Effects of Ion-Regulated Mechanisms on Calcite Precipitation in the Enzyme-Induced Carbonate Precipitation Treatment of Loess

Abstract

This study examines the effects and mechanisms of different Enzyme-Induced Carbonate Precipitation (EICP) treatments on loess structure improvement. The study focuses on ordinary EICP and three modified methods using MgCl₂, NH₄Cl, and CaCl₂. A series of unconfined compressive strength (UCS) tests, scanning electron microscopy (SEM), energy-dispersive spectroscopy (EDS), and elemental mapping were used to assess both macroscopic performance and microscopic characteristics. The results indicate that ordinary EICP significantly enhances loess particle bonding by promoting calcite precipitation. MgCl₂-modified EICP achieves the highest UCS (820 kPa) due to delayed urea hydrolysis and the formation of aragonite alongside calcite, which results in stronger and more continuous cementation. In contrast, NH₄Cl reduces urease activity and reverses the reaction, which limits carbonate precipitation and weakens structural cohesion. Excessive CaCl₂ leads to a “hijacking mechanism” where hydroxide ions form Ca(OH)₂, restricting carbonate formation and diminishing the overall enhancement. This study highlights the mechanisms behind enhancement, degradation, and diversion in the EICP process. It also provides theoretical support for optimizing loess subgrade reinforcement. However, challenges such as uneven permeability, environmental variability, and long-term durability must be addressed before field-scale applications can be realized, necessitating further research.

1. Introduction

Loess is an aeolian sediment formed in arid climates, predominantly found in Northwestern (NW) China [1,2,3]. Its main mineral components are quartz, feldspar, and mica. Although loess resources are abundant in China, their use in engineering is limited due to poor geotechnical properties such as loose structure, low strength, fine particle size, and high compressibility [4,5]. Untreated loess is highly susceptible to collapse when exposed to water, leading to a significant loss of soil structure integrity [6]. With rapid infrastructure development in loess-rich areas, there is an increasing need for effective soil reinforcement to meet modern construction standards [7,8]. Traditional loess stabilization techniques mainly rely on physical and mechanical methods, which can cause environmental pollution, disturb surrounding soils, and incur high costs [9,10,11,12,13,14]. Therefore, there is an urgent need for an environmentally sustainable, efficient, and cost-effective method for loess subgrade improvement [15].

Microbially induced carbonate precipitation (MICP) has emerged as a promising method for soil reinforcement, gaining increasing global research attention [16,17,18]. The process works through microbial metabolism, which produces urease, an enzyme that catalyzes the hydrolysis of urea [19,20]. The resulting carbonate ions then react with calcium ions to form poorly soluble calcium carbonate crystals with cementitious properties. These crystals fill interparticle voids and bind soil grains, improving the soil’s mechanical integrity [21,22]. In 1973, Boquet et al. [23] first documented microbial biomineralization in natural environments such as lakes, caves, and bones. Later, Whiffin et al. [24] formally introduced this process as a soil improvement technique, coining the term MICP. Van Paassen et al. [25] demonstrated the feasibility of MICP for large-scale ground improvement by using Sporosarcina pasteurii in a grouting procedure to reinforce sandy foundations. Similarly, Zhang et al. [26] applied MICP to cohesive soils, achieving significant strength enhancement. Their results showed that higher cementation solution concentrations improved bonding, while elevated pH values enhanced soil cohesion. Despite its effectiveness, MICP faces two main challenges that limit its broader application [27,28,29]. First, urease-producing bacteria are relatively large and require significant amounts of oxygen, restricting their use to coarse-grained soils like sand and limiting the depth of reinforcement. Second, cultivating these microbes demands strict environmental control, including sterility, temperature regulation, and pH stability, which are difficult to maintain in the field. To address these challenges, researchers have found that urease can also be extracted from certain plant seeds [30,31,32]. This has led to the development of Enzyme-Induced Carbonate Precipitation (EICP), which uses plant-derived urease as an alternative to microbial cultivation, making it a more feasible solution for practical soil improvement [33,34].

EICP technology uses small free urease enzymes to induce calcium carbonate precipitation, allowing it to penetrate finer-grained soils while minimizing the risk of clogging [35,36,37]. This makes EICP especially effective for reinforcing fine-grained soils, such as low-permeability loess. As a new soil improvement method, EICP has gained significant attention from both researchers and engineers. Zhang et al. [38] combined EICP with lignin, a naturally occurring biopolymer derived from plant cell walls, to stabilize silty soils. Lignin’s unique properties, including its ability to enhance soil cohesion and improve structural integrity, contributed to significant improvements in soil strength and wind erosion resistance. This combination of EICP and lignin demonstrates the potential of EICP in reinforcing fine-grained soils, such as loess, by improving both their mechanical properties and resistance to environmental degradation. Yuan et al. [39] used EICP to treat soils with varying initial dry densities and conducted dynamic triaxial tests. The results showed that the treated samples exhibited lower dynamic strain under equivalent cyclic stress, indicating a notable improvement in soil strength after EICP treatment. Zhang et al. [40] evaluated EICP for Pb-contaminated loess without adding an external calcium source, relying on in-situ ionic equilibria. Shen et al. [41] reinforced loess slopes and improved erosion resistance using EICP. Jia et al. [42] stabilized coal fly ash via urease-induced CaCO₃. Yuan et al. [39] quantified gains in sandy soils and underscored the role of particle size and pore networks. Jia et al. [43] compared polymer-cured and bio-cured fly ash and highlighted trade-offs. In EICP-based soil stabilization, calcium chloride and calcium acetate are commonly used as calcium sources [44]. These salts provide abundant Ca²⁺ ions that facilitate efficient calcite precipitation. However, excessive concentrations of these salts can negatively affect the soil matrix, vegetation, and groundwater quality. As environmental concerns grow, traditional ground improvement methods are increasingly unable to meet engineering demands from a sustainable development perspective.

