The test soil samples were obtained at a depth of 2 m from a project on Chuansha Road in Pudong New Area, Shanghai. By testing, the soil is clay, and the particle content of less than 0.075 mm is more than 90%, of which the clay content is more than 30%. The soil specific gravity is 2.67, the plasticity index is 23%, organic matter content was 2.65%, and the pH value is 7.9. After transporting the collected soil samples to the laboratory, air dried naturally, removing obvious garbage such as leaves and plastic bags. Then dried in a constant temperature drying box at 75°C and sieved by 2 mm. The preparation process is shown in Figure 1.

According to the standard of Soil Environmental Standard (GB15618-2018), CuSO₄·5H₂O and ZnSO₄·7H₂O were added to the soil, so that the concentration of Cu and Zn exceeded the standard value by two four times, both of which were 1,000 mg kg⁻¹. Then, add deionized water until the soil is saturated to form 50% soil moisture content, and stir well. To fully adsorb heavy metals, the soil was cultivated in a constant temperature and humidity environment for 1 month, at which point adsorption equilibrium is reached (Kamal et al., 2021).

To detect the soil samples’ heavy metal content before and after pollution, the inductively coupled plasma (ICP) test method was used. The original soil samples were found to have very low Cu and Zn content, 30.1 ± 5.7 mg kg⁻¹ and 142.3 ± 8.9 mg kg⁻¹, respectively. After 1 month of contamination, the Cu and Zn concentrations were 967.1 ± 20.5 mg kg⁻¹ and 1139.4 ± 31.2 mg kg⁻¹, which basically reached the 1,000 mg kg⁻¹ concentration requirement set for the test, and the leaching test could be started.

The removal rate of heavy metals is a fundamental consideration in the study of remediation effectiveness. The calculation equations are shown in Equations 1, 2. M h m = M s × B (1) W = B − N × M s M h m × 100 % (2)

N -Cu, Zn mass ratio after leaching (mg kg^-1);

M s -Soil mass (kg);

M h m -Total Cu and Zn (mg);

B -Cu, Zn mass ratio (mg kg^-1);

W -Heavy metal removal rate (%).

Due to the non-homogeneity of the soil, the soil samples had different internal leaching orifices and different contact times with the leaching solution, resulting in slight differences in heavy metal removal rates at different locations in the same soil layer. Figure 2 shows the average removal rate and its standard deviation for each location of the three model buckets. The average difference in the removal rate between CA-round 1 and CA-round 2 was more than 50%, and the difference between CA-round 2 and CA-round 3 was about 10%, which strongly supported the conjecture that a large number of citric acid molecules in CA-round 1 of leaching failed to take part in the reaction with the heavy metals, but were degraded by soil microorganisms (Wang et al., 2020; Li et al., 2018). The highest Cu and Zn removal rates were all in the bottom part of CA-round 3, reaching 85.0% ± 0.5% and 89.1% ± 0.3%, respectively. In CA-round 2, Cu and Zn removal rates reached 77.0% ± 1.9% and 78.4% ± 0.8%, removing most of the heavy metals. The addition of one leaching cycle extended the treatment duration by 164 h, yet only resulted in marginal improvements of 8.0% and 10.7% in heavy metal removal rates. This demonstrates that after CA-round 2, further time increases show no significant enhancement in heavy metal removal efficiency. Moreover, each additional leaching cycle doubles the consumption of cleaning chemicals. Considering economic factors, CA-round 2 is sufficient to meet remediation requirements for contaminated soil.

In addition, there was a clear pattern of removal rates between locations - the closer to the bottom, the higher the removal rate. The reason was that citric acid could effectively remove heavy metals from the surface of soil particles in the top and middle sections, but these metals remained in the pore fluid due to the poor drainage condition, and were not wholly removed, causing a low removal rate. On the contrary, the bottom drains quickly and has a high removal rate.

The reaction mechanism is shown in Figure 3. Citric acid is a low molecular weight organic acid that is a natural complexing agent, that can ionize three H⁺ in water. The hydrogen ions could, on the one hand, undergo an ionic replacement reaction with Cu²⁺ and Zn²⁺ adsorbed on the surface of soil particles, so that the Cu²⁺ and Zn²⁺ were free in the pore liquid, and the diffusion electric double layer of the polluted soil particles thickened in the process of replacement, which further led to the disintegration of agglomerates (Xie et al., 2021), and, on the other hand, provided an acidic environment for the leaching to make the mineral compositions change (Liu et al., 2023), as shown in Figure 3 (Ⅱ) (Ⅲ). Cu²⁺ and Zn²⁺ free in the pore liquid would interact (complexation) with another product (HA^2-, A^3-, H₂A⁻) produced by citric acid ionization to form a water-soluble complex, and the water stability of the complexed product was MHA⁻>MHA>MHA⁺ (Ke et al., 2020). The complexation process and the equation are shown in Figure 3 (Ⅴ). At the same time, the acidifying effect of citric acid weakens the electrostatic attraction between soil particles, which leads to the disintegration of agglomerates, which is manifested as an increase in the percentage of clayey grain content. At this point the soil compression coefficient increases, the pore ratio rises and permeability is enhanced. This structural change has a positive effect on heavy metal removal, and the elevated pore connectivity accelerates the infiltration and diffusion of the citric acid solution, resulting in fuller contact between heavy metals and citric acid.

