In the control of contaminated soil, cut-off walls are often used beneath existing structures or in areas where contaminated soils are widespread, created by mixing soil in situ with cement. These cut-off walls are designed to limit the spread of contaminants into the groundwater and therefore require low permeability. Recently, the permeability of cement-treated soil subjected to the flow of a magnesium solution was investigated. As a result, it was shown that the permeability decreased significantly with increasing flow rate of the magnesium solution, suggesting that magnesium salts contribute to the reduction of permeability in cement-treated soil. In this study, cement-treated soil with added magnesium sulfate was investigated to develop a low-permeability material. As a result, it was confirmed that the permeability of the cement-treated soil decreased as the magnesium sulfate content increased. Instrumental analysis showed that the amount of ettringite formation increased with higher magnesium sulfate content. The reduction in the permeability of the specimens with magnesium sulfate was attributed to the filling of the pores and the increased complexity of the pore structure caused by the formation of ettringite.
Notation
- CCaO
calcium oxide content in ordinary Portland cement
- dp
pore size diameter
- i
hydraulic gradient
- k
permeability coefficient
- M/C
molar ratio of magnesium sulfate to calcium oxide in the cement
- mCaO
mass of calcium oxide contained in ordinary Portland cement
- mMgSO4·7H2O
mass of magnesium sulfate heptahydrate
- mOPC
mass of ordinary Portland cement
- NCaO
the number of moles of calcium oxide
- NCaO_M/C
the number of moles of NCaO (mol), which corresponds to the M/C (%)
- nCaO
molar mass of calcium oxide
- nMgSO4·7H2O
molar mass of magnesium sulfate heptahydrate
- Q
flow rate
- qu
unconfined compressive strength
- Vdp
cumulative pore volume
- ε
axial strain
- σ
unconfined compressive stress
Introduction
When contaminated soil is present over a large area directly beneath an existing structure, cut-off walls are often used to limit the diffusion of contaminants into the groundwater (Devarangadi et al., 2022; Shu et al., 2022). One of the most important factors in the construction of cut-off walls is permeability. To improve impermeability, a soil–cement made by mixing local soil with cement, is usually used as a cut-off wall. In addition, bentonite-based cut-off walls such as soil–bentonite and cement–bentonite are also commonly used (Carreto et al., 2016; Yang et al., 2017; Zhou et al., 2023). For impermeable soil–cement walls, the permeability is reduced by the formation of cement hydrates that fill the voids in the soil. Generally, the coefficient of permeability used in the development of cut-off walls varies depending on the application; in Japan, it is typically either below 1.0 10−8 m/s or 1.0 10−9 m/s (Hioki et al., 2013). This standard is often not met with a small amount of cement, so large amounts of cement are often added. Even if the cement content is significantly increased from 100 to 300 kg/m3, the permeability coefficient decreases only by one order of magnitude (Japan Cement Association, 2021). Therefore, depending on the soil type, the performance requirements of a cut-off wall may not be met.
Various types of cement-based solidification materials have been developed by adding specific components to ordinary Portland cement (OPC) to provide various functions to the soil. However, while a considerable number of studies have focused on applications such as improving soft soils (Niu et al., 2024; Yaghoubi et al., 2018) and reducing the leaching of hazardous substances such as hexavalent chromium (Erdem et al., 2016; Trezza and Ferraiuelo, 2003), very few studies have focused specifically on cement-based solidification materials developed explicitly to reduce permeability.
Recently, the permeability of cement-treated sandy soil subjected to the flow of a mixed aqueous solution of magnesium sulfate and magnesium chloride was investigated. The results showed that permeability decreased significantly with increasing flow rate, suggesting that magnesium salts contribute to the reduction of permeability (Hara et al., 2022). Sulfates are known to cause deterioration in cement-based materials such as soil cement and concrete. Excessive supply of sulfates from the ground or external environment can cause deterioration of cementitious materials, including expansion, cracking, and strength loss, and the permeability of soil cement may also increase (Jin et al., 2019; Verástegui-Flores and Di Emidio, 2014). On the contrary, small amounts of sulfate may reduce the permeability of soil cement; however, the reduction is minor, and no significant changes have been observed (Asghari et al., 2014; Jha and Sivapullaiah, 2016). These studies examined the improvement effects on soils containing sulfates and the sulfate resistance of soil cement. The objective of this study was to develop a low-permeability material by mixing magnesium sulfate with cement. To this end, the changes in the permeability of cement-treated soil with magnesium sulfate were investigated. In addition, field emission-scanning electron microscope (FE-SEM), mercury intrusion porosimeter, and thermogravimetric-differential thermal analysis (TG-DTA) were performed to discuss the mechanism of the change in permeability.
