Introduction

Mining and metal processing activities provide the raw materials necessary for metallurgical industries to manufacture products such as electronics, energy storage, and transportation, which are essential devices for modern lifestyles. These activities offer numerous benefits to society, including economic development (Ali et al. 2017). For instance, the Brazilian mining sector contributes nearly 12% to the country’s gross domestic product (Brazilian Economy Ministry 2022; Gardioli et al. 2023). However, soil pollution caused by potentially toxic elements (PTE) is an undesirable side effect of mineral exploration activities (i.e., mining and ore refinement) and had been an environmental concern in many parts of the world (Ettler 2016; Kalisz et al. 2022; Kennedy et al. 2023), not only in loco (i.e., mine spot) but also spreading to the surrounding areas through windblown dust deposition (Ono et al. 2016; Naz et al. 2018; Entwistle et al. 2019). Once these PTE are remobilized from soil and sediments, they might migrate to groundwater, plants, animals, and finally humans, posing a potential threat to public health (Chen et al. 2022; Li et al. 2023).

Soil contamination is a major environmental concern that requires action in prevention and urgent remediation to avoid PTE spreading and possible damages to ecosystems. There is a strong interest in developing strategies and methods to ameliorate contaminated soils and prevent PTE spreading, such as landfilling and excavation; however, these actions are often expensive and environmentally unfeasible (Vareda et al. 2019; Rajendran et al. 2022). The amendments reduce the PTE mobility due to adsorption, complexation, and precipitation mechanisms, making them less available for human and plant uptake and less prone to leaching into groundwater (Bolan et al. 2003; Kumpiene et al. 2008). Utilizing soil amendments for in situ PTE immobilization is profitable and environmentally friendly option because (i) it is low invasive, as it is focused on soil surface (i.e., ~ 20 cm, after soil tillage), (ii) it is applicable for a wide number of inorganic pollutants (Zhai et al. 2018; Vrînceanu et al. 2019; Gao et al. 2023), and (iii) it is affordable ($40–$65/m3 of soil (Martin and Ruby 2004), €45–€170/m3 (Liao et al. 2022)), if compared to other remediation options such as electrokinetic ($26–$295/m3 (Federal Remediation Technologies RoundTable 2009), ~ $200/m3 (Reddy and Cameselle 2009)), vitrification ($300–$500/m3 (USEPA 2004), 120–680/m3 (Reddy and Cameselle 2009)), phytoremediation ($25-$100/m3 (USEPA 2004), $626–$2322/m3 (Robinson et al. 2015), ~ $250/m3 (Reddy and Cameselle 2009), or solidification ($87–$190/m3, (Wan et al. 2016)). It is noteworthy that updated values of remediation cost are ignored or not presented in recent literature.

Several amendment studies have been dedicated to evaluating the PTE’s ability to immobilize in soil and decrease its environmental risks, such as the application of phosphate (He et al. 2013), alkaline materials (Lee et al. 2009), biochar (Puga et al. 2016), and biosolids (Basta et al. 1997). However, soil contamination with multiple PTE is a crucial challenge for soil remediation technologies because of the interactions among the soil properties, the chemical behaviors of each PTE (e.g., solubility and redox variations), and the soil amendment reactions. All these factors may lead to conflict when attempting immobilization multiple contaminates (Fang et al. 2016; Wang et al. 2019). In our previous study, we evaluated the effectiveness of some soil amendments for immobilizing lead (Pb) in mine-waste impacted soil and concluded that phosphate and lime decreased Pb desorption and increased the amount of Pb in the fraction with the lowest mobility. However, biochar and biosolids had the opposite effect, resulting in an increase in Pb desorption (Gomes et al. 2022).

For this study, we evaluated the addition of phosphate, lime, biochar, and biosolids to decrease the mobility of the Pb and Zn in an impacted soil under the secondary forest located in the vicinity of an abandoned, heavily polluted mine site in Brazil previously assessed (Gomes et al. 2022). The efficacy of metal immobilization was evaluated by PTE desorption using stirred flow kinetic and sequential extraction approaches. Additionally, the alterations in Zn speciation resulting from soil amendment reactions were investigated using X-ray absorption near edge structure (XANES) spectroscopy. Hence, we expanded the significance and interpretation of remediation strategies presented in our previous study for a heavily contaminated site (PTE in gram per kilogram), for a contamination for both Zn and Pb observed in a forest soil in measured in milligram per kilogram. We expect that the outcomes of our research contribute to the understanding of remediation practices in variated levels of soil contamination by multiple PTE.

