This cement was suitable for stabilising various soft soils, making it appropriate for use in diverse soil conditions, including sandy, cohesive, silt, organic soils, and mud. The specific physicochemical characteristics of this cement are provided in Table 2.

Consolidation took place in a tube that was 15 cm in diameter. The slurries were subjected to one-dimensional vertical pressure, called reconstituted pressure. Specifically, the samples underwent consolidation at reconstituted pressures of 50 kPa. The pressure of 50 kPa was chosen as it is representative of the low effective overburden stress found in shallow deposits. Some researchers applied reconstituted pressure based on the pre-consolidation pressure at an undisturbed state and then varied the reconstituted pressure to assess its effects (Kasama et al., 2000; Khalid et al., 2019; Tsuchida et al., 2009). This pressure was exerted incrementally over 10 days, allowing the slurries to consolidate and drain. This process simulated the gradual build-up of pressure and the simultaneous strengthening of the material through cementation and consolidation, as would occur under natural conditions. After 10 days of reconstitution pressures, the specimen from the consolidation tube was extracted and trimmed to the desired dimensions. Specimens were trimmed to their final dimensions using a fine-gauge wire saw to minimise disturbance. Unconfined compression test (UCT) specimens were prepared to 10 cm × 5 cm with a dimensional tolerance of ±1 mm, while consolidation specimens were trimmed directly into a 6 cm × 2 cm ring. The specimens were sealed with plastic wrap and kept in desiccators to prevent water loss due to evaporation. They were then left to cure at a controlled temperature of 25°C.

Determining the minimum cement content required to affect soil structure is challenging due to influencing factors like water content, curing time, and cement type. Tang et al. (2001) proposed a method using trial tests to estimate this minimum content (c₀) based on the unconfined compressive strength (q_u), expressed in Equation 1.

where k is the strength coefficient (kPa/%) and c₀ is the threshold cement content. Khalid et al. (2019) applied this approach to Shanghai marine clay. To estimate c₀ without extensive testing, a correlation with liquid limit (w_L) was used in Figure 2, based on data from six marine clays (Khalid et al., 2019; Udaka et al., 2013). It is acknowledged that due to mineralogical differences, the direct applicability of this correlation to the pure kaolin used in this study may be limited. However, in the absence of an established correlation for pure kaolin, this equation was used as a pragmatic tool to estimate a reasonable starting range for c₀. With an R² of 0.84, this equation reliably predicts c₀. In this study, c₀ was estimated using Equation 2 as 2%, and cement contents of 1%, 2%, and 3% (by dry weight) were selected. The term ‘lightly cemented’ used in this study describes a range of low cement contents sufficient to induce soil structuration, but low enough that the material’s mechanical behaviour remains analogous to that of a natural structured clay, rather than a high-strength, concrete-like composite.

The geotechnical properties of soil are primarily affected by the arrangement of its particles’ orientation. This arrangement can lead to anisotropy in the fabric, which causes the geotechnical properties to vary depending on the direction. The impact of fabric anisotropy on the constitutive characteristics of soils might be particularly significant, affecting how they respond to different stresses and strains. The specimens were cut to align with the vertical loading direction and are referred to as vertically cut. The other specimens were trimmed along the horizontal directions and are referred to as horizontally cut. The illustration of how specimens are trimmed with vertical and horizontal directions is shown in Figure 3.

Several studies have investigated the mechanical properties, focusing on the consolidation behaviour of lightly cemented clay in reconstituted soils (Chew et al., 2004; Kamruzzaman et al., 2009; Kasama et al., 2000; Khalid et al., 2019; Lei et al., 2022; Quang and Chai, 2015; Tsuchida et al., 2009; Wang et al., 2016; Wang and Korkiala-Tanttu, 2016). The compression behaviour of soft natural clays is often defined by an inverse ‘S’ curve, as shown in studies such as Butterfield (1979), Nagaraj et al. (1990), and Liu and Carter (1999). This distinctive shape is attributed to the consolidation yield stress formed during the soil's deposition and subsequent stages. If the effective vertical stress is below this consolidation yield stress, the clay exhibits low compressibility, known as the pre-yield state. However, compressibility significantly increases once the stress surpasses the consolidation yield stress, moving the soil into the post-yield state. This transformation reflects a dramatic change in how the soil compresses under increased stress.

