1 Introduction

Characterized by its uniform particle size distribution, low cohesion, and a propensity to expansion and settlement, sulfur-rich soil fails to provide adequate support for structural foundations, rendering it unsuitable for the construction of infrastructures. Sulfur-rich or sulfide soil is a variant of silty soil formed under waterlogged conditions and containing sulfide minerals (0.1–2% Sulfur) and typically 5–10% organic matter. Consequently, this leads to an elevated degree of saturation, exceeding the liquid limit of the soil, hampering its workability. Moreover, the High degree of saturation prevents consolidation, further increasing the soil’s tendency to settle under load. Sulfide soil strength parameters were specifically studied in several research projects (Larsson 1990; Westerberg et al. 2005, 2015), as sulfide soils can differ from other fine-grained soils due to their composition and organic content (Westerberg et al. 2015). Sulfide soil, recognized for its chemical composition and high organic matter, creates a distinct set of challenges within the fields of geotechnical and environmental engineering. Saye et al. (2013) emphasized the necessity of developing project-specific correlations for organic soils. However, research on organic soils is substantially less than that focusing on inorganic soils (Westerberg et al. 2015).

This type of soil appears harmless when saturated. However, when these soils are drained artificially for agriculture, infrastructure development, or excavation, they can become a significant environmental concern. This risk arises because sulfide soils turn to potential acid sulfate soils (PASS) since they undergo a chemical reaction when exposed to oxygen due to drainage or excavation. This reaction later can transform them into acidic sulfate soils (ASS). This transformation has a dramatic impact on the soil’s pH, causing it to drop significantly. This acidic environment triggers the release of potentially toxic metals naturally present in the soil minerals (Fanning et al. 2017; Joukainen and Yli-Halla 2003; Nyman et al. 2023; Yli‐Halla et al. 1999).

Excavation is frequently required for upper layers of sulfur rich soil in geotechnical projects due to the challenges posed by it, such as settlement, failure, and frost heave resulting from its mentioned characteristics. Managing the volume of excavated soil becomes crucial as it needs to be transported and deposited in landfills, consuming time, space, and financial resources. Due to its low strength and susceptibility to settlements, sulfide soil often requires replacement with gravel, while excavated soil is disposed. In Sweden, the cost for soil disposal ranges from 20 to 50 €/t, and due to the limited number of waste management facilities capable of handling sulfide soil, transportation costs of 10–20 €/t are common. Other techniques for stabilizing sulfur-rich soil in situ, such as implementing lime cement columns, have been employed to attain the desired mechanical properties. However, the efficacy of these methods is dependent upon various factors, including soil composition, the type and quantity of binding agent utilized, and the mixing methodology, among other relevant variables (Larsson 1999; Larsson et al. 2005). It is crucial to consider the economic aspects of projects, the depth of the soil layer and specific project requirements may necessitate excavation, despite endeavors aimed at in situ soil stabilization.

Therefore, it is assumed that by reusing sulfide soil as a geotechnical material, an environmentally and economically viable solution may be offered. It not only minimizes the need for landfill space but also contributes to sustainable and cost-effective geotechnical practices. While in-situ stabilization of sulfide soil has been a method to improve the soil characteristics (SGI 2015), less attention has been devoted to the stabilization of excavated sulfide soil with the aim of reusing it in geotechnical applications, e.g., noise barrier embankments and levees. In response to these challenges, some studies have been undertaken to improve the properties of silty soil. These studies have explored methods such as strengthening it with various materials, aimed at enhancing consistency limits and overall strength (e.g., de Jesús Arrieta Baldovino et al. 2021; Solihu 2020; Zhang et al. 2020; Zohair et al. 2019)).

Sulfur-rich soil presents unique challenges. From an environmental standpoint, it requires high water saturation to prevent oxidation and possible acidic leachate formation. However, for geotechnical applications, the soil needs physical improvement to be suitable for construction. Compaction, a common technique, is ineffective due to the fine, silty nature of the soil and its high-water content. One possible solution can be to stabilize the silty sulfur-rich soil at high water content through the introduction of binders. To improve the strength of the soil this approach can potentially enhance the soil’s properties and protect it from oxidation due to keeping the water content high during stabilization, ultimately making it a viable and more resilient material for geotechnical applications.

Admixtures can be used to stabilize soft and/or compressible soils, expansive soils, and wet soils. Cementing and pozzolanic agents, such as Portland cement, bitumen, lime, cement kiln dust (CKD) and lime-fly ash, effectively bond individual soil particles together, and also act to partially remove capillary passages, thereby reducing the potential for moisture movement. According to Federal Highway Administration (FHWA) Care must be taken when using lime and lime-fly ash mixtures with clay soils in seasonal frost areas. Stabilization, as used for expansive soils, refers to the treatment of a soil with such agents as bitumen, Portland cement, slaked or hydrated lime, and fly ash to limit its volume change characteristics. Chemical stabilization can substantially increase the strength of the treated material (Schaefer et al. 2016). Pozzolanic agents like fly ash reduces the potential of a plastic soil to undergo volumetric expansion by a physical cementing mechanism, which cannot be evaluated by the plasticity index. Fly ash controls shrink-swell by cementing the soil grains together much like Portland cement bonds aggregates together to make concrete (Association 2003).

