1 Introduction

Saga prefecture, located in the northwest part of Kyushu Island, Japan, faces significant engineering challenges due to the unique properties of Ariake clay, including high compressibility, sensitivity, and low permeability [1, 2]. In response to these challenges, engineers have implemented the Deep Mixing Method (DMM), a ground improvement technique, in the Saga lowland [3]. However, ensuring the quality of cement columns remains a critical concern in the field, particularly given the limitations of traditional assessment methods such as the phenolphthalein tracer method. This method relies solely on visual observation of the purple color, indicating alkalinity due to cement presence, without providing accurate insights into the compressive strength of the improved soil. Recognizing these limitations, our study introduces a controlled and standardized experimental protocol utilizing commercial kaolin clay. Additionally, we explore the potential substitution of the traditional phenolphthalein tracer method with Handheld XRF, aiming to improve the accuracy and reliability of quality assessment in DMM applications.

The handheld X-ray fluorescence (XRF) is employed as a non-destructive analytical tool to quantify the elemental composition, focusing on key constituents such as Calcium (Ca), Silicate (Si) [4], and Sulfate (S). These elements play a crucial role in the pozzolanic reaction equation during the hydration process, influencing the durability and strength of the material. This mechanism is widely acknowledged to stem from the pozzolanic reaction occurring during the hydration process, succinctly represented as Ca(OH)2 + H4SiO4 → CaH2SiO4·2H2O [4]. The main phases present in the microstructure of the hydrated cement paste can be listed as calcium silicate hydrate (C-S–H), calcium hydroxide (Ca(OH)2) or CH, ettringite, monosulfate, unhydrated (UH) cement particles, and air voids [5]. This approach not only reveals the behaviors of Ca, Si and S over time for strength appreciation, it also facilitates the distinction of well-mixed cement parts, and offers benefits compared to the traditional phenolphthalein tracer method, which fails to assess the state of strength adequately.

The application of handheld XRF in geological studies is well-established in fieldwork [6], however, its limited background in analyzing cement-improved soil especially with portable devices, underscores the crucial point of this study. Utilizing XRF offers advantages such as speed, accuracy, and precision in assessing the chemical composition of samples [7]. Samples were assessed both before and after the mixing process, following a controlled and standardized experimental protocol. This protocol facilitated a thorough evaluation of needle penetration strength (quNP). Specimen preparation poses challenges, as noted by Broton and Bhatty [8], emphasizing the importance of meticulous preparation for accurate quantitative XRF analysis. For this study a small cylindrical mold (SM) measuring 0.03 m × 0.03 m was adopted for enhanced accuracy. The cement dosage of 110 kg/m3 and a W/C ratio of 1, using 10-min mixing times were implemented [9]. Handheld XRF measurements were conducted at different intervals: powder state before the mixing and 0 day, 1 day, 7 days, 28 days at post-mixing.

Results indicate the apparent amount of Ca’s increase proportionally with curing time, coinciding with the rise in average needle penetration strength (quNPA). The escalation of Ca suggests ongoing hydration reactions, resulting in the generation of ettringite, (Ca(OH)2) or CH and C-S–H gel. Conversely, the observed decrease in Si amount over time suggests its integration into the hydration product [10] formed during cement hydration, such as C-S–H gel. Through Scanning electron microscopy (SEM) investigations, the formation of hydrated products such as ettringite, (Ca(OH)2) or CH and C-S–H gel, and, were assessed to identify the morphology and the types of crystals formed during the hydration process [11]. Early strength assessment after 1 day revealed initial strength development in the specimen, with significant strength development observed after 28 days which showed a correlation between the measured Ca, Si and S by the Handheld XRF and the results discovered using SEM. This research constitutes a preliminary exploration into the potential of substituting the traditional phenolphthalein tracer method with the Handheld XRF in the field of soil improvement by cement.

2 Field data for phenolphthalein tracer method in Saga lowland

In Japan, the traditional method involves visually assessing the intensity of the purple color, after spraying the phenolphthalein, which indicates a high alkaline state due to presence of cement in the DMM. Its chemical formula is C20H14O4 meaning that each molecule of phenolphthalein contains 20 carbon atoms, 14 hydrogen atoms, and 4 oxygen atoms [12]. In acidic solutions, there are enough hydrogen ions to keep phenolphthalein colorless. But when the solution becomes slightly basic, the phenolphthalein molecules gain hydrogen ions, leading to a change in their structure and resulting in a pink to purple color [13].

Figure 1 presents field data collected in Saga prefecture to evaluate the strength development of DMM constructions. Despite observing a strong purple color, signaling the alkaline nature resulting from cement presence, the constructed columns lack the required strength for road structure applications. This highlights the limitations of the phenolphthalein tracer method in reliably ensuring strength improvement post-application of cement columns.