Oedometer tests were conducted on specimens following Japanese Geotechnical Society standard 0411-2009 to thoroughly understand the compression characteristics of lightly cemented clay. These specimens were cured for the period of 21 days before testing. The consolidation test results, as shown in Figure 4, indicate a noticeable increase in the yield stress (σ′_y) with the addition of cement, even at a minimal content of 1%. This increase in the yield stress highlights the impact of cement on the compression behaviour of treated clay. The behaviour of the uncemented soils suggests that they are fully remoulded and unstructured, exhibiting characteristics of typical normally consolidated clays. Despite being subjected to a reconstituted pressure of around 50 kPa, they show no signs of structuration or cementation. The findings demonstrate that even a minimal amount of cement can lead to the formation of structured soil and how cementation contributes to soil structuration.

It can be observed that specimens with 2.0w_L exhibit a higher initial void ratio compared to those with 1.5w_L, as shown in Figure 4. This observation aligns with previous research indicating that increased water content tends to increase the initial void ratio (Hong et al., 2010; Horpibulsuk et al., 2004; Riyad et al., 2023; Wang and Korkiala-tanttu, 2016). Cerato and Lutenegger (2004) also indicated that the initial water content of specimens might significantly affect the intrinsic parameters. Another important observation is the influence of cementation: at 1.5w_L, the uncemented specimen shows a relatively low initial void ratio, while specimens with cement additions (1%–3%) exhibit a gradual increase in initial void ratio, suggesting that cementation introduces structuration and creates a more open fabric. However, at 2.0w_L, there is a significant jump in the initial void ratio from 1% to 2% cement, while the difference between 2% and 3% is minimal. This suggests that at high water content, a cement content of 1% is insufficient to stabilise the highly dispersed structure, leading to partial collapse during applied reconstituted pressure and a relatively lower initial void ratio. In contrast, at 2% and 3% cement, the amount of cementation product is adequate to bond and preserve the open fabric formed during mixing and applied reconstituted pressure, resulting in a much higher and relatively larger void structure. This behaviour indicates the presence of a critical cementation threshold beyond which further increases in cement have a limited effect on initial void ratio under high water content conditions.

Figure 5 shows the compression curves for horizontally cut specimens. When compared with Figure 4, which presents vertically cut specimens, it can be observed that the void ratios of reconstituted kaolin are generally similar between the two orientations, with only minor differences. Horizontally cut specimens consistently exhibit slightly higher final void ratios after loading compared with vertically cut specimens, but the differences are not substantial enough to indicate a significant anisotropic effect on compressibility behaviour. This suggests that anisotropy, under the tested conditions, does not strongly influence the consolidation behaviour of lightly cemented kaolin. Similarly, the compression index (C_c) increases with both initial water content and cement content in both orientations. Overall, the results imply that anisotropy has only a minor influence on the compressibility behaviour of lightly cemented kaolin. Detailed results, including comparisons of initial and final void ratios and compression index values for vertically and horizontally cut specimens after 21 days of curing, are presented in Table 3.

The minor anisotropic effects observed in this study are attributed to the interplay between a relatively low reconstitution pressure and simultaneous cement hydration. As demonstrated by Cetin (2004), low consolidation pressures such as 50 kPa are known to induce only a small degree of anisotropic particle orientation in cohesive soils. Furthermore, because the reconstitution pressure was applied gradually over 10 days, it is likely that early-forming cement bonds began to ‘lock in’ the initially random soil fabric during the early stages of sample preparation. This created a stiffer skeleton that resisted significant particle reorientation as the pressure increased to its final value. This dual mechanism, where early-stage cementation provides resistance against a low and gradually applied alignment force, is therefore the most probable cause of the minor anisotropic effects.