The uniaxial compression strength test is widely employed to assess soil shear strength, as it is one of the quickest and most cost-effective methods for measuring undrained shear strength (Hossain et al. 2022). Various soil types have been studied using the uniaxial compression strength test to determine the effects of stabilization (Lindh and Lemenkova 2021; Luo et al. 2022; Pham et al. 2022; Rahmannejad and Toufigh 2018; Teing et al. 2019). Unconfined compressive strength (UCS) is a frequently used parameter for comparing the strength of soils after different treatments and binder dosages.

Various factors influence the UCS of stabilized soil samples, including the initial mixing water content of the soil, the type and amount of binder added, the curing time, and the final porosity of the soil. The importance of these factors and their effects on soil samples have been investigated in various studies. Henzinger et al. (2018) utilized the porosity/binder content concept to predict undrained shear strength (qu) of mixtures. The precision of the prediction was affected by the water content and homogeneity variation in the mixture. Rahmannejad and Toufigh (2018) studied different percentages of epoxy resin mixed with sand at varying water content levels and showed that the amount of epoxy resin added, curing time, and initial mixing water content were critical factors affecting the UCS of sandy soil samples. Increasing the water content had a negative effect on UCS, while increasing resin amount and curing time improved the strength of the soil samples. Moreira et al. (2019) investigated the effects of the porosity (η), dry unit weight of molding, cement content, and the porosity/volumetric cement content ratio (η/Civ) or void/cement ratio on the UCS of silty soil–roof tile waste (RT) mixtures. Their study demonstrated that it was possible to estimate the UCS of cement-stabilized samples based on the porosity/volumetric cement content ratio. Several studies have shown that the relationship between soil sample parameters and qu can be described using an equation or model with variables including porosity, binder, and water content (Pham et al. 2021; Soleimani et al. 2018; Tinoco et al. 2019; Zhang et al. 2020). While similar relationships between parameters have been explored in the broader context of soil stabilization, the specific characteristics of sulfide soil necessitate a dedicated inquiry. Understanding the intricate interactions between parameters such as water content, binder content, and compaction is of paramount importance. Efficient utilization of excavated sulfide soil can improve the success of project by increasing project performance, reducing environmental impact and cost and enhancing sustainability.

(Pham et al. 2021; Soleimani et al. 2018; Tinoco et al. 2019; Zhang et al. 2020). While similar relationships between parameters have been explored in the broader context of soil stabilization, the specific characteristics of sulfide soil necessitate a dedicated inquiry. Understanding the intricate interactions between parameters such as water content, binder content, and compaction is of paramount importance. Efficient utilization of excavated sulfide soil can improve the success of project by increasing project performance, reducing environmental impact and cost, and enhancing sustainability.

While research exists on sulfide soil stabilization, it primarily focuses on either in-situ treatment or stabilizing excavated soil for applications requiring low water content. However, a crucial knowledge gap persists: the feasibility of stabilizing excavated sulfide soil with high water content for reuse in geotechnical applications like embankments. This study aims to bridge this gap by investigating the predictability of final strength in stabilized, high-water-content sulfide soil. It explored the relationships between initial porosity, binder content, and initial mixing water content on the final unconfined compressive strength (UCS) of the stabilized samples. To achieve this, a compaction method specifically designed for high-water-content sulfide soils was developed. Soil samples with five different water content levels were mixed with three varying binder percentages, all compacted with equal energy. After a 16-day curing period, the UCS tests was conducted to evaluate the effects of these studied variables.

The findings of this study have the potential to positively change how excavated sulfur rich soil is utilized. By enabling its reuse in geotechnical applications, this approach can significantly reduce disposal costs and minimize environmental impact associated with transportation and landfilling. This paves the way for more sustainable and cost-effective practices in geotechnical engineering.

2 Material

2.1 Method of Soil Characterization

The studied soil was collected from Umeå, Västerbotten, in northern Sweden. The water content of the soil was determined, using the ratio of water to solid mass of the soil (ASTM 2010b). The plastic and liquid limits were measured according to ASTM (2010c) and fall cone method (Wasti 1987). The particle size distribution of the soil was obtained from sieve analysis and sedimentation following the Swedish standard (International Organization for Standardization 2004). The binder product chosen for the experiment was Multicem (MC), from Cementa AB. Multicem contains 50% Portland cement and 50% cement kiln dust (CKD) which is a by-product of cement manufacturing (Siddique 2006). The utilization of Muticem presents a more environmentally sustainable alternative compared to using solely cement due to its composition, which includes CKD, thereby reducing the environmental footprint associated with conventional cement usage. A modified Proctor compaction test has been conducted to determine the maximum dry density of pure soil before stabilization based on ASTM (2012). A chemical analysis of the soil has been conducted obtained in an accredited laboratory, with a particular focus on the quantification of sulfur content. In addition to sulfur, the concentrations of other elements in the soil were also analyzed.