Fig. 1
figure 1

Use of phenolphthalein (1%) tracer on improved Ariake Clay for quality control assurance (field data)

Full size image

Figure 2 shows a field result after a specified curing period following casting, the improved columns were excavated by boring to obtain test soil. The natural moisture content of the test soil was measured, and then approximately 10 g of wet soil was separated and crushed with a mortar. Subsequently, the wet soil was diluted in a sample bottle to achieve a ratio of wet soil to distilled pure water of 1:5, shaken by hand until it became liquid, and used for pH and electrical conductivity (Ec) measurements.

Fig. 2
figure 2

Comparison of chemical properties of improved Ariake Clay for quality control assurance

Full size image

After 28 days of curing, the unconsolidated nature of the improved soil rendered it impossible to conduct the unconfined compression test for evaluating compressive strength. Despite the presence of cement, the soil displayed high malleability and remained unconsolidated, precluding the assessment of its compressive strength. the water content (Wn) within the improved column ranged from 80 to 150% at the depth of 0 to 13 m, indicating a remarkably high level of moisture that is not conclusive to strength development. The pH levels obtained during prior ground investigation, compared to the unimproved area (natural soil), revealed an increase to approximately 11 to 12, indicative of an alkaline state and the presence of cement. However, no significant strength development was observed. Measurement of Ec showed values from 1 to 4 mS/cm at the depth of 0 to 4 m and from 3 to 6 mS/cm at the depth of 4 to 13 m, suggesting a lower chemical processes within the specimen. However, high electrical conductivity alone does not guarantee increased strength. Additionally, the ignition loss (LOI) was notably high, ranging between 7 to 10%, indicating the presence of organic matter, which can hinder strength development in the improved column.

3 Materials and methods

3.1 Sample description

3.1.1 Commercial Kaolin clay

This study focuses on addressing the engineering challenges posed by the soft cohesive clay, such as Ariake clay and Hasuike clay, encountered in Saga lowland [1, 9]. Given the need for uniform soil samples with consistent quality in terms of basic geotechnical properties, a controlled and standardized experimental protocol was adopted, utilizing commercial Kaolin clay. The selection of commercial Kaolin clay is noteworthy for its widespread utilization in various geotechnical experiments and industrial applications [14].

The assessment involves parameters such as the liquidity index (IL), a critical factor in geotechnical engineering used for evaluating soil behavior under different loading conditions. Additionally, the natural water content (Wn) is considered, contributing to a comprehensive understanding of the soft cohesive clay's behavior.

Soils characterized by higher liquidity index (IL) values typically exhibit greater compressibility compared to those with lower IL values. Specifically, in the Saga lowland, clay types like Ariake and Hasuike clay demonstrate higher sensitivity and compressibility, displaying solid behavior when undisturbed but liquefying upon disturbance [1].

The IL is a crucial parameter used to assess the consistency and compressibility of soil. Mathematically, it is defined as the ratio of the difference between the natural water content of the soil and its plastic limit to its plasticity index (PI). This mathematical representation provides a quantitative measure that aids in controlling and standardizing an experimental protocol aiming to adjust the moisture content of the kaolin clay powder.

$${I}_{L}=\frac{{W}_{n}-PL}{PI}=\frac{75-28.8}{21.95} =2.1$$
(1)

Wn: Natural water content; PL: Plastic limit; PI: Plasticity index.

The kaolin powder was adjusted to 1.5 times the liquid limit (1.5LL) for the purpose of resembling approximately a natural environment deposit of clay soil in Saga lowland [15]. The liquidity index IL is used here for scaling the natural water content of a soil sample. By adding distilled water to the commercial Kaolin clay (Powder) and adjusting the water content to 1.5LL, we achieved a state resembling standard soft clay.

Table 1 outlines the fundamental characteristics of the commercial Kaolin clay (Powder) and the adjusted Commercial kaolin. Upon water content adjustment, parameters such as water content (Wn), Wet density (ρt), Dry density (ρd), Void ratio (e), Degree of saturation (Sr) and Liquidity index IL underwent significant changes.

Table 1 Basic soil properties of commercial Kaolin clay
Full size table

The arrangement of test results follows Mikasa’s classification of soil’s geotechnical engineering properties. The grain size distribution showed a clay particle content of 60%, which means a presence of more fines particles grains resulting in more possibility in chemical reactions occurring in the Kaolin clay as moisture infiltrates promoting cation–anion exchange on soil’s particles.

The particle density (ρs) for of 2747 kg/m3 for kaolin clay suggests that the clay particles are relatively dense or tightly packed within a given volume. A liquid limit of 50.75% for kaolin clay indicates its behavior regarding moisture content around each particle. It suggests that at this level of moisture content, the clay particle transforms from a plastic to a liquid state, meaning it becomes very soft and pliable. This indicates that kaolin clay has moderate to high plasticity and can undergo deformation when subjected to external forces. A wet density (ρt) of 652 kg/m3 suggests that kaolin clay has a relatively low wet density, meaning it is lightweight when saturated. But as water is added, the ρt increased to 1516 kg/m3 indicating that the kaolin clay has a higher compactness or is more densely packed when saturated, which can affect its permeability, compressibility, and other engineering characteristics.