Burland (1990) introduced the intrinsic compression line (ICL) concept to characterise the intrinsic properties of reconstituted clays, which are inherent to the soil and independent of its natural state. The ICL is a unique compression line derived from various reconstituted clays, using the void index (I_v) as a normalising parameter. Burland (1990) further showed that a reasonably distinct ICL could be derived by normalising the data with constant values of $e100*$ and $e1000*$⁠. This normalisation is expressed mathematically in Equation 3.

where $Cc*$ is the intrinsic compression index, e represents the void ratio, $e100*$ and $e1000*$ are intrinsic void ratios of reconstituted soils at corresponding effective vertical stresses of 100 and 1000 kPa on the consolidation curve, respectively.

In the current investigation, the concept of normalisation for reconstituted soils was found to align with Burland’s ICL for specimens with yield stresses below 100 kPa. However, for soils where the yield stress (σ′_y) exceeded 100 kPa, the traditional ICL normalisation method proved unsuitable. Figures 6 and 7 present I_v-log σ′_v curves for lightly cemented kaolin, using vertically and horizontally cut specimens with initial water content of 1.5w_L and 2.0w_L. The figures illustrate variations in soil behaviour based on different cement contents of 1%, 2%, and 3%. To address the normalisation of artificially structured clays, Wang and Korkiala-Tanttu (2016) investigated the artificially structured clays, proposing novel approaches for their normalisation. Their ‘Approach I’ involves determining the void ratio at 100 kPa (⁠$e100*$⁠) by linear extrapolation of the post-yielding compression line in a log-log graph. This method proved effective for artificially structured clays, aligning well with Burland’s ICL even when the yielding stress was greater than 100 kPa. Similarly, for the artificially structured clays in this study, as shown in Figures 6 and 7, applying ‘Approach I’ has resulted in good alignment of the ICL, even for specimens where the yield stress is above 100 kPa. This confirms the suitability of this extrapolation technique for normalising the compression curves of the lightly cemented kaolin.

UCTs revealed that lightly cemented kaolin exhibits predominantly brittle behaviour, with all specimens failing at strains below 1%, as shown in Figures 8 and 9, except specimens without cement show very low strength and do not display a clear peak in the stress-strain curve. While an increase in cement content (1%–3%) enhanced peak strength and stiffness, it also intensified brittleness, manifested by abrupt post-peak stress reductions – particularly at higher cement contents (2%–3%). Specimens with 1% cement displayed semi-brittle characteristics, featuring a modest post-peak decline but still failing at very low strains, distinct from the ductile behaviour typical of uncemented clays, which fail at larger strains.

A comparison of Figures 8 and 9 highlights the significant impact of initial water content on the unconfined compressive strength (UCS) of lightly cemented kaolin. The initial water content played a critical role: specimens reconstituted at 1.5w_L demonstrated higher strength and brittleness compared to those at 2.0w_L, where excess water weakened cementation bonds slightly. Notably, even minimal cementation (1%) eliminated the ductile response of uncemented kaolin, transforming the material into a weakly bonded, brittle matrix akin to stiff clays or weakly cemented sands. These findings underscore that cementation governs the failure mode, with curing time further amplifying strength and brittleness through progressive CSH bond formation.

Figures 10 and 11 present the stress-strain curves of horizontally cut specimens under varying conditions, compared with vertical specimens in Figures 8 and 9. The results show nearly identical peak strength between orientations, with horizontally cut specimens exhibiting only slightly lower UCS values, as shown in Table 4. Both q_u and modulus elasticity (E₅₀) values increase with higher cement content and longer curing (21 against 14 days), confirming improved stiffness and strength over time. While vertical specimens consistently demonstrate marginally higher q_u and E₅₀ values than horizontal ones for equivalent conditions, the differences remain minor, indicating minimal anisotropic effects on the mechanical properties. This observation is consistent with the explanation provided in ‘Effect of anisotropy of compressibility behaviour of lightly cemented kaolin’ section: early-stage cement hydration during gradual loading limited fabric alignment, thereby reducing the potential for anisotropy to significantly influence strength and stiffness.