Sample preparation
DL clay was used as the soil sample in this study. The physical properties of the DL clay are listed in Table 1. The X-ray diffraction (XRD) pattern of the soil sample is also shown in Figure 1. The DL clay is mainly composed of quartz as the main mineral. OPC was used as the cementitious material, and its main chemical components and physical properties are shown in Table 2. The water content of the DL clay was adjusted to 44.8%. The amount of cement added was set at 70 kg/m3. Magnesium sulfate heptahydrate was selected as the magnesium-based reagent used in this study. The molar ratio of magnesium sulfate to calcium oxide in the added cement was defined as M/C, and magnesium sulfate heptahydrate was added to achieve an M/C = 0%, 10%, 20%, 30%, and 50%. The mass of magnesium sulfate heptahydrate to be added is obtained from the following equations.
where mCaO (g) is the mass of calcium oxide contained in OPC, mOPC (g) is the mass of OPC, CCaO (%) is the percentage of calcium oxide in OPC, NCaO (mol) is the number of moles of calcium oxide, nCaO (g/mol) is the molar mass of calcium oxide, NCaO_M/C (mol) is the number of moles of NCaO (mol), which corresponds to the M/C (%), mMgSO4·7H2O (g) is the mass of the magnesium sulfate heptahydrate to be mixed, and nMgSO4·7H2O (g/mol) is the molar mass of the magnesium sulfate heptahydrate.
Physical properties of DL clay
| Property | Value |
|---|---|
| Soil particle density: g/cm3 | 2.645 |
| Liquid limit: % | NP |
| Plastic limit: % | NP |
| Particle size distribution: % | |
| Gravel | 0.0 |
| Sand | 0.7 |
| Silt | 92.8 |
| Clay | 6.5 |
A graph of x-ray diffraction pattern depicts intensity in arbitrary units on the y-axis against 2 theta degree on the x-axis ranging from 5 to 70. The pattern shows multiple sharp peaks labelled with letters Q and P. Q represents Quartz and P represents Pyrophyllite, as stated in the legend. A very tall Quartz peak occurs near 26 degree, indicating the highest intensity. Smaller Quartz peaks appear near 20 degree, 50 degree, 60 degree, and close to 70 degree. Pyrophyllite peaks appear at lower intensity near 10 degree and around 28 degree. The baseline shows minor fluctuations with low intensity signals between major peaks, illustrating a Quartz dominated mineral composition with minor Pyrophyllite presence.X-ray diffraction (XRD) pattern of DL clay
A graph of x-ray diffraction pattern depicts intensity in arbitrary units on the y-axis against 2 theta degree on the x-axis ranging from 5 to 70. The pattern shows multiple sharp peaks labelled with letters Q and P. Q represents Quartz and P represents Pyrophyllite, as stated in the legend. A very tall Quartz peak occurs near 26 degree, indicating the highest intensity. Smaller Quartz peaks appear near 20 degree, 50 degree, 60 degree, and close to 70 degree. Pyrophyllite peaks appear at lower intensity near 10 degree and around 28 degree. The baseline shows minor fluctuations with low intensity signals between major peaks, illustrating a Quartz dominated mineral composition with minor Pyrophyllite presence.X-ray diffraction (XRD) pattern of DL clay
Main chemical components and physical properties of ordinary Portland cement (OPC)
| Classification | Component/property | Value |
|---|---|---|
| Chemical components | CaO: % | 64.48 |
| SiO2: % | 20.20 | |
| Fe2O3: % | 2.94 | |
| Al2O3: % | 5.50 | |
| SO3: % | 1.86 | |
| MgO: % | 0.92 | |
| Physical properties | Density: g/m3 | 3.16 |
| Specific surface area: cm2/g | 3,420 |
The specimens for the permeability tests were prepared according to the following procedure. First, cement and a magnesium reagent were added to the soil samples and mixed using an electric mixer for 10 min. The mixture was then placed in plastic moulds with a diameter of 100 mm and a height of 200 mm and filled to a height of ∼40 mm. The test specimens were cured in the laboratory at 20°C for 26 days. After curing, they were trimmed to a height of 30 mm. Before performing the permeability test, the specimen was immersed in water and cured for 2 days under a vacuum pressure of 98 kPa to saturate the specimen. The total curing time for the specimens was therefore 28 days. For the mechanical tests and instrumental analyses, the test specimens were prepared in a plastic mould with a diameter of 50 mm and a height of 100 mm. The test specimens were slightly overfilled to prevent surface drying and cured in the laboratory at 20°C for 28 days. After curing, the overfilled portion was trimmed off, and the specimens were demoulded from the plastic mould.