Figure 4 reveals the effect of different leaching rounds with citric acid on porosity and permeability coefficient. It is visible that the curves shift upward, indicating that the porosity gradually increases from CA-round 1 to CA-round 3. By observing the decreasing curves of void ratio in Figures 4A, B, there were two stages according to changes: the stage I (0–100 kPa), the porosity decreases by more than 70% rapidly. In this stage, the top and middle soil mainly drained the free water.

The second Ⅱ (100–800 kPa) for the void ratio change was gentler; the top and middle soil samples in this stage were discharged of pore-free water post a certain consolidation degree, with the soil skeleton mainly bearing the upper load (Liu et al., 2023), resulting in the compression of the pores in the particles. In Figure 4C, the cut-off point between the two phases was the vertical load of 200 kPa, which was due to the more obvious influence of the vacuum load on the bottom soil, which gave the soil a certain consolidation yield pressure (Hong et al., 2010), so that the decline stage of the void ratio was extended. Overall, the porosity was positively correlated with the number of leaching rounds. Still, the distribution pattern of the pore development in the soil remained unchanged.

The variation of the permeability coefficient with pressure for different leaching rounds is given in Figures 4D–F. Similarly, the curve could be divided into three phases.

In Stage I (0–100 kPa), the permeability coefficient declined rapidly with the vertical load.

In Stage II (100–400 kPa), the permeability coefficient declined slowly. Within this stage, the difference between different leaching rounds at the same location was more apparent. The reason for this is that as the number of leaching rounds increased, the remove rate increased and the diffuse double layer of fine particles is subsequently thickened (Li et al., 2015), the free water was conversed to bound water. Only 10%–20% of the heavy metals were removed in CA-round 1, leading to a smaller diffusion electric double layer thickness, more significant effective void, and higher drainage efficiency compared to subsequent rounds, so the infiltration coefficient of CA-round 1 had the smallest change by the second stage. In contrast, CA-round 2 and CA-round 3 achieved heavy metal removal rates exceeding 70%, causing bound water clearly increased. At the same time, the inter-particle distance decreased significantly after the first stage, amplifying the effects of diffusion electric double layer thickness and bound water content on the permeability coefficient (Zhang J. et al., 2022). Consequently, after the considerable reduction of free water in the first stage, CA-round 2 and 3 still needed to discharge more bound water and some free water than CA-round 1 in Stage I. This resulted in a more pronounced decrease in the permeability coefficient of CA-round 2 and 3 in stage II, with a gradual increase in the slope.

Stage III (400–800 kPa), the permeability coefficients exhibited a slight decrease, which was related to the porosity reduction caused by the dislocation of the chimeric soil particles that made up the soil skeleton (Yu et al., 2022; Kwangkyun et al., 2013). The box plots in Figures 4D–F are statistically analyzed for the permeability coefficient, with the middle line being the average line. Obviously, increasing leaching rounds contributes to a larger average permeability coefficient. The middle and bottom soil samples experienced a noticeable increase in box volume indicating that multiple leaching of soil samples had a large impact on the middle and bottom soil.

In engineering practice, a constant coefficient of consolidation is usually used for settlement calculations (Zuo et al., 2024); however, numerous researches indicated that the coefficient of consolidation changes constantly during soil compression (Elkateb, 2017; Mejlhede et al., 2023; Cui et al., 2023). Similarly, in Figures 5A–C that with the increase in the amount of citric acid leaching, the soil’s consolidation coefficient curve shifted upward, and the consolidation coefficient increased gradually. The most significant changes were observed during CA-round 1 to CA-round 2, especially in Figure 5C, where the consolidation coefficient increased by more than twice. This is due to the high porosity, the dispersed particles and the fragile soil skeleton structure caused by the multiple leaching rounds, which is more susceptible to compression.

In Figures 5D–F he compression coefficient curve of the soil shifted upward, indicating a gradual increase in compression coefficient with each additional round of citric acid leaching. This was particularly evident in Figure 5D, where the compression coefficient after three rounds of leaching in the stage I increased by about 87% compared to the first round. The reason is similar to the consolidation coefficient, related to porosity and particle variations (Lin et al., 2022).