The morphology and mineral composition of calcium carbonate crystals formed through the EICP process are influenced by multiple factors, resulting in a complex precipitation mechanism. Hammes et al. [44] isolated bacterial strains from various environmental samples to catalyze urea hydrolysis for calcium carbonate formation. They found considerable diversity in urease genes among the strains, which they attributed to the morphological differences in the resulting crystals, caused by variations in urease expression and its interaction with calcium ions. Zhang et al. [26] examined the effects of three calcium sources, namely CaCl₂, Ca(CH₃COO)₂, and Ca(NO₃)₂, on the compressive strength, tensile strength, and water absorption of microbial mortar. Their findings revealed that calcium acetate effectively promoted the formation of calcite, vaterite, and aragonite, thereby reducing the risk of chloride-induced corrosion in steel reinforcement. Abo-El-Enein et al. [45] explored the performance of these same calcium sources at a 1 mol/L concentration, combined with urea and bacteria, for sand stabilization. Their results showed that calcium chloride produced higher crystallinity, greater precipitation yield, and stronger cementation, suggesting that it was the most effective calcium source among the three. While many existing studies on MICP technology focus on materials like mortar and sandy soils that require immediate stabilization, providing insights into optimal treatment protocols and reinforcement strategies, research on EICP is still limited. There is a lack of focus on the precipitation mechanisms and the processes controlling its strengthening and degradation behavior. Furthermore, studies on the application of EICP for loess subgrade reinforcement are particularly scarce.

Building on the foundational work of Putra et al. [46,47], Almajed et al. [30], and Carmona et al. [37], this study explores the optimal substrate concentration for the EICP process through a series of test tube experiments. The enhancement mechanism of carbonate precipitation during the EICP process has been widely studied, with several investigations indicating that magnesium ions play a critical role in boosting the precipitation rate [48,49]. However, the biochemical reaction involved in the EICP process is highly complex, with various factors, such as calcium and magnesium concentrations, temperature, pH, and the presence of specific ions, influencing the process. While numerous studies have focused on improving the shear strength of sandy soils using EICP, limited research has addressed the mechanisms affecting carbonate precipitation, especially in finer-grained soils such as loess. This study seeks to bridge this gap by investigating the effect of different salt concentrations (MgCl₂, CaCl₂, and NH₄Cl) on the enhancement and degradation of carbonate precipitation, particularly in the context of loess stabilization. By exploring the optimal conditions for the EICP process and the precipitation mechanisms, this research contributes to a deeper understanding of EICP’s potential for improving loess subgrades and further promotes its practical application. The optimized formulations were then applied to loess specimens to assess improvements in subgrade strength and stability, facilitating the practical application and wider adoption of EICP technology. This study further examines the impact of EICP treatment on the engineering performance of loess subgrades, providing valuable experimental data and theoretical insights to support future field applications and the ongoing development of this technology.

This study advances EICP-based stabilization of fine-grained loess by resolving how specific ions regulate precipitation pathways and, in turn, mechanical performance. We systematically and separately vary the concentrations of MgCl₂, CaCl₂, and NH₄Cl to delineate additive-specific enhancement and degradation regimes, showing that Mg²⁺ moderates kinetics and promotes aragonite formation, yielding the largest strength gains. Collectively, these results offer a chemistry-first design strategy for EICP in low-permeability loess and distinguish this work from prior studies that emphasize sandy soils and short-term strength improvements without clarifying precipitation mechanisms in fine-grained matrices. While this study resolves short-term, ion-specific controls on carbonate precipitation and early-age strength in fine-grained loess, it does not yet evaluate long-term durability. Because design decisions for subgrade applications hinge on performance under environmental stressors, we outline a durability program comprising multi-cycle freeze–thaw exposure and controlled pH excursions. Retained UCS, stiffness degradation, permeability, carbonate yield, and microstructural/phase evolution (SEM/XRD/FTIR) will be quantified. The present paper establishes the mechanistic basis and practical dosing windows; a companion study will report durability and serviceability outcomes.

2. Materials and Methods

In this study, MgCl₂, CaCl₂, and NH₄Cl were chosen based on their ability to provide essential ions for the EICP process. Magnesium ions, in particular, are known to play a crucial role in elevating carbonate precipitation during the reaction, as highlighted in previous studies. The selection of these salts aimed to optimize the precipitation process by ensuring a balance of calcium and magnesium ions while avoiding concentrations that could adversely affect the soil matrix or groundwater quality. Concentrations of the salts were carefully determined through preliminary tests to achieve the most efficient precipitation without causing clogging or excessive ion concentration. To investigate the performance and underlying mechanisms of EICP for loess reinforcement, a comprehensive experimental program was developed, as shown in Figure 1. The methodology consisted of two primary stages: ordinary EICP, modified EICP, and their application in loess treatment tests. In Step 1, a baseline EICP process ((a) ordinary EICP) was carried out to identify the optimal concentrations of urea and calcium chloride under standard conditions. Building on these findings, Step 2 investigated the effects of elevated concentrations of specific additives, namely magnesium chloride ((b) MgCl₂), ammonium chloride ((c) NH₄Cl), and calcium chloride ((d) CaCl₂), on carbonate precipitation efficiency and crystal morphology, examining each additive separately. These experiments aimed to understand the enhancement or inhibition effects caused by each ionic species. Following the solution-level investigation, Step 3 involved applying both the standard and modified EICP solutions to loess specimens to assess their effectiveness in improving soil structure and strength. In Step 4, the treated samples were analyzed using scanning electron microscopy (SEM) and energy-dispersive spectroscopy (EDS) to characterize the microstructural evolution and mineral composition resulting from calcium carbonate precipitation. Finally, Step 5 included unconfined compressive strength (UCS) tests to quantify the mechanical improvements achieved through the various EICP treatments. This sequential methodology enabled a thorough exploration of EICP reaction behavior under controlled conditions and provided valuable insights into its practical application for enhancing the engineering properties of loess. The following sections present the detailed results from the test tube experiments and their extension to loess-based reinforcement.