Methods
Permeability test
In general, the coefficient of permeability of cohesive soils is extremely low. When permeation is performed using rigid-wall columns, preferential flow paths tend to form along the side walls, making it difficult to accurately assess the permeability of the soil. Therefore, previous studies have been conducted using permeameters with flexible walls where a confining pressure is applied across a membrane to suppress the formation of such flow paths (Shaker et al., 2022; Zhu et al., 2023). Accordingly, permeability tests were performed in the present study, using a flexible-wall column permeameter. A schematic representation of the permeameter used in this study is shown in Figure 2. To maintain a constant hydraulic head during the test, a Mariotte tube was used in the water supply tank.
The schematic diagram depicts a constant head permeability test apparatus illustrating the flow of water through a soil specimen under controlled hydraulic conditions. The specimen is positioned horizontally inside a rigid cell and enclosed by a rubber membrane, with porous stone plates placed on both sides to allow water passage. Water enters and exits the specimen through connected tubing, establishing a hydraulic head indicated between an inlet water container and the specimen level. On the right side, a water reservoir is connected to a Marriott tube, which maintains a constant water level. An air compressor supplies pressure through an air regulator and gauge to control the system. Labels identify Hydraulic head, Rubber membrane, Specimen, Porous stone, Marriott tube, Water, Air regulator, and Air compressor, illustrating the components and their functional connections.Schematic representation of the flexible wall permeameter used in this study
The schematic diagram depicts a constant head permeability test apparatus illustrating the flow of water through a soil specimen under controlled hydraulic conditions. The specimen is positioned horizontally inside a rigid cell and enclosed by a rubber membrane, with porous stone plates placed on both sides to allow water passage. Water enters and exits the specimen through connected tubing, establishing a hydraulic head indicated between an inlet water container and the specimen level. On the right side, a water reservoir is connected to a Marriott tube, which maintains a constant water level. An air compressor supplies pressure through an air regulator and gauge to control the system. Labels identify Hydraulic head, Rubber membrane, Specimen, Porous stone, Marriott tube, Water, Air regulator, and Air compressor, illustrating the components and their functional connections.Schematic representation of the flexible wall permeameter used in this study
Permeability tests were conducted with the hydraulic gradient set to 25. The hydraulic gradient varied from 5 to 25 only in the case of M/C = 0% to verify the validity of Darcy’s law. If the water pressure during permeation exceeds the confining pressure, leakage may occur through the gap between the membrane and the test specimen. Under this experimental condition, the maximum hydraulic head difference was 75 cm, which corresponds to a hydraulic pressure of ∼7.4 kPa. The confining pressure was therefore set at 30 kPa to prevent leakage.
The experimental procedure was as follows. The degassed porous stone was installed in the permeability test apparatus and flushed with water to remove the air. Specimens were then placed in the apparatus. A filter paper and a urethane filter were inserted between the porous stone and the specimen. The specimen was then covered with a grease-coated membrane and secured in place using an O-ring. These procedures were carried out under water in a large tank to ensure that no air entered the specimen or the apparatus, thus maintaining a saturated condition. A confining pressure of 30 kPa was then applied using the air compressor, and the water supply valve was opened to start the water flow. For each condition, three measurements were performed.
Analysis with the field emission-scanning electron microscope
As a pre-treatment, each sample was subjected to vacuum freeze-drying. The samples were freeze-dried in their coarsely crushed state for 24 h in a vacuum. The specimens for electron microscopy observation were shaped into a 3–5 mm square. SEM images were recorded with the JSM-7600F (JEOL).
Unconfined compression test
For each condition, three specimens were used for the unconfined compression test. The tests were performed in accordance with JIS A 1216:2020. To ensure uniform load application during testing, the top and bottom ends of the specimens were capped with gypsum. Three specimens were prepared for each set of conditions.
Mercury intrusion porosimeter
For the application of a mercury intrusion porosimeter, it is necessary to remove all fluids from the pores while minimising the structural changes of the specimen during pre-treatment. Oven drying has been confirmed to affect the pore structure, whereas freeze-drying has been suggested as a suitable method for studies of the pore structure of cement-based materials using the mercury intrusion porosimeter (Gallé, 2001). Therefore, vacuum freeze-drying was used as the drying method in the present study. Samples larger than 5 mm were removed from the specimens and carefully placed in a vacuum freeze dryer. The drying process was carried out for ∼24 h.
Thermogravimetric-differential thermal analysis
TG-DTA was performed with the Thermo Plus EV02 (Rigaku). The sample was oven-dried at 110°C for 24 h and then ground using a mortar and pestle. The measurement was performed in a temperature range of 110°C–800°C, with a heating rate of 20°C/min and an air flow of 300 ml/min. The powdered sample was stored in a desiccator at 20% relative humidity until analysis.