2.2 Material Properties

The results of the soil characterization are presented in Table 1. The original water content was 50% with PI (Plasticity index) = 7%, which categorized the soil under the medium plasticity type of soil (Prakash 2002). The Particle distribution curve is presented in Fig. 1a. The soil is silty soil (M) according to Unified Soil Classification System. Figure 1b presents the proctor compaction curve.

Table 1 Physical properties of the studied soil
Full size table
Fig. 1
figure 1

(a) Particle size distribution curve. (b) Proctor soil compaction test graph for modified methods

Full size image

The chemical composition of the soil is presented in Table 2. The values are based on mg/kg of material dry weight. The findings indicate that the values fall within the standard range as per the Compendium Remediation of Contaminated Sites in Sweden (Swedish EPA 2021). Upon comparing the chemical composition of the sulfide soil with Swedish criteria for assessing polluted soil, it was found that the concentration of elements remains below the guideline limit for “less sensitive soil,” permitting various land uses (Swedish EPA 2021). Barium is the sole element surpassing the criteria, yet its level is typical in organic-rich soil. Since large areas of sulfide soil are utilized for agriculture in Sweden and Finland without issue, the soil is not classified contaminated. However, due to elevated sulfur and iron levels, it is classified as sulfur-rich soil.

Table 2 Chemical composition of tested soil
Full size table

3 Method

The experimental plan is summarized in Table 3. A five-step experimental plan was followed to achieve the study’s objective. The uniaxial compression test was conducted after curing the samples for a specific period.

Table 3 Experimental plan
Full size table

3.1 Soil Preparation

Since the available mold size for uniaxial compression test samples was 5 cm in diameter and 10 cm in height, the soil was sieved to segregate particles larger than 25 mm (50% of the mold diameter). The soil was sifted by sieve size 25 mm to separate larger particles. Since the original soil was wet, sieving it with smaller net was not feasible. As drying such soil before sample preparation was not an option because of risks of changing the soil structure during the drying process, particles larger than 1/6 of mold diameter (ASTM 2006) were removed by hand, based on ocular control. The segregated soil was dried in the lab environment at temperatures between 21 and 24 °C for 10 days to limit risk of soil aggregation occurring at higher temperatures. To obtain samples with different water contents, soil samples were collected at different times under the drying period. The collected samples were preserved in a closed container to maintain constant water content. During a 10-day period, four soil samples with different water content were collected. The samples water contents are presented in Table 4.

Table 4 Water content (ω) of the soil after each step of drying
Full size table

3.2 Mixture Preparation

The soil with five different water content levels were mixed with three Multicem (binder) contents (MC0 = 4%, MC1 = 6%, MC2 = 8%).

To assess pH changes resulting from varying Muticem amounts, a series of experiments were conducted. Each experiment involved preparing a mixture of soil-binder and water with a water-to-solid material ratio of 5 (Al-Jabban 2019). The soil mixtures Muticem in percentages ranging from the minimum of 2% to increments of 4%, 6%, 8%, and 10%. For each mixture, the binder percentage was determined by a percentage of the dry weight of the soil to maintain a constant ratio between binder and soil particles at different initial mixing water content. The binder mixture and water were mixed thoroughly and allowed to stand in the laboratory for 1 h, with intermittent mixing every 10 min. After 1 h of mixing, the pH of the solution was measured.

The preparation of the mixtures for the stabilizing purpose was started by mixing pure soil with a stand mixer to gain a uniform soil mass. Then the binder was added to the soil and mixed with it. The blending process was continued until a homogeneous mixture was achieved based on ocular assessment.

3.3 Sample Preparation

The samples were prepared in cylindrical tubes measuring 5 cm in diameter and 10 cm in height. Due to a high water content, compacting the samples with a Proctor hammer became impractical as water escaped from the mold after each hammer impact. To address this issue, a smaller compactor (as shown in Fig. 2a) was employed for sample compaction. Each sample was compacted in 5 layers, with each layer being hand-hammered 25 times (refer to Fig. 2b). It is noteworthy that this hammer is considerably lighter than the standard hammer used in Proctor compaction, indicating a significantly lower compaction energy compared to the Proctor method. The sample preparation procedure remained consistent across all samples, regardless of their water content and binder percentage. Samples were prepared within 1 h after the mixing process. Furthermore, in order to compare the outcomes of hand-compacted molding (HC-M) with the modified Proctor test (MP) on the samples, the proctor test was conducted on samples containing varying Multicem contents. It was taken into account that at higher water content levels, especially at 34%, some water loss occurred in the samples. This was done to examine how hand-compacted molding influences the dry density of the samples at various initial mixing water contents.

Fig. 2
figure 2

a Hammer, b sample layers, c prepared sample

Full size image

3.4 Curing Time

After preparation, the samples were sealed and kept in a humid container to maintain the water content. Samples were cured for 16 days in storage at 22 °C temperature. It was suggested that samples should be submerged under water before the test to reduce suction (Consoli et al. 2011). This step was omitted since samples with higher water content tended to disintegrate after exposure to water.