A dry density of 648 kg/m3 represents the compactness of the clay particles in their natural state, without any added moisture. As water is added, the dry density changes to 866 kg/m3 suggesting that the kaolin clay is relatively dense and compacted, which can contribute to higher strength, stiffness, and reduced permeability. The current state of the kaolin soil with IL < 1, is reflected as powder state, after adjusting the water content to 75% the IL reached 2.1. A liquidity index of 2.1 suggests that the kaolin clay is highly plastic and may require special measures to manage deformation and achieve compaction during construction.

3.1.2 Type of binder

Table 2 shows the cement-based binder known as Ustabiler 10 (US10), specifically designed for the cohesive soil prevalent in the Saga lowland [9]. Developed to address challenging soil types and site conditions not easily solidified by conventional cement (Ordinary Portland Cement), as illustrated in Fig. 3, US10 offers distinct advantages with categorization of the cement-based binder within a framework, showing its relative position compared to other materials in terms of application, properties, or effectiveness. Table 3 outlines the chemical composition comparison between Ordinary Portland Cement and US10. US10 is particularly valued for its ability to promote flash setting, facilitating early strength development in cement slurry. This property is attributed to its elevated Sulfur trioxide (SO3) content, a key component of gypsum (Calcium Sulfate Dihydrate CaSO4·2H2O) [16]. Gypsum, an additive material in the clinker grinding process [17], contains water within its molecular structure and comprises approximately 23.3 percent Ca and 18.5 percent sulfur (S).US10 has less Si and more SO3 compared to ordinary Portland cement. Due to the higher content of ettringite in the cement-based binder, it incorporates soil moisture as crystalline water [9] as shown in Table 2.

Table 2 Test results of US-10 (provided by cement company)
Full size table
Fig. 3
figure 3

Categorization of the cement based binder, (recreated after [18])

Full size image
Table 3 Chemical composition of ordinary portland cement and cement-based binder
Full size table

The utilization of US10 in this research significantly amplifies the early strength, stiffness, and durability of soil, making it adaptable to a wide range of engineering endeavors spanning diverse soil compositions and environmental settings. Notably, its versatility and cost-effectiveness make US10 an appealing option for projects constrained by budget considerations, thereby fostering its extensive adoption in soil stabilization and ground improvement initiatives throughout Saga prefecture.

3.2 Specimens

3.2.1 Type of mold

To enhance the precision of chemical component measurements, small size specimen types were employed in this study, each serving a specific purpose to ensure rigorous comparability throughout the experimental process. The small Mold (SM), as depicted in Fig. 5, with a cylindrical shape, measuring 0.03 m × 0.03 m. Originally designed for a chair-cap under a chair's foot, its small size aligns with the standards of X-ray fluorescence sample cup measurements [7] as illustrated in Fig. 4. Specimen preparation poses challenges, as noted by Broton and Bhatty [8], emphasizing the importance of meticulous preparation for accurate quantitative XRF analysis. Crafted from flexible rubber, this mold facilitates the easy detachment of real specimens after the curing period, contributing to the efficiency of the experimental procedures.

Fig. 4
figure 4

Type of mold SM 0.03 m × 0.03 m

Full size image

3.2.2 Specimen’s preparation

The experimental approach employed in this investigation aimed to assess both the mechanical and chemical characteristics of commercial kaolin before and after mixing. Specifically, quNP for the SM, and X-ray fluorescence analysis were conducted both before and after improvement with the cement-based binder US10. The procedural steps involved in specimen preparation and testing adhered to the guidelines stipulated by the Japanese Geotechnical Society standards [19].

In Saga prefecture, due to the unique properties of clayey soils, construction design guidelines have been established in accordance with the recommendations of the Public Works Research Center [9, 20]. The application of the DMM in lowland areas of Japan follows empirical standards with a water-to-cement ratio (W/C) of 1.0. The cement dosage and the design standard strength (\({\overline{q} }_{u}\)) range from 50 to 150 kg/m3 and 500 to 1000 kN/m2, respectively [9, 20]. These parameters are empirically set to ensure the desired strength of the soil–cement column.

3.2.3 Pre-test measurement

Before the main experiment, an analysis of the powder state for each sample was performed using the standard X-ray fluorescence sample cup, illustrated in Fig. 5. The commercial kaolin powder, commercial kaolin powder + distilled water, cement-based binder (US10) powder, and cement-based binder (US10) powder + distilled water was subjected to handheld XRF analysis. Table 4 shows the sample and the weight (only sample) for the pre-test measurement. Each sample underwent thorough mixing for 10 min using a mixing bowl in Fig. 6, following the guidelines outlined in JGS 0821–2009 [19], to ensure proper homogeneity before being introduced into the molds.