Experimental results
Permeability test
The relationship between the coefficient of permeability and the M/C is shown in Figure 3. The permeability coefficients of the cement-treated soil specimens with M/C = 0%, 10%, 20%, 30%, and 50% were k = 6.91 10−8, 6.08 10−8, 1.83 10−8, 1.21 10−8, and 6.18 10−9 m/s, respectively; the permeability coefficient decreased with the addition of magnesium sulfate. The required permeability coefficient of less than 1.0 10−8 m/s for the cut-off wall was achieved under the condition of M/C = 50% in this study. However, it should be noted that the addition of higher amounts of magnesium sulfate may lead to increased permeability due to the excessive formation of ettringite.
The graph depicts permeability coefficient k in metres per second on the y-axis plotted on a logarithmic scale from 1.0 times 10 to the power of negative 9 to 1.0 times 10 to the power of negative 7, against M C percent on the x-axis ranging from 0 to 50. Discrete data points connected by straight lines show permeability decreasing as M C increases. At M C equals 0 percent and 10 percent, permeability lies close to the upper range near 1.0 times 10 to the power of negative 7 metres per second. A sharp reduction occurs between 10 percent and 20 percent, followed by a gradual decrease toward 50 percent. A dashed horizontal reference line marks 1.0 times 10 to the power of negative 8 metres per second, which the plotted trend crosses between 30 percent and 50 percent.Relationship between the coefficient of permeability and the M/C
The graph depicts permeability coefficient k in metres per second on the y-axis plotted on a logarithmic scale from 1.0 times 10 to the power of negative 9 to 1.0 times 10 to the power of negative 7, against M C percent on the x-axis ranging from 0 to 50. Discrete data points connected by straight lines show permeability decreasing as M C increases. At M C equals 0 percent and 10 percent, permeability lies close to the upper range near 1.0 times 10 to the power of negative 7 metres per second. A sharp reduction occurs between 10 percent and 20 percent, followed by a gradual decrease toward 50 percent. A dashed horizontal reference line marks 1.0 times 10 to the power of negative 8 metres per second, which the plotted trend crosses between 30 percent and 50 percent.Relationship between the coefficient of permeability and the M/C
The relationship between flow rate and hydraulic gradient for the M/C = 0% is shown in Figure 4. A high correlation was confirmed between the flow rate and hydraulic gradient, indicating that Darcy’s law is applicable. Darcy’s law requires laminar flow; however, as flow velocity increases, the seepage flow in soil may transition to turbulent flow. Generally, soil materials with lower permeability have higher flow velocities at which they transition to non-Darcy flow. Therefore, since Darcy’s law was confirmed to hold up to a hydraulic gradient of 25 for the specimen with the highest permeability (M/C = 0%), it can be concluded that Darcy’s law also applies to specimens with M/C = 10% or higher.
The graph depicts flow rate Q in cubic metres per second on the y-axis, ranging from 0 to 1.5 times 10 to the power of negative 8, plotted against hydraulic gradient i on the x-axis, ranging from 0 to 25. Four data points at hydraulic gradient values of 5, 10, 20, and 25 show a steady linear increase in flow rate. The corresponding flow rates rise from about 2.0 times 10 to the power of negative 9 at 5 to about 1.3 times 10 to the power of negative 8 at 25. The data points are connected by a straight line, and a label near the line indicates a coefficient of determination R-squared equals 0.999, showing an almost perfect linear relationship between flow rate and hydraulic gradient.Relationship between flow rate and hydraulic gradient for the M/C = 0%
The graph depicts flow rate Q in cubic metres per second on the y-axis, ranging from 0 to 1.5 times 10 to the power of negative 8, plotted against hydraulic gradient i on the x-axis, ranging from 0 to 25. Four data points at hydraulic gradient values of 5, 10, 20, and 25 show a steady linear increase in flow rate. The corresponding flow rates rise from about 2.0 times 10 to the power of negative 9 at 5 to about 1.3 times 10 to the power of negative 8 at 25. The data points are connected by a straight line, and a label near the line indicates a coefficient of determination R-squared equals 0.999, showing an almost perfect linear relationship between flow rate and hydraulic gradient.Relationship between flow rate and hydraulic gradient for the M/C = 0%
Field emission-scanning electron microscope analysis
Figure 5 presents the FE-SEM images at 5000 magnification. Needle-like crystals were observed under all conditions, which are believed to be ettringite. With increasing M/C, a larger amount of ettringite was formed, and in the case of M/C = 50%, the ettringite crystals appeared larger. It was found that the expansion due to ettringite formation in cement-treated specimens increases exponentially with the sulfate content (Knopp and Moormann, 2016). In addition, a phase was observed at M/C = 0%, which was presumably calcium silicate hydrate; however, it was no longer recognisable when the M/C was increased to 50%.