3.5 Unconfined Compression Test

The employed uniaxial compression device had 250 kg maximum capacity under constant displacement control. A compression load plate with a rate of 1 mm/min, equal to 1% of the height of the samples was applied to the specimens (ASTM 2006). The continuous displacement to specimens continued up to the failure of samples. The maximum load recorded by the apparatus was considered for the calculation of the undrained shear strength of the soil.

3.6 Specific Gravity

The porosity of specimens was measured to fulfill this study’s aim. In previous studies (Consoli et al. 2011) porosity of samples was measured by particle density of mixture. They employed a mathematical relation between particle density of pure soil and binder and percentage of binder to calculate the particle density of mixture. In this study, the specific gravity of cured samples was measured directly after uniaxial compressive test according to ASTM (2010a) for each sample, and mathematical calculation was disregarded.

4 Results and Discussion

Exposure of sulfide soil to oxygen leads to oxidation, which causes acidification and aggregation of soil particles. Additionally, the chemical compounds present in sulfide soil can alter its structure upon drying, potentially affecting its mechanical properties. Despite these changes being known, the impact on the mechanical properties of the soil remains poorly understood. To prevent any possible chemical changes that could influence the stabilization result during the drying process and subsequent mixing with water to prepare the soil with varying water contents, it is recommended to exercise ocular control rather than completely drying the soil and then sieving it. Furthermore, to avoid any changes in soil texture and properties, the soil was gradually dried at room temperature.

The use of cement for soil stabilization offers numerous advantages in construction projects. However, its effectiveness can be significantly compromised in environments with sufur-rich soils. This phenomenon which called sulfate attack disrupts hydration, alter microstructure, and weaken bonding between soil particles. Consequently, sulfur-induced reactions lead to volume changes, cracking, and reduced durability, compromising load-bearing capacity. Understanding these mechanisms is crucial for optimizing binder selection and dosage in sulfur-rich environments to enhance long-term performance (Kalıpcılar et al. 2016; Rollings et al. 1999).

The primary culprit in sulfate attack is the presence of sulfate ions (SO42−). These ions originate from various sources, including Pyrite (FeS2). Pyrite undergoes oxidation in the presence of oxygen and water, leading to the formation of sulfuric acid (H2SO4) or iron sulfate (FeSO4) (Burland et al. 2012). Which considering the fact that soil and cement need to be mixed and during this mixing process oxygen will be introduced to the mixture even though the water content of the soil is high.

Sulfate attack manifests through reactions between the sulfate ions and the hydration products formed during cement’s interaction with water. These hydration products play a critical role in the strength and stability of the stabilized soil. The key targets include:

  • Calcium hydroxide (Ca(OH)2) This readily reacts with sulfate ions, forming gypsum (CaSO4·2H2O).

  • Calcium-Silicate-Hydrates (C-S–H) The primary binding agents in cement, C-S–H are also susceptible to degradation under prolonged exposure to sulfate ions (Alam et al. 2012).

While the fundamental chemical reactions involving sulfur consumption remain similar, the inclusion of CKD in the cement mixture may impact the overall reactivity and effectiveness of the binder in stabilizing sulfur-rich soils. Further research and experimentation are necessary to fully understand the influence of CKD on sulfur consumption reactions and to optimize the formulation of cement mixtures for soil stabilization in sulfide-rich environments.

Based on the determined sulfur content in the soil, the mass of sulfur was converted to moles utilizing its molar mass. Concurrently, the molar mass of the Muticem mixture, comprising CKD and Portland cement, was calculated. Employing the stoichiometric ratio of sulfur to cement, the mass of Muticem required to react with all sulfur present in the soil was determined. This calculation, yielding a quantity of approximately 28 g of Muticem per kilogram of soil, signifies a critical step towards effective soil stabilization.

In addition to determining the minimum Muticem quantity required for sulfur consumption, despite having a relatively neutral pH initially, the pH of the soil is crucial for cement-based stabilization. During hydration reactions, the soil pH should fall within the range of 8–12 to ensure effective binder performance. When lime or cement is applied to the soil, it raises the pH of the system. This results in the release of calcium and the dissolution of clay alumina and silica, leading to the formation of pozzolanic compounds (Talluri 2013). Results demonstrated that the addition of 4% Muticem successfully raised the soil solution pH above 8. As the result, it was determined that Muticem dosages of 4%, 6%, and 8% would be examined further to observe their stabilization effects on the soil. This pH testing offers critical insights into optimizing Muticem dosage to achieve the desired pH range for effective soil stabilization. Ensuring pH alignment within the specified range enhances the efficacy of the cement-based binder, promoting increased soil strength and stability in sulfide-rich environments.