Materials and methods

Contaminated forest soil

The soil sample was collected surrounding (less than 1 km distance) a discontinued Zn exploitation area sited in Vazante city, State of Minas Gerais, Brazil (approximately to geographic coordinates 17° 55′ 3″ S, 46° 49′ 2″ W, see Fig. 1SM). This area is covered by a secondary forest of the Brazilian “Cerrado” biome, known as the savanna-like formations, which is composed of xeromorphic, sclerophyllous, and exotic species such as Eucalyptus.

Twenty subsamples (~ 500 g) of soil were collected from the superficial soil layer (0–20 cm) at six points along a transect throughout the forest site (Fig. 1SM). The composite sample, hereafter referred to as the forest soil, was air-dried at room temperature (~ 22 °C) and passed through a 2-mm sieve. A portion of the forest soil sample was dispersed by a sodium hexametaphosphate solution (5 g L−1) in mechanical agitation (16 h), and then, the particle size distribution was measured according to Stokes law (Pansu and Gautheyrou 2007). We determined 14% sand, 9% silt, and 77% clay in the soil sample. The major mineral phases identified by X-ray diffractometry (XRD) analysis in the silt and clay fractions were kaolinite, gibbsite, muscovite, goethite, and hematite, while quartz is the main mineral phase in the sand fraction (data not shown).

Amendments and incubation time

Individual portions of the original forest soil (100.0 g) were amended with each soil amendment at different rates. Monobasic ammonium phosphate (NH4H2PO4, analytical grade), as a phosphate amendment, was applied evenly to the soil in a 0.5:1 (quarter dose), 1:1 (half dose), and 2:1 (full dose) molar ratio of PO43− to the sum of PTE (i.e., Pb, Cd, and Zn both in molar base) as suggested elsewhere for soil amelioration (Cao et al. 2009). Likewise, lime (CaCO3, analytical grade) was added as an alkaline material, at a 0.25:1 (quarter dose), 0.5:1 (half dose), and 1:1 (full dose) molar ratio of CO32− to the sum of PTE, as suggested by Basta and McGowen (2004). The biochar was made from sugarcane (Saccharum officinarum) straw pyrolyzed at 450 °C, as detailed by Feola Conz et al. (2017). The rates used were 2.5, 5, and 10% weight-based (m:m; wt%) of biochar, as evaluated previously (Puga et al. 2015, 2016; Ali et al. 2020). Additionally, we tested the residual biosolids produced by a station of water treatment in Piracicaba city (State of São Paulo, Brazil). It was composted, stabilized, and analyzed following the official Brazilian guidelines (MAPA 2014). The rates used for biosolids were 5, 10, and 20% weight-based (mass/mass), as tested by Madejón et al. (2010) and Elkhatib and Moharem (2015). Each combination of amendment type (phosphate, lime, biochar, and biosolids) × amendment rate was carried out in triplicate.

The samples with amendments were well homogenized, and water was added to reach 70% of the maximum water retention capacity (MWRC), along with the forest soil without any amendments (control). All samples were kept incubated at 25 ± 2 °C for 60 days, with the water refilled every 3 days based on mass loss (i.e., water loss ~ 30% MWRC) to conserve the moisture level suitable for chemical reactions.

Soil chemical analysis, desorption, and sequential extraction

Soil chemical analyses were conducted following the incubation period. For all the analyses mentioned below, we replicated the same procedures and protocols used previously (Gomes et al. 2022). Pb and Zn desorption kinetic was measured by the stirred flow method. In brief, 200 mg of the samples (initial forest soil and it amended) was placed in the reactor, a 12-mL stirred flow chamber (Fig. 2SM), and a 25-mm diameter cellulose filter membrane with a 0.45 µm pore size was used in the reaction chamber. The solid and desorption solutions (Mehlich-III: 0.2 M CH3COOH + 0.25 M NH4NO3 + 0.013 M HNO3 + 0.015 M NH4F + 0.001 M EDTA) were stirred in the reactor using a magnetic stirrer at 300 rpm, and the liquid portion of it was then flowed through the chamber, passing the filter at a rate of 1 mL min−1 with a piston displacement pump designed for use in a high-performance liquid chromatography system, been that the liquid portion was collected at 2-min intervals. The concentration of Pb and Zn was measured by an ICP-OES (Pb emission line at 220.353 nm, calibration curve from 0 to 20 mg L−1 Pb; and Zn emission line at 213.856 nm, calibration curve from 0 to 30 Zn mg L−1; all calibration curves gaining R2 values greater than 0.992). Posteriorly, amount of Pb and Zn desorbed was calculated (Yin et al. 1997) and plotted as cumulative Pb or Zn desorption (in % of the total). The first-order kinetics was used to model the correlation between the amount of PTE desorbed over time, as previously demonstrated (Sparks 2003; Barreto et al. 2023).