The kaolinite crystals are primarily pseudo-hexagonal, accompanied by plates, some larger book-like forms (booklets), and vermicular stacks (Wilson et al., 2014). In Figure 12, the uncemented kaolin shows areas where platelets are stacked in a ‘booklet-like’ formation, alongside some edge-face arrangements characteristic of flocculated clay structures. Consistent with an SSA-based classification of a well-crystallised material, the presence of these numerous stacked ‘booklets’ in the SEM images confirms the kaolin’s high crystallinity. The SEM images also reveal a complex surface structure on the basal planes, with micro-anhedral crystallites frequently adhering to larger particles and broken crystallite edges. Most particles exhibit irregular or rounded outlines, and their edge steps are often curved.

As outlined by Delage and Lefebvre (1984) and depicted schematically in Figure 13, pore spaces are categorised into intra-aggregate and inter-aggregate. Intra-aggregate pores average of kaolin ≈0.1 µm in size, representing irregular packing within soil aggregates. Larger inter-aggregate pores are formed by the stacking of these aggregates.

The observed porosity in the samples often exhibits a bimodal distribution, featuring both significant inter-aggregate and intra-aggregate porosity, a finding consistent with previous microstructural studies on kaolin (Gao et al., 2020a; Hattab and Fleureau, 2010; Wang and Xu, 2007; Yu et al., 2016; Zheng, 2023). Figure 14 provides SEM images of reconstituted samples at 1.5w_L with varying cement contents. In Figure 14(a), even with 1% cement content, the presence of ettringite is visible, alongside clearly distinguishable intra-aggregate (white rectangles) and inter-aggregate (white circles) pores. This indicates that even a minimal cement addition initiates changes in the soil’s microstructure through the formation of new mineral phases. As the cement content increases to 2% (Figure 14 (b)), ettringite crystals become more pronounced, and both pore types remain evident. At 3% cement content (Figure 14 (c)), a notable observation is the increased filling of inter-aggregate voids by ettringite. This visual evidence from the SEM images, demonstrating the progressive occupation of larger pore spaces by cement hydration products, aligns with findings from pore size distribution (PSD) analysis [mercury intrusion porosimetry (MIP)] presented in section ‘Pore size and void measurement by mercury intrusion porosimetry’, where increased cement content leads to a denser and more refined microstructure due to the filling of pore spaces by C–S–H and ettringite.

Analysing SEM images of soil structure is crucial for understanding various properties of soil, including porosity, particle size, and the arrangement of soil particles. Several researchers have conducted pore size measurements on kaolin using SEM image analysis, including Wang and Xu (2007), Hattab and Fleureau (2010), Yu et al. (2016), and Gao et al. (2020a). ImageJ, an open-source image processing programme, is widely used for this purpose due to its key advantages. It offers tools for quantitative image analysis, enabling precise measurement of soil structure characteristics such as pore size, particle shape, and distribution. In addition, it supports effective segmentation and thresholding techniques. The threshold value of pore segmentation based on the overflow criterion method by Wong et al. (2006).

In soil SEM image analysis, ImageJ measures pore size based on the brightness intensity of pixels and corresponding brightness thresholds. Dark areas are interpreted as pores, while lighter areas represent solids. During threshold adjustment, the dark regions in the SEM image are highlighted in red. After the threshold is applied, the SEM image is transformed into a binary image. The software then identifies and measures the size and shape of these black-and-white regions, providing quantitative data on pore sizes. After that, the pores were extracted into an elliptical shape, which is commonly used in pore size analysis (Gao et al., 2020a; Zheng et al., 2023). Selected images are shown in Figures 15 and 16, illustrating the process of quantifying pore sizes in soil specimens using ImageJ. Figure 15 represents a specimen with a water content of 1.5w_L, while Figure 16 corresponds to a specimen with 2.0w_L. The larger white areas observed in Figure 16 suggest increased pore sizes, highlighting the relative influence of water content on pore size.