4.1 Relation Between Water and Cement Content and Compressive Strength

Figure 3 assumes a linear relationship between the undrained shear strength (qu) and binder percentages at each specific initial mixing water content. Equations qu0 to qu4 correspond to the undrained shear strength of stabilized samples at the highest to lowest initial mixing water contents, respectively. It was observed that qu increased with the addition of more binder, regardless of the soil’s initial mixing water content. Furthermore, samples with an initial mixing water content of 50% had the lowest qu values among the prepared mixtures. However, the strength of the prepared samples was found to improve by lowering the initial mixing water content of the soil to 24%. The strength increase corresponds to the slope of the respective lines (qu0 to qu4). At a water content less than 24%, the rate of strength development was found to be faster for mixtures with higher binder percentages than those with higher initial mixing water content (34% and 50%). One hypothesis for this finding is that samples with a 24% water content could be compacted to a higher density, which led to an increase in the sample’s strength. Another explanation could be attributed to the cementation bond that occurred at a lower water content in the presence of different binder dosages.

Fig. 3
figure 3

Relation between qu with binder percentage for soil with various water content, (MC:Binder percentage, \({B}_{n}\) Line equation constant)

Full size image

Figure 4 presents a comparative analysis of dry density achieved through hand compaction and the modified Proctor test on soil samples with varying binder content and water content. As anticipated, the modified Proctor test consistently generates higher dry densities across all sample groups (sample groups defined by different binder content levels). This observation aligns with established principles, as the Proctor method is specifically designed to deliver a standardized and higher compaction energy compared to hand compaction.

Fig. 4
figure 4

Relation between water content and dry density of samples with different percentages of binder

Full size image

However, beyond the observed difference in density between the two methods, noteworthy trends emerge from the data. The data demonstrates a positive correlation between binder content and dry density for both compaction methods. As the binder content increases, a corresponding increase in dry density is observed. This suggests that the presence of binder enhances the packing efficiency of soil particles, resulting in a denser compacted material. An intriguing observation is the convergence of dry density results between the Proctor and hand compaction methods at the extreme water content levels (34% and 15%). This suggests that when the water content deviates significantly from the optimum water content, the influence of compaction energy on the final density becomes less pronounced. Conversely, samples with water content closer to the optimum water content (24%) exhibit a more substantial difference in dry density based on the compaction method employed. This highlights the critical role of optimum water content in maximizing the impact of compaction energy on achieving higher dry densities.

According to Fig. 4, the dry density of the samples increased as the water content decreased at all binder percentages. The samples had the highest dry density and correspondingly the highest qu at initial mixing water contents around 24%. Subsequently, the dry density decreased as the initial mixing water content was further reduced. As a result, it was found that further decreasing the water content beyond 24% resulted in a lower qu. When binder is added to the soil, the resulting mixture becomes drier than pure soil because the binder is a dry substance, and the hydration reactions between the binder and water consume water. For instance, Table 5 shows the average water content of samples after mixing preparation and curing at 34% initial mixing water content. After 16 days of curing, the amount of water in the soil reduced to a level lower than the water content after mixture preparation. Since the sample tubes were sealed and preserved in an airtight container with high humidity, the difference in water content between the mixture after preparation and after 16 days of curing time could be attributed to the hydration reactions in the presence of the cement-based binder in the mixture.

Table 5 Comparison of initial mixing water content of soi at 34% with water content of soil after curing time
Full size table

4.2 Relation Between Porosity, Binder Content and qu

Figure 5 illustrates the effect of adding various amounts of binder to soil mixtures on porosity. The results show that porosity decreased with increasing binder content at all initial mixing water contents. The reduction in porosity was due to the decrease in water content caused by the addition of the binder, which facilitated compaction. Figure 4 shows that the addition of binder allowed for higher densities in mixtures at all initial mixing water contents compared to pure soil. The formation of cementation bonds between the binder and soil particles filled the voids and cracks in the sample, leading to a reduction in porosity and an increase in density. Baldovino et al. (2021) suggested that the addition of binder enhanced inter-cluster bonding and reduced the pore space in the soil structure, which supports this explanation. In Fig. 5, an inverse relationship between porosity and binder content in the prepared samples was demonstrated. From this, it can be concluded that adding a specific amount of binder to the soil can reduce porosity by a specific amount. However, the changes in porosity corresponding to changes in binder percentage were not as significant as when the water content was closer to the optimum amount. This finding is crucial for practical applications, as it indicates that achieving specific characteristics in soil-binder mixtures may require careful consideration of the water content and binder percentage.

Fig. 5
figure 5

Relation between porosity and binder content in different initial soil water content

Full size image

Figure 6 displays the relationship between porosity and volumetric binder content, which is the ratio of binder volume to the total volume in each cured sample. This ratio was calculated using the following equation:

$${MC}_{iv}=\frac{{V}_{MC}}{{V}_{Total}}$$
(1)

where \({MC}_{iv}\) is volumetric binder content, \({V}_{MC}\) is volume of the binder in each sample tubes, \({V}_{Total}\) is total volume of each sample tube.