The sequential extraction of Pb and Zn was conducted following a modified method from the (Silveira et al. 2006) protocol. Briefly, 1 g of air-dried soil was placed in 50-mL polycarbonate centrifuge tubes and mixed in a stepwise fashion with the following solutions: 0.1 M CaCl2 (15 mL, 2 h at 22 ± 2 °C) to target metals in a exchangeable form (F1); 1 M NaOAC (30 mL, pH 5, 5 h at 22 ± 2 °C) to target carbonate (F2); 40% NaOCl (5 mL, pH 8.5, 30 min at 80 °C) to target organic-matter bound (F3); 0.2 M oxalic acid + 0.2 M NH4-oxalate (30 mL, pH 3, 2 h at 22 ± 2 °C) to target poor crystalline iron oxide bound (F4); and, finally, the solid fraction was digested by the hydrochloric-nitric acid mix (i.e., EPA3051a (USEPA 2007)) method to measure the residual form (F5). The concentrations of Pb and Zn in each extract were measured by ICP-OES. A post hoc multiple comparison for observed means was carried out in Statistica 13 software (StatSoft Inc.) using Fisher’s LSD method at 5% (p < 0.05).

Zn spectroscopic speciation

Zn was chosen among the metal contaminants because it was the element with the highest total concentration in the samples, high enough to offer a suitable spectrum quality for XANES analysis. Soil samples that received the highest rate of each amendment were grounded at 50 mesh. The Zn K-edge XANES data was collected at Beamline XAFS2 at the Brazilian Synchrotron Light Laboratory (LNLS, https://lnls.cnpem.br/facilities/xafs2-en/). The Si (111) monochromator energy was calibrated to 9659 eV based on the first inflection point in the K-edge derivative spectra from a Zn foil. Spectra were acquired in fluorescence mode using a Canberra 15-element Ge solid-state detector. We gathered five spectra per sample through the energy range of − 130 to 340 eV in relation to the Zn K-edge energy (E0 = 9659 eV), with the following energy steps: 0.5 eV in the pre-edge (− 130 to − 35), 0.2 eV across the absorption edge region (− 35 to 65), and 1 eV in the post-edge region (65 to 340) with 1 s of acquisition time.

The scans were merged, aligned, and processed by subtracting the backgrounds using a linear pre-edge function between − 120 and − 45 eV and a linear or quadratic function between + 30 and + 110 eV from E0, followed by a flattening function in the post-edge region for normalization. Linear combination fitting (LCF) was then performed across the − 20 to + 80 region to identify singular Zn species and their relative proportions on the sample Zn spectrum, following the method described by Manceau et al. (2012). The Zn standards used in this study were previously reported in Khaokaew et al. (2012) (Fig. 5SM). All XANES analysis steps were processed in the Athena Software package (Newville 2001; Ravel and Newville 2005).

Results

Soil and amendment characterization

The amount of Pb present in the soil sample was around 16-fold higher, while the Cd content was around 20-fold higher than the reference quality values suggested by the Environmental Agency of the State of São Paulo (CETESB 2016). These values are considered hazardous as their concentrations exceeded the intervention values accepted for agricultural areas (Table 1). Moreover, the amount of Zn was around 14-fold higher than the reference quality values, surpassing the prevention value, which is a threshold in Brazilian legislation that indicates a high likelihood of damage to suitable soil ecosystem functions and groundwater quality (Table 1) (CETESB 2016). Pb and Zn were not present in the biochar, whereas biosolids had a significant amount of these elements and were slightly alkaline, and both had a high cationic exchange capacity (Table 2).

Table 1 Chemical properties1 of forest soil impacted by atmospheric deposition of mine-waste at distinct dose of phosphate, lime, biosolids, and biochar, after 60 days of incubation
Full size table
Table 2 Chemical properties and elemental composition of biochar and biosolids
Full size table

The phosphate amendments reduced the pH by 0.4 unit, thus concomitantly increasing exchangeable acidity likely because of NH4+ oxidation (i.e., from NH4H2PO4) (Dong et al. 2021), while lime increased pH by 0.5 unit and biochar by 0.7 pH unit (Table 1). Except for the phosphate addition, the amendments decreased the potential acidity (H + Al) and improved the alkalinity features of the soil. The total carbon contents were significantly increased by biochar (+ 120%) and biosolid treatment (+ 65%), which is expected because the biochar had ~ 65% of C in total elemental composition while biosolids had just 15% (Table 2). The highest phosphate dose increased the P available contents ~ 34 times, and biosolids increased 14 times at the highest rate. Lime and biosolid amendments increased Ca contents by ~ 70% and ~ 144%, respectively (Table 1).