The multi-panel scanning electron micrograph depicts microstructural features of cement-treated material at different M C contents labelled a to e. Panel a shows M C equals 0 percent with visible calcium silicate hydrate and needle-shaped ettringite distributed between particles. Panel b shows M C equals 10 percent with a denser surface and fewer distinct needle features. Panel c shows M C equals 20 percent with an interconnected network of elongated ettringite crystals. Panel d shows M C equals 30 percent with compacted hydration products filling pore spaces. Panel e shows M C equals 50 percent with abundant plate-like and needle-shaped crystals forming a dense and complex structure. All panels are recorded at similar magnification with a scale bar of 1 micrometre, illustrating progressive microstructural densification with increasing M C content.FE-SEM images at 5000 magnification
The multi-panel scanning electron micrograph depicts microstructural features of cement-treated material at different M C contents labelled a to e. Panel a shows M C equals 0 percent with visible calcium silicate hydrate and needle-shaped ettringite distributed between particles. Panel b shows M C equals 10 percent with a denser surface and fewer distinct needle features. Panel c shows M C equals 20 percent with an interconnected network of elongated ettringite crystals. Panel d shows M C equals 30 percent with compacted hydration products filling pore spaces. Panel e shows M C equals 50 percent with abundant plate-like and needle-shaped crystals forming a dense and complex structure. All panels are recorded at similar magnification with a scale bar of 1 micrometre, illustrating progressive microstructural densification with increasing M C content.FE-SEM images at 5000 magnification
Unconfined compression test
Figure 6 shows the stress–strain curves for each M/C. It was found that cement-treated soil specimens with added magnesium sulfate showed higher stress values than the M/C = 0% specimens, beyond an axial strain of 3.0%. Figure 7 illustrates the relationship between the unconfined compressive strength and the M/C. The results of this experiment showed that even when the M/C value was increased, no significant change in the unconfined compressive strength was observed. Figure 8 also shows the relationship between failure strain and the M/C. A trend of increasing failure strain was observed with higher M/C. A series of direct tensile tests on fibre-reinforced soils demonstrated that failure strain increased with higher fibre content (Divya et al., 2014). Accordingly, it is suggested that ettringite with a high aspect ratio played a fibre-reinforcing role within the cement-treated soil and increased the failure strain.
The graph depicts unconfined compressive stress sigma in kilonewtons per square metre on the y-axis plotted against axial strain epsilon in percent on the x-axis, ranging from 0 to 8. Five curves represent M C equals 0 percent, 10 percent, 20 percent, 30 percent, and 50 percent as indicated in the legend. All curves rise steeply at low strain, reach a peak stress, and then decrease with increasing strain. The highest peak stress of about 250 kilonewtons per square metre occurs for M C equals 30 percent at around 3 percent strain. Lower peak stresses are observed for M C equals 0 percent, 10 percent, 20 percent, and 50 percent. After the peak, all curves show progressive stress reduction, illustrating strain softening behaviour with varying M C content.Stress–strain curves for each M/C
The graph depicts unconfined compressive stress sigma in kilonewtons per square metre on the y-axis plotted against axial strain epsilon in percent on the x-axis, ranging from 0 to 8. Five curves represent M C equals 0 percent, 10 percent, 20 percent, 30 percent, and 50 percent as indicated in the legend. All curves rise steeply at low strain, reach a peak stress, and then decrease with increasing strain. The highest peak stress of about 250 kilonewtons per square metre occurs for M C equals 30 percent at around 3 percent strain. Lower peak stresses are observed for M C equals 0 percent, 10 percent, 20 percent, and 50 percent. After the peak, all curves show progressive stress reduction, illustrating strain softening behaviour with varying M C content.Stress–strain curves for each M/C
The graph depicts unconfined compressive strength q u in kilonewtons per square metre on the y-axis plotted against M C percent on the x-axis, ranging from 0 to 50. Data points with vertical error bars show the variation in measured strength. The strength is about 215 kilonewtons per square metre at 0 percent, decreases to around 195 kilonewtons per square metre at 10 percent, increases to about 225 kilonewtons per square metre at 20 percent, and reaches a maximum of approximately 255 kilonewtons per square metre at 30 percent. At 50 percent, the strength decreases to around 210 kilonewtons per square metre, indicating an optimum M C content near 30 percent for maximum unconfined compressive strength.Relationship between unconfined compressive strength and the M/C
The graph depicts unconfined compressive strength q u in kilonewtons per square metre on the y-axis plotted against M C percent on the x-axis, ranging from 0 to 50. Data points with vertical error bars show the variation in measured strength. The strength is about 215 kilonewtons per square metre at 0 percent, decreases to around 195 kilonewtons per square metre at 10 percent, increases to about 225 kilonewtons per square metre at 20 percent, and reaches a maximum of approximately 255 kilonewtons per square metre at 30 percent. At 50 percent, the strength decreases to around 210 kilonewtons per square metre, indicating an optimum M C content near 30 percent for maximum unconfined compressive strength.Relationship between unconfined compressive strength and the M/C