Fig. 6
figure 6

Variation of porosity versus volumetric binder content

Full size image

The amount of binder used was based on the dry matter of soil, resulting in varying volumes of the additive depending on the initial mixing water content and dry density of the samples. The relationship between porosity and volumetric binder content, which represents the ratio of the volume of binder to the total volume of each cured sample, is shown in Fig. 6. The trend of decreasing porosity with increasing volumetric binder content applies for all initial mixing water contents. The higher volume of binder at lower initial mixing water content can explain the difference in the rate of strength increase shown in Fig. 3. Specifically, the soil with a 24% initial mixing water content had the highest volume of binder, least porosity, and exhibited the greatest strength enhancement with increasing binder content.

The presented findings in Fig. 7 reveal the relationship between qu and porosity of the samples. It can be observed that a decrease in porosity generally leads to an increase in qu. Furthermore, it is noteworthy that the decline in strength due to porosity exhibited a correlation with initial mixing water content. Figure 7 also enables a comparison of the porosity of samples resulting from different initial mixing water contents. Samples with an initial mixing water content of 50% displayed higher porosity and the lowest qu at all binder dosages, in comparison to the other samples. The line corresponding to the soil with 50% water content further demonstrated that the porosity of samples varied less when compared to other soils with lower water contents, implying that the mixture was difficult to compact at this water content. On the other hand, a decrease in water content to 24% resulted in a maximum variance of 45% in porosity with increasing binder content. This illustrates that binder dosages at the optimum water content (24%) have a greater impact on enhancing the characteristics of the mixture, including porosity and strength, compared to other initial mixing water contents.

Fig. 7
figure 7

Relation between unconfined compressive strength and porosity in different soil initial mixing water content

Full size image

The effect of binder content and porosity on the qu of the studied soil was analyzed separately. As depicted in Figs. 3 and 7, it is apparent that the qu of the studied samples is affected by the porosity and binder percentage for each initial mixing water content. This conclusion is consistent with the findings reported by Consoli et al. (2011), who observed a nonlinear relationship between qu and porosity of the silty soil samples stabilized with cement. However, in this study, a linear correlation between these parameters was observed (Fig. 7). This difference could be attributed to the variation in soil type, sample preparation methods, and curing time. The current study samples were prepared under equal compaction energy, resulting in different dry densities, depending on the binder and water content. Additionally, the samples were cured for 16 days, as opposed to the 7 days in the Consoli et al. (2011) study, and were not saturated before the UCS test. The results indicate that adding binder increased qu, while increasing porosity decreased it. Thus, these two variables have different impacts on the strength of the stabilized soil. Studies conducted by Consoli et al. (2011) and Santana et al. (2021) have used the relation between qu versus porosity/binder ratio to determine the effect of these variables simultaneously on the qu of soils mixed with a binder, in other words, how changes in the ratio of these variables affect the strength of the samples.

For this purpose, a variable was used which is the ratio of porosity to volumetric cement content, and it was calculated using the following equation:

$$\frac{\eta }{{MC}_{iv}}=\frac{\left(\frac{{V}_{v}}{{V}_{Total}}\right)}{\left(\frac{{V}_{MC}}{{V}_{Total}}\right)}=\frac{{V}_{v}}{{V}_{MC}}$$
(2)

where \({V}_{v}\) is the voids volume in the sample.

The relationship between qu and porosity/volumetric binder content ratio was investigated and is presented in Fig. 8. It can be observed that an increase in porosity/volumetric binder content ratio led to a decrease in qu for all water contents. The lowest qu values corresponded to the highest porosity/volumetric binder ratio. This trend was evident for soil samples with the highest water content, which was equal to the original water content of the soil. The effect of density was also noticeable in the graph, with samples that had the highest density (24% initial mixing water content) having the highest qu and lowest \(\frac{\eta }{{MC}_{iv}}\) ratio. Additionally, the graph clearly shows the effect of binder content on qu, with samples containing lower binder content corresponding to higher porosity/volumetric binder ratio dots on each initial mixing water content curve. As more binder was added, \(\frac{\eta }{{MC}_{iv}}\) decreased, and qu of samples improved.

Fig. 8
figure 8

qu of soil versus (porosity/volumetric cement)

Full size image

Furthermore, Fig. 8 indicates that there is a mathematical relationship between qu and porosity. It can be seen that for each water content, qu is a function of \(\frac{\eta }{{MC}_{iv}}\). The curves also suggest that the relationship between these variables is not unique and varies for each initial mixing water content. The function is as follows:

$$q_{u} = A\left[ {\frac{\upeta }{{MC_{iv} }}} \right]^{ - a}$$
(3)