2.1. Test Tube Experiments

The materials used in the test tube experiments included deionized water, urea with a purity of 99%, calcium chloride (CaCl₂) with a purity of 96%, and magnesium chloride (MgCl₂) and ammonium chloride (NH₄Cl), both with a purity of 98%. The urease enzyme used to catalyze urea hydrolysis had an enzymatic activity of 1.1 U/mg. In the first step, CaCl₂ was added to the prepared urea solution, and urease was introduced to catalyze the hydrolysis of urea, leading to the precipitation of carbonate. This method is referred to as the conventional EICP method, as shown in Figure 2 [34]. Additionally, the effects of MgCl₂, NH₄Cl, and varying CaCl₂ concentrations were considered in the experiments. The mass of CaCO₃ precipitated under both conventional and modified EICP conditions was measured after 72 h. The precipitation ratio (PR) of CaCO₃ was compared across treatments to quantify the efficiency of carbonate precipitation [46,47].

Urea at a concentration of 1 mol/L is hydrolyzed by urease to produce carbonate ions (as shown in Equations (1) and (2)), which then react with calcium ions released from the hydrolysis of 1 mol/L CaCl₂ (Equation (3)) to form calcium carbonate precipitates.

CH₂N₂O + 4H₂O→H₂CO₃ + 2NH₄⁺ + 2OH⁻

(1)

HCO₃⁻ + OH⁻→CO₃²⁻ + H₂O

(2)

Ca²⁺ + CO₃²⁻→CaCO₃

(3)

Yuan et al. [48] conducted biomineralization experiments with a fixed CaCl₂ concentration of 1 mol/L. Their results demonstrated that the amount of CaCO₃ precipitate increased with rising urea concentration but plateaued once the molar ratio of CaCl₂ to urea reached 1:1. Beyond this point, further increases in urea concentration resulted in only marginal gains in precipitation. Similar trends have been observed by Putra et al. [46,47,49], Almajed et al. [30], and Carmona et al. [37]. Based on these findings, this study used synchronous variation in substrate concentrations when investigating the effects of substrate levels. Specifically, when examining the influence of NH₄⁺ concentration, CaCl₂ and urea concentrations were varied simultaneously while maintaining a 1:1 molar ratio to maximize the potential for CaCO₃ precipitation. This approach also allowed for differentiation between precipitation reductions caused by calcium ion deficiency and those caused by the inhibitory effects of ammonium ions on the biomineralization process. For the investigation of Mg²⁺ concentration effects, the urea concentration was kept constant, while the combined concentration of CaCl₂ and MgCl₂ was adjusted to equal the urea concentration. In this case, increasing Mg²⁺ concentrations reduced Ca²⁺ concentrations, thereby highlighting the role of Mg²⁺ in promoting or inhibiting calcium carbonate precipitation.

The detailed experimental setup for the test tube experiments is presented in Table S1. The experimental procedure, as shown in Figure 2, involved the following steps: First, the dry weight of each test tube was measured. Then, a fixed volume of deionized water was added, followed by the sequential addition of urea, CaCl₂, and either MgCl₂ or NH₄Cl, according to the experimental design (see Table S1). Each reagent was fully dissolved with stirring before the next was introduced. Finally, urease was added to initiate the hydrolysis reaction. After all reagents were added, the test tubes were placed in an incubator and maintained at a constant temperature for 12 h. Following incubation, the resulting precipitate and suspension were separated, filtered, dried, and weighed. Specifically, after each reaction, the slurry was centrifuged and the supernatant retained for chemical analysis. The solid fraction was rinsed three times with ethanol to remove soluble salts without dissolving carbonate, then dried to constant mass at 60 °C and weighed. Blank controls were processed identically and subtracted. The mass of CaCO₃ precipitated was thus obtained gravimetrically. The precipitation ratio (PR) was calculated using Equation (4).

PR = (Actural precipitation mass/Theoretical precipitation mass) × 100

(4)

Additionally, pH and electrical conductivity (EC) were continuously monitored during the first 70 min of the reaction using a pH meter and conductivity meter, respectively.

2.2. Soil Specimen Preparation

The soil used in this study was a disturbed loess sample collected from the surface layer of a representative site on the Loess Plateau. The Loess Plateau, located in NW China, is one of the most significant geographical features in Asia. It is a vast, semi-arid region characterized by thick layers of loess, a wind-deposited silt. The plateau spans over 600,000 square kilometers and covers parts of several provinces, including Shaanxi, Shanxi, Gansu, and Ningxia. The region is known for its unique landscape, with deep gully systems and steep slopes, which have resulted from centuries of erosion due to rainfall and wind. The Loess Plateau has been heavily affected by soil erosion, which has significantly influenced its agricultural practices and local development. It is also an area with substantial cultural and historical significance, being the cradle of ancient Chinese civilizations. The Loess Plateau plays a crucial role in understanding soil properties, erosion mechanisms, and land reclamation strategies. The loess samples used in this study were collected from a slope near Shiziyuan Village in Xi’an, Shaanxi Province, at depths of 3.0–4.5 m. After sampling, the soil was air-dried at room temperature, and organic matter such as plant roots and debris was removed. The dried soil was then pulverized and passed through a 2 mm sieve to obtain a homogeneous sample suitable for laboratory testing. Basic physical properties of the soil were determined in accordance with the Chinese Standard for Geotechnical Testing Methods (GB/T 50123-2019) [50]. The measured parameters included natural water content, liquid limit, plastic limit, specific gravity, maximum dry density, and optimum moisture content. These results are summarized in Table S2. The particle size distribution curve (Figure 3a) shows that the loess consists of 3.3% sand, 87.4% silt, and 9.3% clay, indicating a silt-dominated composition. The median particle size (D50) is approximately 0.026 mm. The measured liquid limit and plasticity index are 31.6% and 12.1%, respectively (Figure 3b). According to the Unified Soil Classification System (USCS) [51], the loess is classified as low-plasticity clay (CL).