Fig. 5
figure 5

Standards X-ray fluorescence sample cup (recreated after [7])

Full size image
Table 4 Samples pre-test measurement
Full size table
Fig. 6
figure 6

Mixing conditions: a 12 pieces of mold; b mixing bowl

Full size image

The mixing of US10 + Kaolin + Water corresponds to the conditions of the actual experiment. Specimens were measured during the pre-test phase and are regarded as the collected data for 0 day.

3.2.4 Actual experiments

To determine the required number of specimens and the percentage increase in preliminary tests, the volume of improved soil per batch (referred to as the volume of the mixing) is taken into consideration. Per batch basically obtains 12 specimens [21]as shown in Fig. 6.

Tables 5 and 6 present a comprehensive overview of the mixing conditions, with the wet density of the commercial kaolin sample determined as 1516 kg/m3. Calculations for the soil sample volume and the quantity of cement-based binder are conducted through Eqs. 1 and 2, respectively. The computation for the soil sample volume is expressed as follows:

Table 5 Mixing conditions
Full size table
Table 6 Mixing conditions based on the mass and W/C = 1
Full size table
$$\text{Soil sample required}=\text{Specimen volume V x Number of specimens x Wet density }({\uprho} \text{t})\text{ x Premium ratio}$$
(2)

Similarly, the calculation for cement-based binder is given by:

$$\text{Cement}-\text{based binder required }=\text{ Volume of specimen V x Number of specimens x Amount of cementitious solidifier x Premium ratio}$$
(3)

The premium ratio, set to 1.1, functions as a multiplier, typically a percentage, addressing potential variations, wastage, or other factors [22]. It is applied to the other factors in the equations, providing the total mass of soil required for the specified number of specimens, factoring in considerations like wet density and the applied premium ratio. This approach ensures a meticulous and thorough calculation accounting for potential variations and contingencies in the experimentation process. All the specimens were well moved to the constant temperature room after preparation as shown in Fig. 7.

Fig. 7
figure 7

Molded specimens inside constant temperature room

Full size image

3.3 Handheld XRF measurements

3.3.1 Handheld XRF specifications

The handheld XRF is an X-ray fluorescence analysis tool which measures the chemical components of a sample based on the atom’s excitation. This radiation's energy is characteristic of the atom present in the sample.

Figure 8 displays the S1 TITAN MODEL 800, a handheld XRF device manufactured by BRUKER. This model is equipped with a range of specifications, including excitation levels ranging from 6 to 50 kV and a 5-position automatic filter changer. The detector utilized is an SDD (graphene window), enabling the analysis of elements within the range of Magnesium (Mg) to Uranium (U). The spot size options for analysis are 8.5 or 3 mm. This particular device belongs to the 2nd generation of the S1 TITAN series and was produced in the year 2013 [23]. Figure 9 illustrates the X-ray fluorescence (XRF) process, where X-rays displace electrons from their atomic orbital positions, releasing characteristic energy bursts for specific elements [23, 24]. This energy release is detected by the XRF instrument's detector, which registers the signals and categorizes them by element. These signals are then collected in a multi-channel analyzer, sequentially handling each X-ray at high speed. The XRF machine's detector can manage 1 million counts per second, facilitating the determination of the sample's composition [23]. Extending the measurement time improves statistics, precision, and peak-to-background ratios, resulting in enhanced detection limits. Data was collected based on the percentage of each atom’s amount.

Fig. 8
figure 8

Handheld XRF MODEL: S1 TITAN MODEL 800: a Handheld XRF equipment; b X-ray tube and detector (from [23])

Full size image
Fig. 9
figure 9

X-rays displacement of electrons and signal generation (from [23])

Full size image

3.3.2 Measurements

Figure 10 illustrates the procedural aspects of the measurement process. Handheld XRF measurements were systematically conducted at various intervals, encompassing the powder state pre-mixing and subsequent post-mixing time points: 0 day, 1 day, 7 days and 28 days. The measurement duration for each specimen was set at 30 s, with 3 data points collected per measurement, resulting in a total measuring time of 90 s per specimen. During measurement, the handheld XRF device is positioned on a standing support, orienting the X-ray tube and detector upward. The surface of the targeted specimen is then placed on the Mylar thin film, serving as a thin layer between the X-ray tube, detector, and the measurement point. This configuration optimizes the efficiency of X-ray transmission and data collection.

Fig. 10
figure 10

Handheld XRF measuring of specimens

Full size image

In the measurement setup, a Mylar thin film, a polyester film frequently employed in X-ray fluorescence (XRF) applications, was utilized, as depicted in Fig. 10. Mylar serves as a window material for XRF detectors, facilitating the transmission of X-rays while protecting the detector from external influences. Its thin and transparent nature allows X-rays to pass through with minimal absorption, ensuring efficient elemental detection. Mylar was chosen based on its favorable characteristics, including transparency to X-rays, mechanical strength, and stability.