The compressive modulus of each test group is shown in Figures 5D–F, which are inversely proportional to the compression coefficients. The overall compression modulus gradually decreased with leaching rounds increasing. It is well known that the compressibility of the soil skeleton is related to its particle composition, bound water content, the thickness of the double electric layer, and other micro-physical and chemical properties (Jongkwan et al., 2021; Aria et al., 2022; Oliveira and Massao, 2015; Hao et al., 2022; Wu et al., 2023). Overall, as the soil particles interact with citric acid in different rounds changes the microscopic properties of the soil considerably, in turn, affects the engineering properties of the soil.

Figure 6 compared the particle distribution of different leaching rounds at the same location, and observed the effect of the leaching round on soil particle movement. Figures 6A–C corresponded to the top, middle and bottom layers of the soil sample, respectively. The particle composition changed continuously with the increase of the leaching rounds, and the local enlarged images showed the cumulative particle content at the 0.005 mm (clay) position. It can be found that the clay content at the top was CA-round 1 > CA-round 2 > CA-round 3, while the bottom showed the opposite pattern, indicating that the migration movement of clay particles gradually intensified with the leaching rounds, and more clay particles migrated to the bottom. The clay content in the middle part shown in Figure 6B was CA-round 3 > CA-round 1 > CA-round 2, which was due to the increase of porosity from CA-round 1 to CA-round 2, the top and middle clay particles migrated a large amount to the bottom part under the action of pore fluid (Yang et al., 2023). The migrating clay content was greater than the clay content decomposed by large particles, so that the overall clay content decreased (Jez and Lestan, 2016). From CA-round 2 to CA-round 3, the effect of the agent was more obvious, the particle decomposition produced vast clay particles, and the top clay particles migrated to the middle part and could not be discharged in time, so the clay content in the middle part increased, exceeding the initial content.

Figure 6D showed the composition of soil particles in different locations after CA-round 3, it can be seen that clay content was bottom> middle > top. Because the bottom drainage board was the destination of clay migration, the clay particles converged downward after 3 leaching rounds, with the greatest clay content near the bottom drainage board.

The microstructures of the soil samples with different leaching cycles and the original contaminated soil samples at magnifications of 400, 1,000, 2,000 and 5,000 are presented in Figure 7. In Figures 7A, B, compared to the original contaminated soil, the soil samples in CA-round 1 showed a decrease in the aggregates size. As leaching cycles increased, Figures 7C, D depict a gradual loosening of the microstructure. The surface of the soil samples was relatively flat, with a significant reduction in folds, an increase in surface-to-surface contact, and a pronounced macroporous structure, which is consistent with the results of the pore ratio tests. In Figures 7E–L, it was more obvious that with leaching rounds increased, the soil microstructure became looser, fine particles increased, and there were more pores, indicating the strong corrosive impact of citric acid leaching. Thus, the aforementioned pattern of the permeability coefficient occurred. At the same time, a distinctive needle-stick structure emerged. As the number of leaching rounds increases and the reaction continues, the structures become more pronounced and more numerous. This structure mainly originated from the chemical action of citric acid, its acidic qualities and strong complexing ability, and this reaction prompted the recombination and precipitation of internal substances in the soil, which led to the formation of a unique needle-cluster morphology. The needle-stick structures act as miniature supporting skeletons, interspersed between the soil particles, providing additional support to the soil mass. Remarkably, in Figures 7M–P, the 5,000 × SEM images show significant agglomeration of the soil particles where these substances occur, suggesting that the cluster structure has a positive effect on the agglomeration of the soil particles. In addition, it has been found that such structures are gel particles, which are associated with an increased content of silica, as a result of the change in mineral composition after drenching, and it has been noted that such substances help to stabilize the structure (Wen et al., 2022).

Citric acid, as an external substance, greatly changed the pore fluid environment of the contaminated soils during leaching, then altered the mineral content of the soil (Ke et al., 2022). The top part of the soil was in direct contact with the leaching agent, resulting in the highest concentration of the agent in the pore fluid, the XRD analysis results were shown in Figure 8. The data in the figure are the average of the experiments.

Quartz constituted the primary component in the soil both pre- and post-leaching, with a concentration exceeding 50%, showing a gradual rise with each subsequent leaching cycle. The content of chlorite also increased with the increase in leaching. attributed to the transformation of illite into chlorite in a Mg²⁺-rich environment (Jie et al., 2017). The acidic environment of citric acid would cause the Ca²⁺ and Mg²⁺ on the dolomite and calcite surface to be desorbed and free from pore fluid, creating favorable conditions for the illite transformation, so the illite content steadily declined to 5.3%. A related study (Ke et al., 2022) showed that illite and feldspar minerals were dissolved in an acidic environment and produced SiO₂, Al₂O₃, Fe₂O₃, etc., so the content of quartz and hematite increased. The feldspar mineral content decreased overall with each leaching round, although there was a slight increase in the first round likely due to insufficient acidification, so there was no significant dissolution, but rather, some albite was formed under the environmental conditions (Meng et al., 2023). With the increase in leaching, the soil environment was gradually acidified, the dissolution of feldspar minerals intensified, and the content decreased.