PTE desorption kinetic and sequential extraction

The pseudo-first-order kinetic model reached a good fit to the values of cumulative Zn and Pb (Figs. 1 and 2) extracted by Mehlich-III solution, as shown by the high coefficient of determination (Table 3) and by the low values for reduced χ2 (Fig. 3SM). The parameters that estimate the constant velocity rate of kinetic (k1 (min−1), Table 3) were similar in our observations. In general, regardless of the soil amendment, 50% of the total Pb released was released after ~ 28 min, and 50% of the total Zn released was after ~ 19 min. The parameters that estimate the maximum amount of metal release at equilibrium, assuming time →  + ∞, describe the effect of amendment treatments on the metal’s solubility.

Fig. 1
figure 1

Experimental data (dots) of cumulative Zn desorption (% of the total Zn in the soil shown in Table 1) by Mehlich-III solution in a stirred flow system. Soil samples were collected in a contaminated forest (control) and after its remediation by A phosphate, B biochar, C lime, and D biosolids at different rates. The first-order kinetics was adjusted (see example in Fig. 3SM), and the estimated maximum amount of Zn desorbed (qmax) at equilibrium (time →  + ∞) is presented for each treatment. The qe values followed by the same letter, for each treatment and control, share the same 95% confidence interval. The “aa” later means that qe value of this treatment was higher than the qe value estimated for control. The complete results are shown in Table 3

Full size image
Fig. 2
figure 2

Experimental data (dots) of cumulative Pb desorption (% of the total Pb in the soil shown in Table 1) by Mehlich-III solution in a stirred flow system. Soil samples were collected in a contaminated forest (control) and after its remediation by A phosphate, B biochar, C lime, and D biosolids at different rates. The first-order kinetics was adjusted (see example in Fig. 3SM), and the estimated maximum amount of Pb desorbed (qmax) at equilibrium (time →  + ∞) is presented for each treatment. The qe values followed by the same letter, for each treatment and control, share the same 95% confidence interval. The complete results are shown in Table 3

Full size image
Table 3 Desorption kinetic parameters of Pb and Zn desorption of the contaminated forest soil (control) and after amendment application
Full size table

In the pseudo-first-order kinetic model, ⃰qmax means the maximum amount of metal (% of the total metal content presented in Table 1) desorbed at equilibrium, when time (min) →  + ∞; and k1 (min−1) is the velocity rate constants. ⃰ ⃰R2 is the coefficient of determination of the model to data fitting. The fitting example is shown in Fig. 3SM. The qe values followed by the same letter, for each treatment and control, share the same 95% confidence interval. The “aa” letter means that qe value of this treatment was higher than the qe value estimated for control.

Lime was the most effective amendment to decrease Zn desorption, reducing it by 17% at full rate, 21% at half rate, and 27% at 1/4 recommended lime dose. Conversely, the addition of phosphate in full rate increased Zn mobility by + 10% (Fig. 1 and Table 3) and was effective in decreasing Pb desorption, reducing it by 17% at full rate, 19% at half rate, and 22% at 1/4 phosphate recommended dose (Fig. 2 and Table 3). On the other hand, lime was not efficient at reducing Pb release at full or half dose. For Pb, the worst amendment was biosolids at a rate of 20%, which increased the Pb released by + 20%. At a 10% biosolid rate, however, the lowest Pb desorption was observed (Fig. 2 and Table 3).

The Zn sequential fractionation stressed only variations in the F1 fraction (Fig. 3), related to the exchangeable, more environmentally available fraction. The values observed in the F1 fraction increased with high phosphate and biochar rates. Lime treatments did not affect Zn fractions, and the addition of biosolids decreased the exchangeable Zn fraction over the rate increment, following the results observed in the desorption experiment (Fig. 1). Otherwise, the proportions of Pb throughout the sequential fractionation were not modified noticeably by soil amendments (Fig. 4SM), and the only exception was the decrease in exchangeable Pb fraction (F1) over the increase in phosphate rate (Fig. 4SM). All these results observed in sequential fractionation agreed with those observed in the desorption experiment (Figs. 1 and 2).