3.4 The needle penetration test (NPT)

NPT, originally developed in Japan by MARUTO Co., as an alternative for the indirect estimation of the UCS of soft rocks [25, 26], was employed to assess the strength development of the improved soil. This test, widely utilized for determining the compressive strength properties of cement-based binder mixtures [27], adhered to conventional and standardized equipment, with calculations following Japanese geotechnical standards [28].

In the NPT, the needle is inserted into the soil until it reaches 1 cm, at which point the load is measured., and the penetration depth is measured from the position of the presser on the penetration scale [29, 30] Uchida et al., (2004) explored the applicability of needle penetration tests on improved clay soil, determining that the unconfined compressive strength could be accurately estimated from the test results. This strength evaluation method accelerates and simplifies the evaluation process for DMM, making it more rational and efficient. It enables the design of suitable combinations and holds potential for future application across a broader range of soils targeted for improvement [30].

$$\begin{array}{l}quNP=41.8xNP-4\\NP\:=\:100\;\mathrm N/\mathrm{cm}\;\mathrm x\;\mathrm{depth}\;(1\;\mathrm{cm})\end{array}$$
(4)

Figure 11 illustrates the needle penetration test equipment used in this study, exclusively applied to the SM specimens. This equipment measures the material's resistance to the penetration of a standard needle, offering insights into its ability to withstand long-term stresses.

Fig. 11
figure 11

Needle Penetration test equipment (From [30])

Full size image

3.5 Scanning electron microscopy (SEM) on 0 day, 1 day, 7 days and 28 days

Samples were prepared based on the standards requirements of the SEM machine. The samples were meticulously prepared according to the specifications outlined by the SEM machine standards. SEM samples were carefully crafted into cubical shapes measuring 5 mm on each side. Freeze-drying was conducted over a period of 24 h, during which the weight of the sample was monitored at 5-h intervals until a constant mass was achieved. The sample was initially subjected to freezing in a nitrogen solution before being transferred to a freeze dryer, also known as a lyophilizer (Model FDU-1200 EYELA). Within the freeze dryer, the sample was maintained at a temperature of -30 °C while subjected to a vacuum pressure of 20 Pa. The lyophilizer is commonly used to remove moisture from samples before performing SEM. This process involves freezing the sample to very low temperatures and then subjecting it to vacuum conditions, allowing the frozen water to sublimate directly from solid to gas without passing through the liquid phase. Freeze drying is preferred for delicate samples that may be damaged by traditional drying methods, as it helps preserve their structure and integrity. The EPMA-1720 Electron Probe Microanalysis was used for the SEM.

3.6 Data analysis

In the data analysis phase, Microsoft Excel serves as our primary tool for assessing the data and generating graphs. For the handheld XRF measurements, meticulous preparation is crucial, involving the creation of three samples for each specimen to ensure data accuracy. This tripartite approach enables a thorough evaluation of precision, facilitating the determination of an average value. We also consider the coefficient of variation, a statistical measure that helps assess the consistency and reliability of the obtained results. By incorporating this measure, we enhance the overall robustness of our experimental outcomes.

4 Results

4.1 Handheld XRF measurements

4.1.1 Ca tendencies

Figure 12 presents the handheld XRF measurements conducted on Small Mold (SM) specimen, capturing observations from the initial powder state to subsequent stages post-mixing. Through an analysis of chemical constituents, notable findings were obtained, shedding light on how the specimens evolved after mixing in comparison to their original powder state characteristics.

Fig. 12
figure 12

Specimens behaviors based on Ca over time

Full size image

Notably, the calcium (Ca) composition displayed significant disparities across various specimen compositions. Initial assessments revealed negligible Ca content (0%) for both Kaolin samples, contrasting with higher Ca levels of 51.47% for Cement.

Furthermore, the combination of Cement + Kaolin yielded an intermediary Ca content of approximately 20%. However, subsequent mixing with water (cement + kaolin + water) led to a considerable reduction in Ca content, drop** to less than 10% at 0 day. Nonetheless, with extended curing time, the Ca content exhibited an upward trajectory, escalating to 20% at 1 day, 33% at 7 days, and reaching a peak of 46% at 28 days. These suggest a direct correlation between prolonged curing duration and an apparent increase in Ca content, indicating active involvement in the hydration process. Additionally, there is a discernible tendency for the Ca content to revert towards its original powder state composition, approximating 51.47%. This phenomenon highlights the dynamic interaction between hydration kinetics and specimen composition.

4.1.2 Si tendencies

Figure 13 illustrates the handheld XRF measurements conducted on SM specimen over time, spanning from the powder state to post-mixing stages. Following a thorough analysis of chemical components, the resulting observations shed light on specimen behaviors post-mixing, with reference to their original powder state, as previously discussed in the Ca tendencies section.

Fig. 13
figure 13

Specimens behaviors based on Si over time

Full size image

Initial Si compositions varied across different specimen compositions. Kaolin clay sample exhibited Si content around 78.4%, contrasting with substantially lower Si content of 26% for Cement. Moreover, the combination of cement + kaolin yielded a Si content of approximately 59.1%. However, subsequent mixing with water (cement + kaolin + water) resulted in a noticeable an apparent increase to 75% at 0 day for Si. As curing time progressed, Si content exhibited a decreasing trend, reaching 62.8% at 1 day, 50.9% at 7 days, and 40% at 28 days for SM specimens.