2.3. Experimental Procedure for Loess Treatment Using the EICP Method

This study examines the early-age mechanical response to ion-specific EICP treatments. Specimens were cured at a controlled moisture and temperature and tested in unconfined compression. No freeze–thaw or chemical exposure cycles were performed. To evaluate the feasibility of the EICP technique for loess reinforcement, a series of laboratory experiments was conducted, following the procedure shown in Figure 4.

Loess samples were first air-dried, crushed, and sieved through a 2 mm mesh. A 2000 g mass of the prepared soil was measured, and the required quantities of dry soil, deionized water, calcium sources, urease, and urea were calculated to formulate the EICP solution. The pH of the solution was adjusted to 8. The dry soil was thoroughly mixed with the prepared solution in a non-absorbent mixing tray and then compacted into cylindrical molds (Φ39.1 mm × 80 mm), using a triaxial compaction hammer. Each specimen was compacted in three layers, with the surface of each layer roughened before adding the next layer to ensure good interlayer bonding. The compacted specimens were cured in a humidity-controlled chamber for 4 days before EICP grouting. After curing, the specimens were oven-dried at 80 °C to remove any residual moisture. To promote reagent infiltration and ensure uniform distribution, vacuum saturation was applied. A stable vacuum was maintained in a sealed chamber. The EICP solution, containing urea, calcium chloride, and urease at concentrations optimized in preliminary test tube experiments, was infused at a controlled rate until the loess matrix was fully saturated. After injection, the samples were sealed and placed in a constant temperature chamber at 30 °C for 72 h to allow sufficient time for urea hydrolysis and calcium carbonate precipitation.

After curing, the specimens were again oven-dried at 80 °C to deactivate the enzyme and stabilize the internal structure. To preserve the natural properties of the loess, bulk material used for baseline characterization (particle size distribution, index properties) was air-dried at ≤40 °C, gently disaggregated, and passed through a 2.0 mm sieve; no chemical purification was performed. A low-temperature drying step at 80 °C was applied only after EICP curing when (i) terminating urease activity and (ii) stabilizing mass for carbonate quantification and microscopy sample preparation. Mechanical tests on untreated and EICP-treated specimens were conducted at controlled moisture states (sealed curing, then conditioning to target water content) and were not oven-dried at 80 °C prior to testing. The 80 °C step is below typical oven-drying temperatures used in geotechnical standards (≈105–110 °C) and is intended to deactivate urease without altering the mineral phases or fabric of this silt-dominated loess. Unconfined compressive strength (UCS) tests were then conducted to evaluate the mechanical performance improvements induced by the EICP treatment. For each treatment variant, three independent cylindrical specimens (n = 3) were prepared and tested in unconfined compression following GB/T 50123-2019 [50]. We tested three independent specimens per variant (n = 3), for a total of 15 UCS tests across the five conditions. Additionally, selected samples were analyzed using scanning electron microscopy (SEM) and energy-dispersive spectroscopy (EDS) to examine the morphology and spatial distribution of the carbonate precipitates.

2.4. SEM and EDS Analysis

To investigate the microstructural changes and identify the mineralogical composition of the precipitates formed during the EICP treatment, scanning electron microscopy (SEM) coupled with energy-dispersive X-ray spectroscopy (EDS) was employed (see Figure S1). After the UCS tests, representative specimens were selected from the middle sections of the treated samples to ensure consistency. The selected fragments were oven-dried at 60 °C for 24 h to prevent thermal decomposition of the carbonate precipitates. Once dried, the samples were gently fractured along natural planes to expose fresh internal surfaces and mounted on aluminum stubs using conductive carbon tape. To enhance conductivity and image quality, the samples were sputter-coated with a thin layer of gold (approximately 10 nm) using a vacuum coater. SEM observations were carried out with a JEOL JSM-IT500 microscope, operated at an accelerating voltage of 15 kV [3]. Multiple regions of each sample were scanned at various magnifications to capture the morphology of calcite crystals, their bonding with soil particles, and their distribution patterns. Elemental composition was confirmed via EDS point analysis and mapping, with particular focus on the Ca, C, and O signals, which are indicative of calcium carbonate precipitation. Powder X-ray diffraction (XRD) was performed qualitatively on finely ground, back-loaded specimens of natural loess and standard EICP-treated loess using Cu Kα radiation (Bruker AXS X-ray, D8 Advance). Scans were collected from 5–90° (2θ) with a step size of 0.02°. Phase identification was carried out using the PDF-4+ database (ICDD, 2023) [52] implemented in the Bruker EVA v14.2 software package, and reference codes for the identified phases were applied for comparison (Albite: PDF#09-0466; Quartz: PDF#46-1045; Calcite: PDF#05-0586).

3. Results

3.1. Ordinary EICP Process

In the ordinary EICP process, the temporal variations of electrical conductivity (EC) and pH under different concentrations of urea-CaCl₂ are shown in Figure 5. Initially, the EC showed only minor changes, which were attributed to the dissolution of urea, rather than its ionization. When CaCl₂ was introduced into the urea solution, the EC increased rapidly due to the ionization of CaCl₂, which produced a large number of charge-carrying ions. Subsequently, the hydrolysis of urea, catalyzed by urease, generated NH₄+ and CO₃²⁻ ions (as described in Equations (1) and (2)). The CO₃²⁻ ions then reacted with Ca²⁺ ions, leading to the precipitation of carbonate and the consumption of CO₃²⁻ ions (as shown in Equation (3)). The influence of urea hydrolysis on EC is more significant than that of carbonate precipitation, and it is the primary factor driving the continuous increase in EC throughout the EICP process. The EC continues to rise until the reaction reaches completion. In contrast, the introduction of CaCl₂ has minimal effect on pH. However, when urease is added to the urea-CaCl₂ solution, a significant change in pH is observed. This change is due to the generation of OH⁻ ions. As the EICP process progresses, the pH starts to decrease after reaching a peak value of approximately 8.9. This gradual decline is caused by the consumption of OH⁻ ions during carbonate precipitation.