The graph depicts failure strain epsilon in percent on the y-axis plotted against M C percent on the x-axis, ranging from 0 to 50. Data points with vertical error bars show the variation in failure strain. The failure strain is about 2.2 percent at 0 percent, increases to around 3.3 percent at 10 percent, and reaches a maximum of approximately 4.1 percent at 20 percent. The strain then decreases to about 3.3 percent at 30 percent and further reduces to around 3.0 percent at 50 percent. The trend indicates increasing deformability up to 20 percent M C, followed by a gradual reduction at higher M C levels.Relationship between failure strain and the M/C
The graph depicts failure strain epsilon in percent on the y-axis plotted against M C percent on the x-axis, ranging from 0 to 50. Data points with vertical error bars show the variation in failure strain. The failure strain is about 2.2 percent at 0 percent, increases to around 3.3 percent at 10 percent, and reaches a maximum of approximately 4.1 percent at 20 percent. The strain then decreases to about 3.3 percent at 30 percent and further reduces to around 3.0 percent at 50 percent. The trend indicates increasing deformability up to 20 percent M C, followed by a gradual reduction at higher M C levels.Relationship between failure strain and the M/C
Mercury intrusion porosimeter
Figure 9 shows the cumulative pore volume curves of the cement-treated soil specimens. No significant difference in total pore volume was observed when M/C was varied. Figure 10 presents the pore size distribution of the cement-treated soil specimens. Sharp peaks were identified at 3.1, 2.5, 2.1, 1.7, and 1.9 μm for M/C = 0%, 10%, 20%, 30%, and 50%, respectively, indicating a general trend of decreasing peak pore diameter with increasing M/C. According to the proposed classification, pores with diameters of 0.01–1 μm are defined as micropores, 1–10 μm as mesopores, and those larger than 10 μm as macropores (Kamon, 1977). Figure 11 illustrates the proportion of each pore size category for the cement-treated soil specimens with varying M/C. From M/C = 0%–30%, the proportion of micropores in the total pore volume increased from 28% to 47%. However, at M/C = 50%, the micropore ratio decreased to 35%. Therefore, it was difficult to evaluate the permeability of cement-treated soil with added magnesium sulfate based solely on the total pore volume and pore size distribution.
The graph depicts cumulative pore volume in millilitres per gram on the y-axis plotted against pore size diameter d p in micrometres on the x-axis shown on a logarithmic scale from about 0.005 to 100. Five curves represent M C equals 0 percent, 10 percent, 20 percent, 30 percent, and 50 percent as indicated in the legend. All curves begin at higher cumulative pore volume values at small pore sizes and decrease gradually before a sharp drop occurs around 1 micrometre. The curve for M C equals 0 percent maintains a higher cumulative pore volume at larger pore sizes compared with higher M C levels. Increasing M C shifts the curves downward and slightly leftward, indicating a reduction in total pore volume and changes in pore size distribution as M C increases.Cumulative pore volume curves of the cement-treated soil specimens
The graph depicts cumulative pore volume in millilitres per gram on the y-axis plotted against pore size diameter d p in micrometres on the x-axis shown on a logarithmic scale from about 0.005 to 100. Five curves represent M C equals 0 percent, 10 percent, 20 percent, 30 percent, and 50 percent as indicated in the legend. All curves begin at higher cumulative pore volume values at small pore sizes and decrease gradually before a sharp drop occurs around 1 micrometre. The curve for M C equals 0 percent maintains a higher cumulative pore volume at larger pore sizes compared with higher M C levels. Increasing M C shifts the curves downward and slightly leftward, indicating a reduction in total pore volume and changes in pore size distribution as M C increases.Cumulative pore volume curves of the cement-treated soil specimens
The graph illustrates pore size distribution with d V divided by d log d p in millilitres per gram on the y-axis and pore size diameter d p in micrometres on the x-axis plotted on a logarithmic scale from about 0.01 to 100. Five curves represent M C equals 0 percent, 10 percent, 20 percent, 30 percent, and 50 percent as listed in the legend. All curves gradually increase from low pore sizes and rise sharply to a main peak near 1 to 2 micrometres. The highest and sharpest peak appears for M C equals 0 percent, while increasing M C broadens the peak and slightly shifts its position. After the main peak, all curves rapidly decrease and remain close to zero at larger pore sizes.Pore size distribution for the cement-treated soil specimens
The graph illustrates pore size distribution with d V divided by d log d p in millilitres per gram on the y-axis and pore size diameter d p in micrometres on the x-axis plotted on a logarithmic scale from about 0.01 to 100. Five curves represent M C equals 0 percent, 10 percent, 20 percent, 30 percent, and 50 percent as listed in the legend. All curves gradually increase from low pore sizes and rise sharply to a main peak near 1 to 2 micrometres. The highest and sharpest peak appears for M C equals 0 percent, while increasing M C broadens the peak and slightly shifts its position. After the main peak, all curves rapidly decrease and remain close to zero at larger pore sizes.Pore size distribution for the cement-treated soil specimens