This observed trend suggests that with prolonged curing time, the Si content decreases apparently in parallel with the hydration process. However, unlike the tendencies observed for Ca, the Si content does not revert to its original powder state composition.

4.1.3 S tendencies

US10 is highly valued for its ability to accelerate setting and enhance early strength in cement slurry. This feature is linked to its high amount of Sulfur trioxide (SO3), a key constituent of gypsum (Calcium Sulfate Dihydrate CaSO4·2 H2O) [16].

Figure 14 presents the handheld XRF measurements for both SM specimens, delineating observations from their powder states to post-mixing stages. In terms of sulfur (S) composition, initial readings indicated 0% S content for kaolin, contrasting with higher S content of 7.6% for cement. The combination of cement + kaolin yielded an intermediate S content of 2%. However, after mixing with water (cement + kaolin + water), a notable decrease to less than 1% was observed at 0 days. As curing time progressed, further reductions in S content were evident, reaching 0.7% at 1 day, 0.5% at 7 days), and eventually declining to 0% at 28 days for SM specimens.

Fig. 14
figure 14

Specimens behaviors based on S over time

Full size image

This trend suggests that with prolonged curing time, S’s apparent content decreases in tandem with the hydration process. Notably, there is no noticeable return of S content to its original powder state.

4.2 Needle penetration strength quNP on 1 day, 7 days and 28 days

For a comprehensive interpretation aligned with the compressive strength of each specimen, the converted unconfined compressive strength quNP (kN/m2) derived from the needle penetration test were analyzed.

Figures 15 and 16 illustrate the strength tendencies observed through the needle penetration test. At each curing interval, three specimens were evaluated, and the average value was determined, considering the coefficient of variation (CV) for each quNP. For SM, the average based on CV values were 8% at 1 day, 10% at 7 days, and 9% at 28 days.

Fig. 15
figure 15

quNP of each specimen

Full size image
Fig. 16
figure 16

Average quNPA per curing time

Full size image

As depicted in Fig. 17, the average strength (quNPA) development trends for were as follows: 144.10 kN/m2 at 1 day, 238.65 kN/m2 at 7 days, and 301.32 kN/m2 at 28 days.

Fig. 17
figure 17

Identification of hydrated products based on morphology and appearance: a identification of (Ca(OH) 2 at 8.3mm resolution; b identification of (Ca(OH) 2 and C S H at 8.7mm resolution; c identification of (Ca(OH) 2 and C S H at 9.8mm resolution; d identification of (Ca(OH) 2 and C S H at 9.1mm resolution. (modified after [ [31])

Full size image

4.3 Scanning electron microscopy (SEM) for 1 day, 7 days and 28 days

SEM Analysis was employed to investigate the microstructural characteristics and identification of hydrated products over time of specimens at different stages of curing: 1 day, 7 days, and 28 days. SEM allows for high-resolution imaging of the surface morphology and microstructure of materials, providing a better understanding into the formation and distribution of hydrated products. Figure 17 shows the identified hydrated products resulting from the use of nano calcium-silicate-hydrate modified by polycarboxylate superplasticizer, called nano C-S–H-PCE cementitious materials to promote the rapid development of compressive strength at early age [31]. The morphology of CH in Fig. 17 suggests CH as a crystallization in the hexagonal-rhombohedral system, typically forming hexagonal plates or short prisms with perfect basal cleavage. The atomic structure of C-S–H gel is like capillary pores or the naturally occurring mineral tobermorite [32].

We examine the three primary hydrated products typically generated during the pozzolanic reaction in cementitious materials: ettringite, (Ca(OH)2) or CH and C-S–H gel, and. Figure 18 provides SEM assessment images of kaolin and cement powder at a resolution of 10 µm, offering understanding into the initial state of each material before mixing.

Fig. 18
figure 18

Assessment of kaolin and cement powder through SEM at 10µm resolution: a Kaolin powder; b Cement powder

Full size image

Figure 19a and b, present the 1 day state of the specimen. During the early stages of cement hydration, ettringite typically forms first, and the formation occurs through a reaction between calcium aluminate phases in cement and sulfate ions, particularly in the presence of excess gypsum (CaSO4·2H2O). Its crystals appear as needle-like or plate-like structures, and their presence indicates the early stages of hydration. Alongside ettringite formation at 1 day, the presence of kaolin particles and calcium hydroxide (Ca(OH)2) or CH is observed. Kaolin particles are visible as aggregate shapes with a white-grayish appearance, consistent with their appearance after 1 day of curing. However, there was no significant presence of C-S–H observed.