The actual mass of carbonate precipitation and the precipitation ratio (PR) as a function of urea-CaCl₂ concentration are shown in Figure 6. The relationship follows a bell-shaped trend, indicating that neither excessively low nor high urea-CaCl₂ concentrations result in maximum carbonate precipitation. At low CaCl₂, Ca²⁺ is insufficient and carbonate precipitation is limited. At high urea and CaCl₂, urease becomes rate-limiting, slows urea hydrolysis, and reduces carbonate precipitation. As urea and CaCl₂ increase, the precipitation ratio (PR) declines, indicating higher efficiency at lower concentrations. These findings suggest that achieving both maximum carbonate precipitation and maximum PR simultaneously is not possible. However, the intersection point of the two curves may provide a compromise, allowing for relatively high values of both metrics. In the conventional EICP process, this optimal balance is observed at a urea-CaCl₂ concentration of approximately 0.3 mol/L.

3.2. Modified EICP Process

3.2.1. Effect of Magnesium Ion Addition

Based on the optimal substrate concentrations determined from preliminary test tube experiments, a series of EICP tests was conducted to investigate the effect of MgCl₂ addition on calcium carbonate precipitation behavior. The total concentration of divalent cations (Ca²⁺ + Mg²⁺) was maintained at 0.3 mol/L, while the molar ratio of CaCl₂ to MgCl₂ was systematically varied. Figure 7 shows the temporal variations in electrical conductivity (EC) and pH during the EICP process under different MgCl₂ concentrations. The results indicate that increasing the Mg²⁺ concentration enhances urease activity, accelerating urea hydrolysis. This results in more efficient carbonate precipitation, as evidenced by a more pronounced decline in EC and a higher peak pH compared to the control group without MgCl₂. These observations suggest the formation of a greater amount of CaCO₃, along with minor amounts of magnesium carbonate MgCO₃. As shown in Figure 8, the amount of CaCO₃ precipitated initially increased with the MgCl₂ concentration, reaching a maximum at 0.02 mol/L, after which it slightly declined. In contrast, the precipitation efficiency continued to increase and peaked when the molar ratio of CaCl₂ to MgCl₂ reached 1:1. This trend aligns with the findings of Putra et al. Based on the combined analysis of precipitation mass and efficiency, an MgCl₂ concentration in the range of 0.05–0.07 mol/L is recommended to achieve optimal precipitation performance in the EICP process.

3.2.2. Effect of Ammonium Ion Addition

As shown in Figure 9, the addition of ammonium chloride to the urease-calcium chloride solution increased the solution’s conductivity, but did not significantly alter the overall trend. However, as the ammonium ion concentration increased, the peak pH value gradually decreased. This effect occurs because ammonium ions reduce the dissociation of ammonia (NH₃) in water, resulting in lower hydroxide ion (OH⁻) concentrations. Additionally, NH₄⁺ ions neutralize OH⁻ ions, further consuming hydroxide ions and shifting the reaction equilibrium in the reverse direction. The resulting decrease in pH suppresses the formation of carbonate ions (CO₃²⁻), thereby reducing both the quantity and efficiency of calcium carbonate precipitation (see Figure 10). These findings demonstrate that NH₄⁺ exerts a significant inhibitory effect on the EICP process.

3.2.3. Effect of Calcium Chloride Addition

The fundamental components used in this process typically consist of urea and CaCl₂. To better understand the effect of CaCl₂ addition on carbonate precipitation, experiments were conducted in test tubes using higher concentrations of the basic components. The temporal variations of EC and pH under the influence of CaCl₂ addition are shown in Figure 11. As the concentration of CaCl₂ increases, the changes in EC become more significant. It can also be observed that EC remains relatively stable throughout the urea hydrolysis process, suggesting that carbonate precipitation counterbalances the effect of urea hydrolysis. In contrast, the influence of CaCl₂ concentration on pH appears negligible. However, during urea hydrolysis, pH rapidly increases initially, followed by a slow decline as the process nears completion. This is because urea hydrolysis releases OH⁻ ions rapidly, leading to an increase in pH. As carbonate precipitation begins, OH⁻ ions are consumed, causing the pH to peak and then decrease. The maximum pH of 8.6 occurs 30 min after the start of the process. Under the effect of CaCl₂ addition, precipitation induced by the improved EICP process reaches its maximum amount (see Figure 12). Hydrolysis forms Ca(OH)₂, indicating the consumption of OH⁻ ions. This consumption leads to a shortage of CO₃²⁻ ions, reducing carbonate precipitation. Additionally, as OH⁻ ions are consumed, the pH decreases, causing the freshly formed Ca(OH)₂ to release OH⁻ ions to counteract the pH decrease. This explains the delayed appearance of the peak pH. Compared to the PR of the improved EICP method with NH₄Cl addition, the effect of CaCl₂ addition further reduces the PR by converting all Ca²⁺ ions into carbonate precipitates.