The graph depicts the percentage contribution of Micropore, Mesopore, and Macropore across M C values on the y-axis labelled M C colon percent, with horizontal bars extending from 0 to 100 percent on the x-axis. For M C equals 0 percent, Micropore contributes 28 percent, Mesopore 68 percent, and Macropore 4 percent. For M C equals 10 percent, Micropore is 42 percent, Mesopore is 54 percent, and Macropore is 4 percent. For M C equals 20 percent, Micropore is 45 percent, Mesopore is 53 percent, and Macropore is 3 percent. For M C equals 30 percent, Micropore is 47 percent, Mesopore is 49 percent, and Macropore is 4 percent. For M C equals 50 percent, Micropore is 35 percent, Mesopore is 61 percent, and Macropore is 5 percent.Proportion of each pore size category for the cement-treated soil specimens
The graph depicts the percentage contribution of Micropore, Mesopore, and Macropore across M C values on the y-axis labelled M C colon percent, with horizontal bars extending from 0 to 100 percent on the x-axis. For M C equals 0 percent, Micropore contributes 28 percent, Mesopore 68 percent, and Macropore 4 percent. For M C equals 10 percent, Micropore is 42 percent, Mesopore is 54 percent, and Macropore is 4 percent. For M C equals 20 percent, Micropore is 45 percent, Mesopore is 53 percent, and Macropore is 3 percent. For M C equals 30 percent, Micropore is 47 percent, Mesopore is 49 percent, and Macropore is 4 percent. For M C equals 50 percent, Micropore is 35 percent, Mesopore is 61 percent, and Macropore is 5 percent.Proportion of each pore size category for the cement-treated soil specimens
Thermogravimetric-differential thermal analysis
Figure 12 shows thermogravimetric (TG) curves and differential thermal analysis (DTA) curves. The TG curves showed a trend of decreasing mass loss with increasing M/C. As the samples used for the analysis were oven-dried at 110°C during the pre-treatment process, any mass loss below this temperature range was not considered. Ettringite is known to decompose at temperatures between 70°C and 100°C (Alonso and Fernandez, 2004; Annerel and Taerwe, 2009; Castellote et al., 2004), and it is likely that ettringite had almost completely disappeared by the time of pre-treatment. In other words, since the mass loss of ettringite was not measured, it can be inferred that the amount of mass loss decreases with increasing M/C.
The two-panel graph depicts thermogravimetric and differential thermal analysis results plotted against temperature in degrees Celsius on the x-axis, ranging from 100 to 800. The upper panel labelled T G shows mass reduction ratio in percent on the y-axis from 96 to 100 for D L clay and samples with M C equals 0 percent, 10 percent, 20 percent, 30 percent, and 50 percent, all showing gradual mass loss with increasing temperature, with D L clay retaining the highest mass throughout. The lower panel labelled D T A shows heat flow in microvolts on the y-axis from 0 to 60, with all curves increasing to a broad maximum and showing two marked thermal events highlighted by dashed vertical lines at 455.0 degree Celsius and 575.0 degree Celsius, where small inflections occur for all M C levels.Thermogravimetric (TG) curves and differential thermal analysis (DTA) curves
The two-panel graph depicts thermogravimetric and differential thermal analysis results plotted against temperature in degrees Celsius on the x-axis, ranging from 100 to 800. The upper panel labelled T G shows mass reduction ratio in percent on the y-axis from 96 to 100 for D L clay and samples with M C equals 0 percent, 10 percent, 20 percent, 30 percent, and 50 percent, all showing gradual mass loss with increasing temperature, with D L clay retaining the highest mass throughout. The lower panel labelled D T A shows heat flow in microvolts on the y-axis from 0 to 60, with all curves increasing to a broad maximum and showing two marked thermal events highlighted by dashed vertical lines at 455.0 degree Celsius and 575.0 degree Celsius, where small inflections occur for all M C levels.Thermogravimetric (TG) curves and differential thermal analysis (DTA) curves
Regarding the DTA curves, a peak at ∼450°C–460°C was observed for the sample with M/C = 0%, and a small peak at a similar temperature range was also noted for M/C = 10%. It has been reported that the thermal decomposition peak of calcium hydroxide occurs between 465°C and 469°C (Vedalakshmi et al., 2003). Therefore, it is assumed that the peaks observed at 450°C–460°C in the M/C = 0% and M/C = 10% specimens are due to the presence of calcium hydroxide. However, for M/C = 20%, 30%, and 50%, no peak was detected in this temperature range. This indicates that calcium was consumed during the formation of ettringite, leading to a decrease in the production of calcium hydroxide. In the DTA curve of DL clay, a peak was observed around 575°C; the peak was at the same location under all conditions. As shown in Figure 1, DL clay is mainly composed of quartz, and a peak corresponding to the crystal transformation of quartz is reported to occur near 575°C (Klimesch and Ray, 1997; Sepulcre-Aguilar and Hernández-Olivares, 2010). It is therefore assumed that this peak is due to the crystal transformation of quartz.