Table 5 presents the results of pore size measurements for soil specimens at varying water and cement contents, including the average pore size area (µm²) and its corresponding percentage of the total area. The observed phenomenon where the specimens that reconstituted at 2.0w_L have larger pore sizes and a higher void ratio compared to the specimens reconstituted at 1.5w_L can be linked to the interplay between water content, soil structure, and the behaviour of cementation. When the initial water content is low, the soil particles are closer, and the particles are more likely to be closely packed, resulting in smaller pore sizes. Higher initial water content leads to more fluid and causes the soil particles to be more dispersed during mixing, leading to a structure with larger initial voids and, hence, larger pore sizes.

In addition to the PSD obtained from SEM images, the pore diameter was measured using the minimum diameter of the elliptical pore shape. This approach was chosen because it aligns with the MIP assumption of measuring pore entrance diameter, allowing for a more direct comparison. The PSD from SEM image analysis in Figure 17 reveals a smaller dominant pore diameter than the MIP results (Figure 18), which is consistent with the findings of You et al. (2017), Gao et al. (2020b), and Zheng (2023). However, it should be note that SEM-derived results are inherently limited (Zheng et al., 2023). The use of SEM 2D images restricts the viewing area and may not capture the most representative pore sizes, while the flat cross-section provides only an incomplete representation of the true 3D pore structure whereas the MIP test, by measuring the entire bulk sample, provides a more statistically robust and comprehensive distribution of the interconnected pore network.

In the present work, SEM and MIP analysis were based on a single image per variable, which may compromise the reliability of the results. For future studies, it is recommended to analyse multiple SEM images from many different regions of the specimen to minimise human error and improve accuracy. For example, Zheng and Baudet (2025) used 15–20 SEM images per variable, yielding ≈5000 pore measurements. In the author’s view, SEM images are more suitable for qualitative assessment of PSD evolution than for rigorous quantitative analysis.

Figures 18(a) and 18(b) present the PSD curves obtained from porosimetry for reconstituted samples with varying water and cement contents. The entrance pore sizes range from ≈0.01 μm to 100 μm, with dominant peaks consistently appearing around 0.3–0.7 μm across all conditions. As shown in Figure 18(a), which covers the full pore diameter range, samples prepared with a higher initial water content (2.0w_L) exhibit noticeably higher peak pore densities and slight shifts of the peaks toward the right, indicating not only a greater number of pores but also slightly larger dominant pore sizes. Figure 18(b), which is a zoomed-in view of the peak region, highlights these differences more clearly, allowing for easier comparison between samples. The results suggest that higher initial water content promotes a more dispersed and loosely packed particle arrangement, resulting in more uniformly sized and slightly larger pores. As the cement content increases from 1% to 3%, the peak intensity decreases for both different initial water content conditions, reflecting the progressive filling of pore spaces by cement hydration products such as C–S–H and ettringite. This leads to a denser and more refined microstructure. The consistently narrow PSD peaks across all samples indicate a relatively uniform pore structure, with water content mainly affecting pore quantity and size, while cement content governs the extent of structural densification and pore refinement.

These results should be interpreted with caution, as MIP primarily measures the size of pore throats; the decrease in peak intensity more accurately reflects a refinement of the pore network, where these connecting channels are progressively blocked by hydration products like C–S–H and ettringite. Furthermore, known MIP limitations such as the ink-bottle effect and pore accessibility issues, which are particularly relevant for fine-grained cemented soils, may also influence the measured peak intensities.

As shown in Table 6, the void ratio determined by MIP (e_m), which is calculated by taking the ratio of the apparent density to the bulk density and then subtracting one, is consistently lower than the total void ratio (e_o) measured by the oedometer test. This discrepancy arises because the MIP technique cannot measure inaccessible or occluded pores enclosed within particle aggregates. Analysing this difference between macro and micro analysis provides deeper insight into the complexity of the pore network formed during the structuration process. The volume of these ‘inaccessible pores’ (e_o − e_m) is substantially larger in specimens with higher initial water content, (e.g. 1.153 for 2.0 w_L compared with 0.781 for 1.5 w_L at 3% cement). This finding indicates that the cementation process within a more dispersed initial fabric creates a more complex and tortuous pore network, effectively trapping a significant volume of voids. This complex microstructure explains the macroscopic compressibility behaviour, where a higher initial water content results in a significantly higher overall initial void ratio (e_o). This linkage confirms that the macro-mechanical properties are fundamentally controlled by the engineered microstructure.