3.3. UCS of the Modified Loess

Figure 13 shows the UCS performance of loess under both ordinary EICP and modified EICP technologies. Unconfined compressive strength (UCS) was determined per GB/T 50123-2019 [50]: the peak stress was used when present; otherwise, the stress at 15% axial strain was taken. For each condition, three independent specimens (n = 3) were tested. We report mean ± standard deviation (SD) and the coefficient of variation (COV = SD/mean × 100%). Across all conditions, replicate variability was modest (COV ≤ 10%), indicating good repeatability (see Table S3). The control group sample, which did not undergo any EICP treatment, exhibited the lowest strength, with a peak stress of approximately 370 kPa and a brittle shear failure mode. After treatment with ordinary EICP, the loess specimen showed a significant increase in peak strength to 580 kPa and exhibited a certain degree of ductility, indicating that CaCO₃ precipitation effectively reinforced the soil structure. The loess specimen treated with MgCl₂-modified EICP showed the most significant improvement in UCS, reaching a peak stress of 820 kPa with a slow stress drop, demonstrating excellent toughness and structural stability. In contrast, the loess specimens treated with NH₄Cl-modified and CaCl₂-modified EICP exhibited lower strengths of approximately 500 kPa and 460 kPa, respectively, showing varying degrees of stress reduction. These results are consistent with the previously observed differences in carbonate precipitation. Moreover, these findings highlight the positive and significant role of introducing different additives in modifying the EICP kinetics, contributing to the improvement of loess subgrade strength and stability.

3.4. Discussion

Figure 14 and Figure 15 present scanning electron microscopy (SEM) images and elemental distribution (EDS) maps of various loess samples. GO, GI, G2, G3, and G4 represent the control sample (natural loess), loess treated with ordinary EICP, and loess treated with modified EICP using MgCl₂, NH₄Cl, and CaCl₂, respectively. The SEM images show that the untreated GO sample exhibits dispersed particles with high porosity and no visible cementation, indicating the absence of any strengthening effects. In contrast, the GI sample, treated with ordinary EICP, displays abundant crystal precipitates on the particle surfaces. Calcium carbonate effectively fills the interparticle voids, forming a relatively stable skeletal structure, which enhances the soil’s cohesion. The G2 sample, treated with MgCl₂-modified EICP, shows the most compact microstructure, with uniformly distributed and well-developed crystals tightly binding soil particles together. This suggests that the addition of MgCl₂ significantly promotes crystal nucleation and growth. Conversely, the G3 and G4 samples show only localized and loosely distributed precipitates with smaller crystal sizes and poor interparticle connectivity. This indicates that the presence of NH₄⁺ and Ca²⁺ may suppress effective precipitation and bonding, limiting the reinforcement potential. Elemental mapping further supports these observations. In the G2 sample, the Ca, C, and Mg elements are densely and uniformly distributed, suggesting that Mg²⁺ facilitates the co-precipitation of multiple carbonate minerals (e.g., aragonite), contributing to enhanced structural integrity. In contrast, the distributions of C and Ca in the G3 sample are relatively sparse, and the C signal in the G4 sample is noticeably weak, reflecting the distinct influence of different additives on the precipitation behavior and stabilization performance of the EICP-treated loess.

The EDS results shown in Figure 15 and Table S4 further validate the previous observations. EDS is semi-quantitative, especially for C. We therefore use it to track relative enrichment and spatial association, and we cross-check with PR, UCS, and phase/morphology. In G0 (natural loess), Ca is low (14.88%) and O is high (38.41%). The signal is silicate-dominated, with no carbonate signature. PR is near baseline and UCS is 370 kPa. In G1 (standard EICP), Ca and C rise to 18.78% and 43.17%. This is consistent with new CaCO₃ at contacts and within pores. XRD (Figure S2) confirms the carbonate phases. PR increases, and UCS reaches 580 kPa (+56.8%). In G2 (Mg-modified EICP), Ca and C are elevated and Mg is detectable (0.38%). The co-occurrence of Ca–C–Mg indicates Mg co-precipitation or lattice incorporation. SEM shows denser bridges and needle-like or spherulitic textures, typical of aragonite/vaterite regimes. These textures explain the highest UCS (820 kPa; +121.6%) through more effective grain-bridging. The trend matches reports that Mg²⁺ slows calcite growth, stabilizes transient phases, and shifts polymorph selection toward aragonite, which can enhance contact efficiency. In G3 (NH₄⁺) and G4 (excess Ca²⁺), the O signal decreases relative to G1–G2. PR also drops, and UCS falls to 500 kPa (+35.1%) and 460 kPa (+24.3%). The mechanisms differ. NH₄⁺ alters the pH trajectory via the NH₄⁺/NH₃ buffer, moderating supersaturation and early nucleation. The result is less extensive cement bridges. Excess Ca²⁺ raises ionic strength and drives rapid local supersaturation. This favors proximal precipitation, crusting, and transport limitation (“clogging”), which lowers the effective, sample-wide bonding density. Two caveats apply. First, absolute C by EDS is uncertain; the direction of Ca enrichment and the Ca/(Si + Al) context are more reliable. Second, the observed benefits of Mg²⁺ are window-dependent. Outside the tested Mg/Ca and curing ranges, yield can decline. Within our conditions, however, the EDS trends, PR increases, and SEM textures cohere with the strength hierarchy G2 > G1 > G3 > G4 > G0. This supports an ion-controlled pathway from polymorph selection and bridge morphology to macroscopic UCS [40,41].

By comparing the experimental results of untreated loess, loess treated with ordinary EICP, and three modified EICP treatments (MgCl₂ (G2), NH₄Cl (GI3), and CaCl₂ (G4)), a systematic analysis was conducted, encompassing the reaction control mechanisms, microstructural differences, and macroscopic strength performance (see Figure 14). This study highlights the intrinsic characteristics of the “promotion mechanism,” “attenuation mechanism,” and “hijacking mechanism” that govern the EICP process. These mechanisms illustrate how different additives influence the precipitation process, with MgCl₂ promoting crystal growth and improving soil strength, NH₄Cl suppressing precipitation and bonding, and CaCl₂ causing a shift in the reaction dynamics, leading to reduced precipitation efficiency (see Figure 16). As the control group, the natural loess clearly exhibits microstructural characteristics of an uncemented state. SEM images reveal loosely arranged particles, high porosity, small interparticle contact areas, and a complete absence of cementitious substances. The EDS spectra show that the contents of Ca and C are only 4.88% and 33.16%, respectively, indicating a minimal presence of carbonate crystals capable of forming binding structures under natural conditions. This microstructure results in poor mechanical performance, as demonstrated by brittle failure and a low peak stress of only 370 kPa in the unconfined compressive strength (UCS) test, reflecting weak structural integrity and low load-bearing capacity. These inherent properties of natural loess highlight the contrast and provide significant potential for structural enhancement through subsequent EICP treatment, which aims to form a carbonate-based cementation network.