Fig. 19
figure 19

Microstructural and identification of hydrated products over time: a 1day specimen at 4µm; b 1day specimen at 2µm; c 7days specimen at 4µm; d 7days specimen at 2µm; e 28days specimen at 4µm; f 28days specimen at 2µm

Full size image

In Fig. 19c and d, the specimen's state at 7 days is depicted. Here, the presence of calcium hydroxide (CH) is more pronounced, appearing as grayish hexagonal plates, while the formation of ettringite slows down. Additionally, a moderate appearance of capillary pores resembling C-S–H gel or the naturally occurring mineral tobermorite [32] is observed. C-S–H gel forms continuously throughout the hydration process and becomes the predominant hydration product in the specimen by 28 days. At 28 days, as shown in Fig. 19e and f, the abundance of CH in hexagonal plates and C-S–H gel capillary pores suggests the formation of properly hydrated products.

5 Discussions

5.1 Handheld XRF measurements and quNP A on 1 day, 7 days and 28 days

In this study, laboratory experiments were conducted to investigate the behaviors of calcium (Ca), silicate (Si), and sulfur (S), which are key elements in the formation of hydrated products. This research aims to comprehensively understand how the behaviors of Ca, Si, and S correlate with the development of strength throughout the experimental process. The presence of water-initiated hydration reactions in the US10, leading to the formation of ettringite, (Ca(OH)2) or CH and C-S–H gel. Some calcium ions from the US10 reacted with water to form calcium hydroxide, while others reacted with silicate ions from the cement and Silicate from the Kaolin clay to form C-S–H gel. Additionally, the higher amount of sulfate in the cement led to the formation of ettringite.

Figure 20 illustrates the overall trends of these chemical components with their correlations alongside quNPA. After mixing water, cement, and kaolin clay on 1 day, the resulting chemical composition showed a calcium content Ca of 20.0% and Si of 62.8%. the early strength assessment revealed a compressive strength quNPA = 144.10 kN/m2 indicating the initial development of strength in the specimen. During the 7 days of curing, Ca showed an apparent increase from 20.0% to 33.6%, while Si noticeably decreased from 62.8% to 50.9%. These changes suggest continued hydration reactions, alongside the increasing of quNPA from 144.10 kN/m2 to 239.65 kN/m2 at 7 days.The quNPA.

Fig. 20
figure 20

Correlations between Ca, Si for XRF and quNPA

Full size image

After 28 days of curing time, the noticeable increase in Ca content reached 46.2% indicating ongoing hydration reactions. More calcium ions from the cement continue to react with water and silicate ions, contributing to the formation of additional ettringite, (Ca(OH)2) or CH and C-S–H gel. The apparent decrease in Si content from 62% to 40.9% suggests its incorporation into the C-S–H gel structure. As the hydration process progresses, Silicate from the Kaolin clay and the cement reacts with calcium ions to form C-S–H gel. Therefore, the apparent amount of Silicate may stabilize or decrease slightly as hydration approaches completion. The compressive strength (qu) further increased for quNPA 239.65 kN/m2 at 7 days to quNPA = 301.32 kN/m2 at 28 days, indicating significant strength development over the curing period.

As a general trend, the apparent amount of Ca increases proportionally with curing time, coinciding with the rise in quNPA. The noticeable escalation of Ca from 20.0% at 1 day to 46.2% on 28 days implies ongoing hydration reactions, wherein Ca ions from cement compounds react with water to generate ettringite, (Ca(OH)2) or CH and C-S–H gel. In the presence of sulfate ions, calcium aluminate phases in cement can react to form ettringite. Ettringite crystals can incorporate calcium ions(Calcium Sulfate Dihydrate CaSO4·2H2O) [16], further increasing the calcium amount in the specimen. This process significantly contributes to strengthening and binding [33] the improved kaolin clay within the soil–cement mixture over time.

Calcium Hydroxide (CH), also known as portlandite, is another hydration product that forms when calcium ions from cement react with water. Typically, it crystallizes into large formations within the pore structure of the concrete [34]. The presence of CH contributes to the gradual rise in calcium amount over time. Furthermore, the calcium hydroxide released during cement hydration fills the interstitial pores within the concrete structure [34], it essentially occupies the empty spaces between the solid particles. As shown in Fig. 21, these filled pores reduce the overall connectivity between the pores, meaning there are fewer pathways for liquids or gases to flow through the improved soil. This reduction in pore connectivity leads to a decrease in the overall porosity of the concrete microstructure [34]. In simpler terms, the presence of CH helps to fill in the gaps within the concrete, making it denser and less porous.

Fig. 21
figure 21

Schematic diagram of particles in specimen after Curing times

Full size image

However, the apparent decrease in Si’s amount from around 62.8% on 1 day to 40.9% on 28 days and a rise in quNPA over time, suggests its integration into hydration products and secondary minerals formed during cement hydration, such as C-S–H gel. Silicate plays a crucial role in the formation of C-S–H gel, which is the dominant binding phase in cement hydrates [35]. As hydration progresses, Si is consumed in the formation of C-S–H gel, leading to a reduction in its amount over time as shown in Fig. 22.