Mechanism of permeability reduction in cement-treated soil with the addition of magnesium sulfate
The results of this study showed that the permeability of cement-treated soil decreased with the addition of magnesium sulfate. It was inferred that the amount of calcium silicate hydrate decreased with increasing M/C, and it was confirmed that the formation of ettringite increased. Figure 13 illustrates a conceptual image of the changes in the pore structure of cement-treated soil due to the addition of magnesium sulfate. As shown in this figure, the decrease in permeability can be attributed not only to pore filling by ettringite but also to the fact that the ettringite formed in an interlocked manner within the pores, resulting in a more complex pore structure that likely impeded the flow of water through the soil.
The two-panel schematic depicts cement treated soil structure comparing the Magnesium sulphate-free and Magnesium sulphate-added conditions. The left panel shows soil particles with pore spaces partially filled by cement hydrates, including Calcium Silicate Hydrate, written as C-S-H, and ettringite, allowing limited water flow through connected pores. Labels identify Ettringite, Soil particle, C-S-H, and Water flow. The right panel shows Magnesium sulphate added, where a larger amount of ettringite forms between soil particles, the amount of C-S-H decreases, and the pore network becomes more complex. Water flow paths are more obstructed, indicating reduced permeability due to increased ettringite formation and altered pore structure.Conceptual image of the pore structure in cement-treated soil with added magnesium sulfate
The two-panel schematic depicts cement treated soil structure comparing the Magnesium sulphate-free and Magnesium sulphate-added conditions. The left panel shows soil particles with pore spaces partially filled by cement hydrates, including Calcium Silicate Hydrate, written as C-S-H, and ettringite, allowing limited water flow through connected pores. Labels identify Ettringite, Soil particle, C-S-H, and Water flow. The right panel shows Magnesium sulphate added, where a larger amount of ettringite forms between soil particles, the amount of C-S-H decreases, and the pore network becomes more complex. Water flow paths are more obstructed, indicating reduced permeability due to increased ettringite formation and altered pore structure.Conceptual image of the pore structure in cement-treated soil with added magnesium sulfate
Conclusions
In this study, cement-treated soil specimens were prepared with varying amounts of added magnesium sulfate. Permeability tests, FE-SEM analysis, unconfined compression tests, mercury intrusion porosimeter, and TG-DTA were performed to investigate the effects of magnesium salts on permeability and the underlying mechanisms. The results are as follows.
The permeability of the cement-treated soil decreased as the amount of added magnesium sulfate increased.
FE-SEM analysis revealed a trend towards increased ettringite formation as the M/C increased.
The results of the mercury intrusion porosimeter showed no significant change in the total pore volume in the cumulative pore volume curve. However, a trend was observed in the pore size distribution where the pore diameter corresponding to the peak became smaller as the M/C increased. In addition, there is a tendency for the proportion of micropores in the range of 0.01–1 μm to increase with increasing M/C.
In the TG analysis, it was found that the mass loss tends to decrease with increasing M/C.
The decrease in permeability with increasing M/C is attributed to the pore filling by ettringite and the increased complexity of the pore structure.
This study confirmed that adding magnesium sulfate reduces the permeability of cement-treated soil. However, since this reduction likely depends on the presence of ettringite, its long-term durability needs to be examined. In addition, only one type of additive and soil was tested here. Therefore, further research using different additives and soils is necessary to develop more effective low-permeability soil cement and to clarify the limitations of this method.
Acknowledgements
This study was conducted with support from the Geo-science Center Foundation and the Japan Society for the Promotion of Science (KAKENHI Grant No. 25K07923). We would like to express our gratitude here.