The ordinary EICP treatment enhances interparticle bonding by promoting the precipitation of CaCO₃, formed through a urease-mediated hydrolysis of urea, which releases OH⁻ and CO₃²⁻ in the presence of sufficient Ca²⁺. This reaction pathway is characterized by the following features: (1) stable reaction kinetics, (2) the formation of calcite as the dominant CaCO₃ polymorph, and (3) uniform crystal deposition along particle surfaces and within pore spaces. SEM images show that the treated sample surface is densely covered with crystalline deposits, while EDS analysis reveals a significant increase in the mass fractions of Ca and C, confirming the formation of a typical particle-to-particle cementation structure. Our loess results agree with EICP gains reported for loess slopes under erosion loadings [41]. They also align with strength improvements seen in fine industrial residues [42] and with the bio- vs. polymer-cured contrasts in fly ash [43]. Compared with sandy soils [39], absolute UCS is smaller, but ion-dependent trends are consistent. Zhang et al. [40] demonstrated EICP without an external Ca source in Pb-loess. Our data complement that work by mapping dose–response under controlled Ca, Mg, and NH₄ additions and by separating ion effects. The unconfined compressive strength (UCS) increased by 57%, demonstrating the effectiveness of this mechanism. This standard reaction pathway serves as a baseline for evaluating subsequent modifications to the EICP process. XRD of the natural loess shows the expected silicate phases, primarily albite and quartz, with no discernible carbonate reflections. In contrast, the EICP-treated specimen exhibits additional peaks indexed to calcite [52], confirming calcium carbonate precipitation under the standard formulation (Supplementary Figure S2).

Mg²⁺ can inhibit calcite step growth and stabilize transient phases. It promotes vaterite and aragonite under specific Mg/Ca and pH windows. Such shifts explain slower early precipitation but better grain-bridging at cure, which increases UCS. NH₄⁺ affects pH and conductivity during ureolysis. Ca²⁺ controls saturation, growth rate, and pore-scale transport. These roles match prior observations in loess and granular systems. The addition of MgCl₂ to the EICP process demonstrates a typical “promotion mechanism” effect. Firstly, Mg²⁺, due to its high hydration energy, delays dehydration and participation in precipitation reactions, which initially suppresses the rapid consumption of CO₃²⁻ and extends the duration of urea hydrolysis and carbonate precipitation. Secondly, Mg²⁺ inhibits the rapid growth of calcite crystals through a “crystal face poisoning” effect, promoting the formation of a new carbonate phase dominated by aragonite. This mineral phase features a denser structure and superior spatial filling capacity. In the EDS spectra, a pronounced Mg peak is observed alongside enhanced Ca and C signals. SEM images reveal crystals enveloping soil particles and filling pore spaces, resulting in the densest microstructure among all treatments. The unconfined compressive strength (UCS) increases significantly, reaching 820 kPa. This mechanism, characterized by “slow release + synergistic precipitation,” markedly enhances the rate and spatial uniformity of carbonate formation without increasing the total salt concentration, thereby substantially improving the structural integrity and strength of the treated loess.

The introduction of NH₄⁺ inhibits the urea hydrolysis process by shifting the reaction equilibrium in the reverse direction, reducing the generation of both OH⁻ and CO₃²⁻. Additionally, NH₄⁺ further suppresses carbonate formation through neutralization reactions with OH⁻. SEM images show sparse precipitates with scattered crystal particles, while EDS analysis reveals a significantly lower O content compared to the G1 group, indicating poor precipitation efficiency. The resulting structure lacks effective cementation, with high porosity, leading to a reduction in UCS to 500 kPa. This mechanism clearly demonstrates the inhibitory effect of NH₄⁺ on both enzymatic activity and the precipitation pathway, which can be characterized as a compound “enzyme attenuation + reverse reaction” mechanism.

Under elevated Ca²⁺ concentrations, although the formation of additional CaCO₃ is theoretically favored, Ca²⁺ preferentially reacts with OH⁻ to form Ca(OH)₂, thereby consuming OH⁻ and inhibiting the generation of CO₃²⁻. This process results in the formation of ineffective precipitates. SEM images show discontinuous and morphologically irregular precipitates, while EDS analysis reveals a decrease in C content, indicating insufficient carbonate formation. As a result, the UCS drops to 460 kPa, reflecting low structural load-bearing capacity. This deviation in the reaction pathway causes the “hijacking” of carbonate precipitation, leading to the formation of an unstable structure and suppressing the reinforcement effect.

In summary, among the different modification groups studied, the natural loess (control) exhibited a loose structure and extremely low strength, providing a baseline for EICP-based reinforcement. Ordinary EICP treatment facilitated the initial formation of calcium carbonate precipitates, significantly enhancing the soil’s structural compactness and load-bearing capacity. Among the modified treatments, MgCl₂ demonstrated the most effective reinforcement by inducing the formation of denser and more uniformly distributed polymorphic carbonate precipitates through crystal phase regulation and delayed reaction mechanisms. In contrast, NH₄Cl inhibited enzyme activity and reversed the reaction direction, reducing carbonate formation and weakening the cementation between particles. Excess CaCl₂ triggered a “hijacking mechanism” that suppressed effective CaCO₃ precipitation and instead promoted the formation of by-products like Ca(OH)₂, which proved detrimental to structural enhancement.