The aforementioned functions can be expressed in terms of two constants, denoted as “A” and “a,” which are unique to each initial mixing water content condition. It is important to note that, while the values of these constants vary between different initial mixing water content conditions, they exhibit a decreasing trend with decreasing water content. This observation highlights the fact that variations in the strength of samples with higher initial mixing water content are more sensitive to changes in the \(\frac{\eta }{{MC}_{iv}}\) ratio than those with lower initial mixing water content. It is worth mentioning that although changes in this ratio may lead to a relative increase in strength, it does not necessarily imply that reducing this ratio will result in achieving the desired strength level. Additionally, it is evident from Fig. 8 and the equations that samples with different initial mixing water content and equal \(\frac{\eta }{{MC}_{iv}}\) ratios exhibit varying levels of calculated strength. These findings suggest that while porosity and binder content are both factors affecting the strength of the mixtures, their relative influence may differ considering their initial mixing water content. It is due to incompressibility of soils with high water content. Therefore, changes in \(\frac{\eta }{{MC}_{iv}}\) value correspond to changes in the amount of binder in the mixtures, while the porosity remains almost unchanged. Previous research has also demonstrated that the coefficients of porosity (η) and a power of \(\frac{1}{{MC}_{iv}}\), which determine the rate of variation of qu, are unique to the type of soil and the quantity of fine particles present in the soil (Consoli et al. 2007, 2011, 2016; Moreira et al. 2019).

Figure 9 illustrates a significant correlation between qu and the coefficients of porosity (η) and \({MC}_{iv}\). Notably, the results demonstrate that qu can be expressed as a power function of \(\frac{\eta }{{{MC}_{iv}}^{0.2}}\) indicating that the influence of porosity on the strength of the stabilized samples is more significant than that of binder dosages. Thus, compaction plays a crucial role in improving the mechanical properties of stabilized samples, as it modifies the strength of the soil.

Fig. 9
figure 9

Adjusting porosity/binder ratio for predicting qu of samples

Full size image

The qu depends on the water content, soil maximum density at optimum water content and binder dosage. It should be noted that the importance of parameters influencing the strength of stabilized samples is different, and their impact on the named parameter is not equal. It is observed that to find a power function for qu that mathematically fits the measured data with the highest accuracy, it is necessary to divide the data into two groups based on water content. In the following, it is found that constant value a is equal 0.2 for both equations with coefficient of determination equal to 0,95 and 0,9 for ω > 24 and ω ≤ 24%, respectively. The equations are as follows:

$$\omega > 24\% , q_{u2} = 3 \times 10^{5} \left( {\frac{\upeta }{{MC_{iv}^{0.2} }}} \right)^{ - 4} , R_{2}^{2} = 0.95$$
(4)
$$\omega \le 24\% , q_{u1} = 3 \times 10^{8} \left( {\frac{\upeta }{{MC_{iv}^{0.2} }}} \right)^{ - 2} , R_{1}^{2} = 0.9$$
(5)

Figure 9 clearly indicates a noticeable difference in the qu of samples, depending on their initial mixing water content. Samples prepared at ω ≤ 24% had a higher qu value than the other group of samples with higher water content. This difference confirms the significant impact of porosity on the strength of stabilized samples. The amount of \(\frac{\eta }{{{MC}_{iv}}^{\text{0,2}}}\) in the equation is controlled by both porosity and binder content. At ω > 24%, the porosity of samples was higher, and reducing the porosity was less feasible due to the water hindering compaction. Therefore, the named parameter was changed by modifying the amount of binder. However, it should be noted that increasing the binder content beyond the optimum amount can adversely affect the strength of the mixture (Horpibulsuk et al. 2003), and this has not been considered in the presented equations. Moreover, considering the economic aspect of a project, the binder amount must be feasible. The observed strength behavior of samples at different water content can be attributed to chemical reactions between the binder and water, creating new structures between soil particles and altering soil strength accordingly. When the water-binder proportion in a mixture is higher, it creates a wider space between cement and soil particles, affecting cementation reactions. This occurrence is associated with water preventing compaction and causes a low qu value for samples with water content higher than 24% (ω > 24%). This finding is crucial for projects that offer a range of options to achieve design requirements, taking into account compaction equipment, binder cost, and estimated project duration. Additionally, considering that the studied silty soil is sulfur-rich, and high-water content plays a significant role in controlling acidification and environmental consequences, knowledge of the relationship between qu and \(\frac{\eta }{{{MC}_{iv}}^{\text{0,2}}}\) can help select the best mixture recipe and adequate compaction energy for the project’s goals.

The current study investigates the relationship between the unconfined compressive strength (qu) and the ratio of porosity (n) to volumetric binder content (MCiv) in stabilized soil. While an exponential relationship between these parameters has been established by Consoli et al. (2016), the results from this study show significant discrepancies. Notably, the stabilized samples in the current research exhibit unconfined compressive strengths approximately ten times less than those reported by Consoli et al. despite comparable binder and water contents.