The mineral composition was analysed using X-ray diffraction (XRD) with a D/teX Ultra 250 powder diffractometers, which operated with a Cu-tube at 40 kV, current at 30 mA, and radiation with CuKα. Figure 19 displays representative XRD patterns of pure kaolin and kaolin–cement mixtures with varying cement percentages. All patterns revealed distinct kaolinite peaks, with the primary peaks at 12.3° and 24.8° of 2θ (7.21 Å and 3.59 Å, respectively) showing decreasing intensity as cement content increased, though they remained clearly visible. The relative decrease in kaolinite peak intensity indicates a progressive reduction in the crystallinity and content of kaolinite.

This observed reduction is consistent with prior research by Herzog and Mitchell (1962), Chrysochoou (2014), and Tabet et al. (2018). Furthermore, Mitchell and Jack (1966) and Choquette et al. (1987) noted that free calcium and calcium hydroxide, products of cement hydration, chemically attack kaolinite particles. This attack leads to the destruction of the kaolinite crystalline structure and the formation of amorphous CSH, which weakens kaolinite reflections at higher cement contents.

This study investigated the development of artificially structured soils through a controlled reconstitution process using lightly cemented kaolin. The results demonstrate that even a minimal cement content of 1% can significantly increase yield stress, indicating the onset of soil structuration. The initial water content plays a decisive role in determining the resulting macrostructure, with specimens prepared at 2.0w_L exhibiting higher void ratios and larger pore sizes compared with those at 1.5w_L. A critical cementation threshold was identified, where 1% cement was insufficient to stabilise the dispersed structure, while 2%–3% cement provided adequate bonding to preserve the open fabric.

Anisotropy was found to exert only a minor influence on the compressibility and mechanical behaviour. This limited effect can be attributed to the combination of low reconstitution pressure and simultaneous cement hydration, which ‘locked in’ the random particle arrangement early in the process, preventing substantial particle reorientation during subsequent loading. Mechanical testing revealed that the lightly cemented kaolin exhibited predominantly brittle behaviour, with strength and stiffness increasing with both cement content and curing time. The higher initial water content reduced strength, while lower water content led to stiffer and more brittle responses.

At the microstructural level, SEM analysis confirmed the formation of ettringite and progressive refinement of the pore network with increasing cement content. The PSD from SEM image analysis reveals a smaller dominant pore diameter than the MIP results. However, these microstructural results require careful interpretation. Future work should incorporate multiple images and samples to enhance quantitative accuracy. The XRD analysis further verified a gradual reduction in kaolinite crystallinity with increasing cement content, confirming chemical alteration by hydration products and the formation of amorphous C–S–H.

Overall, this research establishes a practical and reproducible method for fabricating artificially structured soils in the laboratory. The procedure involves preparing kaolin–cement mixtures with defined water and cement contents, consolidating them under a low reconstitution pressure, and subsequently examining their characteristics evolution through oedometer and UCT, supported by MIP, SEM, and XRD analyses. The simplicity and reproducibility of this process make it suitable for implementation in standard geotechnical laboratories. The proposed approach offers a systematic framework for producing benchmark materials and provides a valuable methodological basis for future studies aimed at standardising artificially structured soil preparation and investigating the mechanical and microstructural behaviour of structured and lightly cemented clays.

The authors sincerely express their gratitude to the Japanese Government for the financial support provided through the Japan MEXT scholarship program, which made the successful completion of this research possible. The authors also wish to thank technician of Saga University of Civil Engineering Department, Mr Akinori Saito, for his assistance with the experimental testing.