This study not only elucidates, at the micro-mechanistic level, how various ions influence carbonate precipitation pathways and crystal morphologies during the EICP process, but also systematically evaluates their macroscopic effects on the mechanical performance of loess. The findings highlight the significant potential of biomineralization technology in the engineering reinforcement of loess regions. In particular, the Mg²⁺-induced “synergistic precipitation mechanism” demonstrates unique advantages in forming a dense cementation network and improving the soil’s load-bearing performance. This approach offers a promising and environmentally friendly solution for stabilizing loess slopes and subgrades in high-altitude, arid, and rain-eroded regions.

However, several critical challenges remain for the practical application of EICP in loess subgrade engineering. First, the inherent heterogeneity and spatial variability of loess porosity can hinder uniform solution infiltration and precipitate distribution, affecting the continuity and stability of the reinforcement layer. Second, the EICP reaction is highly sensitive to environmental factors such as temperature, pH, and ion concentrations, making its performance under field conditions difficult to predict. Third, research has largely emphasized short-term strength, whereas long-term and dynamic performance has not been systematically evaluated; this includes durability, permeability, freeze–thaw resistance, and the response under earthquake loading (e.g., cyclic strength, stiffness degradation, and liquefaction/softening potential). Additionally, the development of efficient and controllable grouting techniques and site-adaptable operational procedures remains a key area for improvement. Future research should focus on enhancing reaction controllability, regulating the spatial distribution of precipitates, exploring multi-component modification strategies, and conducting in situ feasibility trials under realistic field conditions. These efforts will be crucial for advancing the practical, standardized, and scalable application of EICP technology in complex engineering environments, providing robust support for subgrade reinforcement and infrastructure safety in loess regions. Although EICP avoids the clinker-related CO₂ burden of cementitious binders and its precipitation pathways can be tuned through salt chemistry, labeling the approach “environmentally friendly” requires evidence that salt additions do not impose unacceptable loads on soils or shallow groundwater. In particular, chloride and ammonium introduced via CaCl₂, MgCl₂, or NH₄Cl may transiently elevate electrical conductivity and nutrient concentrations. The datasets reported here emphasize mechanism (ureolysis, polymorph selection, grain-bridging textures) and early-age strength rather than durability. To address application-relevant performance, we will (i) subject treated loess to freeze–thaw cycling between sub-zero and ambient conditions and (ii) evaluate pH resistance under acidic–neutral–alkaline solutions. Performance metrics will include retained UCS and stiffness, permeability change, mass loss, carbonate yield, and microstructural/phase stability (SEM/XRD/FTIR). These experiments are underway and will be disseminated separately.

4. Conclusions

The strength enhancement and microstructural evolution of loess treated by ordinary and modified EICP methods were investigated through mechanical testing, SEM-EDS characterization, and reaction mechanism analysis. The results demonstrate that the variation in carbonate precipitation pathways caused by different additives directly influences the bonding effectiveness and macro strength of the treated loess. Based on the results and discussion, the following main conclusions can be drawn:

(1) EICP treatment significantly enhances the strength of loess, with the MgCl₂-modified group demonstrating the best performance. The unconfined compressive strength (UCS) of untreated loess was 370 kPa. Ordinary EICP increased the UCS to 580 kPa, a gain of 210 kPa (56.8%) relative to the untreated baseline. MgCl₂-modified EICP further increased the UCS to 820 kPa, a gain of 450 kPa (121.6%) over untreated and 240 kPa (41.4%) over ordinary EICP, consistent with a moderated reaction rate and the promotion of aragonite. The resulting crystals uniformly fill the pore spaces, producing the densest structure and exhibiting the highest reinforcement potential.

(2) The three modifying agents induce distinct carbonate precipitation mechanisms that control the reaction pathway and cementation efficiency. Mg²⁺ triggers a “promotion mechanism,” facilitating the co-precipitation of CaCO₃ and trace amounts of MgCO₃, resulting in a dual enhancement in both the quantity and rate of precipitation (with a PR exceeding 114%). In contrast, NH₄⁺ significantly inhibits urease activity and OH⁻ release, lowering the pH peak to 8.5 and reducing UCS to 500 kPa, consistent with a “decay mechanism.” Excess Ca²⁺ leads to a “hijacking mechanism,” favoring the formation of Ca(OH)₂ as a by-product, thereby reducing the efficiency of CaCO₃ precipitation and lowering UCS to 460 kPa.

(3) This study elucidates the micro-mechanisms of ion-regulated EICP reactions and their corresponding macroscopic performance, providing theoretical support for environmentally friendly reinforcement of loess subgrades. SEM and EDS analyses confirm that the Mg-modified group exhibits the most continuous crystal distribution and significantly increases Ca and C contents. In contrast, the NH₄⁺ group shows a fragmented structure and sparse precipitation, while the excess CaCl₂ group features irregular crystals and reduced C and O element signals. The findings verify the regulatory role of ionic composition on carbonate crystal formation and soil mechanical properties, offering valuable data to guide formula optimization and grouting strategies for the practical engineering application of EICP technology.

(4) This study does not deal with the dynamic earthquake behavior of loess which has great importance in engineering. Future work will assess durability via multi-cycle freeze–thaw and controlled pH excursions. We will track retained UCS and stiffness, permeability, mass loss, CaCO₃ yield, and microstructural or phase stability using SEM, XRD, and FTIR. We will quantify environmental risk using column-flushing leachate chemistry (EC, pH, Cl⁻, NH₄⁺/NO₃⁻, Ca²⁺, Mg²⁺) and plant germination and early-growth assays. We will advance composition analytics using XRF for major elements and TIC, CHN, or TGA for carbonate carbon. We will develop field-relevant dosing and delivery protocols, since additives were tested separately here, to limit pore clogging and improve treatment uniformity. We will also increase statistical rigor with larger sample sizes and effect sizes with confidence intervals. Together, these steps will translate the chemistry-first insights into validated, durable, and environmentally responsible EICP treatments for loess subgrades.