Analysis of the porosity and dry density data suggests no significant differences between the two studies. However, several potential factors may contribute to the observed disparity in strength. Both studies utilized cement-based binders; however, the current research employs a cement-CKD blend (Multicem). While CKD offers some stabilization benefits, its effectiveness is generally lower compared to pure cement (Sariosseiri et al. 2011). This difference in binder type could partially explain the observed strength variation. A critical distinction lies in the sample preparation procedures. Consoli et al. submerged their samples underwater for 24 h to minimize suction. In contrast, the current study tested samples immediately due to concerns about potential sample destruction with lower density and cement content during submerged curing. Attempts to increase saturation within the sample tubes were unsuccessful, leading to the omission of this step. Consoli’s work, however, lacks information regarding the final degree of saturation achieved after the 24-h submersion period. For samples with higher density in the current study, the degree of saturation exhibited minimal change after 24 h due to their low porosity. The presence of sulfur in the soil used in the current study may be another contributing factor. Sulfur is known to react with cementitious materials, potentially impacting the hydration process and influencing the formation of cementitious compounds, ultimately affecting the overall strength development of the stabilized soil. Further investigation into the specific interactions between sulfur and the cement-CKD binder is warranted.

This study has potential limitations since the discussions are based on laboratory measurements, and various factors can influence the performance of the test apparatus, thereby affecting the certainty of the measured values. The primary limitation of the results is associated with the sample preparation method. The samples were hand compacted, and there is no certainty that the energy of compaction for all samples was exactly equal. Consequently, this variability might have exceeded the tolerances of the sample density. Another significant consideration is that the samples used in this study were not saturated, and the effect of suction on the measured properties should be considered if these results are to be used for design purposes. Suction can substantially affect the strength and stiffness of unsaturated soils, potentially yielding different properties compared to fully saturated conditions. Future research could explore the effect of saturation on the properties of cement-stabilized sulfur-rich soils to provide more comprehensive data for design and construction purposes.

Evaluating the long-term behavior of sulfide soil stabilized with binder is critical for its successful application in geotechnical projects. This assessment should encompass two key aspects: First, Monitoring the unconfined compressive strength (qu) of stabilized soil samples at extended curing times is essential. This provides insights into the long-term efficacy of the binder in enhancing soil strength and durability. Second, Monitoring the pH of the stabilized soil over time is crucial. Changes in pH can indicate the occurrence of pozzolanic reactions, where the binder interacts with the soil to form cementitious compounds and potentially improve strength. Additionally, monitoring pH is necessary to assess the effectiveness of the binder in preventing soil acidification, a potential concern with sulfide soils.

Al-Jabban (2019) showed the immediate effects of adding Multicem on soil pH values. It highlights how increasing the binder content leads to a rise in soil pH. He also mentioned the variations in pH over time for different binder types including Multicem and emphasized the promising results in maintaining soil pH above acidic level with Multicem stabilization.

While leaching can be a concern for soils containing harmful elements, it’s less critical in uncontaminated scenarios. Nevertheless, monitoring the pH at various stages (initial, post-mixing, and post-curing) provides valuable data on potential changes. Maintaining a neutral pH range is crucial for mitigating potential environmental and structural issues associated with sulfide soil. A low pH can not only increase leaching risk, which might be less relevant here but also negatively impact vegetation, aquatic life if leachate enters waterways, and infrastructure integrity. Therefore, the primary focus during stabilization should be ensuring the pH remains within a safe zone throughout the treatment process. This can help minimize potential environmental and structural impacts. Further research is warranted to gain a deeper understanding of the long-term interactions between sulfur-rich soil and cement-based binders. This knowledge is essential for optimizing stabilization techniques and enabling the safe and sustainable use of these soils in construction applications.

5 Conclusion

In this paper, the impact, binder dosage, and initial mixing water content on the properties of stabilized silty soil was studied. A silty soil with a total of five different water contents were mixed with three dosages of Multicem as the binder. Samples were prepared with equal energy of compaction and cured for 16 days. Then, unconfined compressive strength tests were conducted on the samples. The following conclusions were derived from the experiment and discussion, taking into consideration the limitations of laboratory conditions.

It was observed that the undrained shear strength of the samples was directly dependent on the dosages of the added binder. This pattern was observed for all initial mixing water contents at the studied dosages. The effect of increasing binder dosage was more evident when the water content was equal to or less than the optimum water content.

Furthermore, the impact of variations in water content and binder dosage on porosity was identified as an essential variable to enhance the strength of the samples. It was seen that porosity reduction was directly correlated with an increase in the strength of the samples.

The influence of variables, including porosity and binder dosage on soil strength, was evaluated. It was shown that qu follows a function of the ratio of these variables. However, the effect of porosity on the qu was more significant than that of the binder percentage. Moreover, it was revealed that the relationship between qu of the samples and the porosity/binder ratio was not unique and depended on the initial mixing water content of the sample.

By employing \(\frac{\eta }{{{MC}_{iv}}^{\text{0,2}}}\) as a variable, it was shown that the strength of samples at initial mixing water content higher than the optimum level was less than the qu of other samples with equal or lower water content than the optimum considering the fact that samples were not saturated. Therefore, reducing the initial mixing water content of the mixtures and lowering the porosity by compaction are practical options for reaching higher strengths in stabilization with binder.

Additionally, the formula introduced in this paper can help predict the strength of samples based on variables such as water content and binder content. This can lead to higher strength, taking into account soil conditions and possible changes that can be made to the soil.