Fig. 22
figure 22

Schematic structure of crystalline C-S–H (Si consumed for C-S–H gel’s formation) (from [10])

Full size image

According to Cho et al., (2020), to model C-S–H structures effectively, it is essential to account for the variability in chemical composition and its distribution within a cement matrix. This is crucial because the engineering performance of cement hydrates is significantly influenced by the structures and chemical compositions of C-S–H, which are determined by factors such as the w/c ratio and curing conditions [10].

Figure 22 depicts a schematic diagram of the C-S–H crystalline structure [10]. These chains are arranged in a repeating pattern at intervals of three SiO4 tetrahedra. In this arrangement, two adjacent tetrahedra are coordinated to the Ca2+ ions in the layer, while the third tetrahedron bridges two consecutive dimers, forming what is known as bridging tetrahedra [10]. This shows how Si is significantly consumed in the formation of C-S–H gel, leading to a reduction in its amount.

5.2 Handheld XRF measurements and SEM on 1 day, 7 days and 28 days

Figure 23 shows the validation of formed hydrated products during the hydration process over time, we identified crystals that were formed through the SEM imagine based on the morphology of products such as ettringite, (Ca(OH)2) or CH and C-S–H gel. These hydration products contribute to the densification and strengthening of the concrete matrix over time, the formation of these crystalline phases, ultimately leading to improved mechanical properties of the concrete [33].

Fig. 23
figure 23

Correlations between handheld XRF measurements and SEM : a assessment based on 100% of atoms composition; b, c, d SEM on 1day, 7days and 28days respectively, for Ett, CH and CSH appreciation

Full size image

At 1 day, ettringite (Ett) formation is evident, accompanied by a minimal presence of CH and no C-S–H gel. This transformation of Ett is attributed to the apparent decrease in sulfate (S) content from 7.6% in the powder state to less than 1% on 1 day, 7 days, and 28 days. By 7 days, with reduced ettringite formation, the emergence of CH contributes to a gradual apparent increase in calcium amount over time. This decline in pore connectivity results in decreased overall porosity of the concrete microstructure. In simpler terms, CH helps fill gaps within the concrete, making it denser and less porous [34]. The limited presence of C-S–H at 7 days validates its initial formation, correlating with the declining Si amount. By 28 days, (Ca(OH)2) or CH and C-S–H gel are observed in almost equal amounts, explaining the converging tendencies of Ca and Si amount to around 45%. Table 7 shows the overall trends of measured atoms by XRF and Hydrated Products over time.

Table 7 Overall apparent trends of measured atoms by XRF and hydrated products over time. (Apparent increasing: ⇑;Apparent decreasing: ⇓)
Full size table

This application of the handheld surpasses the conventional use of phenolphthalein tracer, offering enhanced precision in evaluating the strength of improved soil based on the key elements from this research.

5.3 Limitations

In this study, Handheld XRF measurements have interesting results; however, it is important to acknowledge their limitations. While it provides valuable understanding into the qualitative amount of chemical components present in the specimen, this tool inherently lacks the precision required for quantitative analysis. Consequently, this research focuses on elucidating the apparent tendencies of the chemical components over time.

Additionally, a notable challenge emerged concerning the size of the specimens. To facilitate strength assessment and enhance precision in measuring the chemical components using handheld XRF, it became evident that the specimen size must be minimized as much as possible. This ensures that the handheld XRF device can effectively capture and analyze the chemical composition of the specimen with greater accuracy and reliability.

6 Conclusions

This study presents a preliminary exploration into substituting the traditional phenolphthalein tracer method with Handheld XRF for quality assurance, addressing the limitations of the former in assessing compressive strength. Standardized parameters, including a cement dosage of 110 kg/m3, a W/C ratio of 1, and a 10-min mixing time, were employed. Handheld XRF measurements were conducted at various stages: at the powder state before mixing, and at 0 day, 1 day, 7 days, and 28 days post-mixing. This analysis aimed to investigate the behavior of Si, Ca, and S elements, which influence the formation of hydrated products crucial for the durability and strength development of improved soil.

  1. (1)

    A significant increase in the apparent amount of Ca from 1 to 28 days indicating its integration into ettringite, (Ca(OH)2) or CH, C-S–H gel, corresponding with the rise in quNPA.

  2. (2)

    A noticeable decrease in Si from 1 to 28 days, coupled with a rise in quNPA, suggesting its integration into C-S–H gel formation.

  3. (3)

    The quNPA significantly increased from 1 to 28 days, corresponding to the observed changes in Si, Ca, and S measured by the handheld XRF.

  4. (4)

    This research suggests the potential use of Handheld XRF as a sustainable substitute for the phenolphthalein tracer method in field-quality assurance relying on the observed atoms (Si, Ca, and S) at standard measurement intervals (1 day,7 days and 28 days) is possible for improved soil in geotechnical engineering.