Fig. 3
figure 3

Relative allocation of Zn among the soil fractions as a function of amendments and rates. F1, exchangeable; F2, carbonate-associated; F3, organic-matter associated; F4, oxide; F5, residual. The relative distribution of Pb is shown in Fig. 4SM. Vertical bars denote stand errors (n = 3). Bars followed by different letters are different by LSD test (p < 0.05); otherwise, *ns, not significant

Full size image

Zn-XAS assessment

The LCF-XANES satisfactorily fitted the sample data, as observed in Fig. 4A, both visually by the low R factor value and by the sum value close to 1 (Table 1SM). The primary Zn species in the non-amended soil were Zn-kerolite (2:1 phyllosilicate, 57%), Zn adsorbed at gibbsite (31%), and in a minor part gahnite (ZnAl2O4, 11%) (Fig. 4B and Table 1SM). A strong reduction of Zn-kerolite fraction from 57 to around 30% was observed regardless of the soil amendments, after the incubation. We also observed an occurrence of Zn associated with Fe minerals (goethite and ferrihydrite), reaching 21% after lime treatment, 45% after phosphate, 25% after biochar, and 43% of the Zn-Fe fraction after biosolid reaction in soil. Phosphate and biosolid treatments eliminated the gahnite fraction and reduced the proportion of Zn adsorbed on gibbsite (Fig. 4B and Table 1SM). It was consistent with the decrease in Zn-Al fraction because 42% of Zn-Al was detected in the control treatment (i.e., unamended soil), while this fraction reached 28% after biosolid addition and just 24% after phosphate addition. Conversely, lime increased the Zn-Al fraction to 50%, while biochar maintained it around 45% (Fig. 4B and Table 1SM).

Fig. 4
figure 4

A Normalized Zn K-edge XANES spectra of contaminated forest soil (control) and after 60 days of incubation with the amendments (i.e., phosphate, lime, biochar, and biosolids). The LCF model fits over the energy range from 9639 to 9739 eV (− 20 to + 80 eV energy range relative to Zn K-edge (E0 = 9659 eV)). Black line means the experimental data, and red dotted line presents the LCF model fits. B Zn speciation chemical proportion based on Zn-XANES LCF. Details of Zn-standard references in Fig. 4SM. See details about the goodness of fit and uncertainty in Table 1SM

Full size image

Discussion

The Mehlich-III components react with the soil and quickly enhance ion mobilization by ionic exchange and ion chelation, which may offset in part the restrictions on time-limiting desorption. As Mehlich-III is an efficient extractor for assessing the mobility of PTE in mine residues and contaminated soils, it is employed for measuring their bioavailability (Plunkett et al. 2018) and the efficacy of reclamation strategies for hazardous elements’ immobilization (McNear et al. 2007). The desorption kinetics of Zn and Pb in the soil (Figs. 1 and 2 and Table 3) provided valuable insights into their mobility. The desorption of total metal contents was around 15% of the total Zn and 12% Pb. However, the time necessary to release 50% of the total amount of Zn (~ 19 min) is lower than that of Pb (~ 28 min) regardless of the amendments, suggesting that the immobilization features are different between these metals.

The metal adsorption/immobilization in soil has been explained by three molecular mechanisms: (i) nonspecific adsorption when metal is located in the diffuse layer, playing as counter-ions; (ii) specific adsorption on minerals or organic fraction surface complexation (Bowers and Huang 1986; Bradl 2004; Bolan et al. 2014; Uddin 2017); and (iii) at high pH, metal precipitating as hydroxides and (or) as carbonates dominate (Kinniburgh et al. 1976; Yong and Phadungchewit 1993; Bradl 2004). We speculate that Pb is more specifically adsorbed in soil solid surfaces than Zn, which is supported by some researchers that observed a selectivity sequence (i.e., sorption lyotropic series (Kinniburgh et al. 1976)), with Pb being preferentially adsorbed over Zn (Kinniburgh et al. 1976; Elliott et al. 1986; Gomes et al. 2001; Moreira and Alleoni 2010). This difference fate between Zn and Pb is stressed by sequential extraction because, even assuming the same Zn-Pb point source (i.e., mine dust), over time, they migrated to Zn, which had a prevalence with the organic fraction (Fig. 3), while Pb was preferentially associated with the residual fraction (Fig. 4SM). Thus, a possible Zn-Pb interaction on soil immobilization or leaching was lacking in experimental evidence. Future studies in µ-scale analysis (e.g., µ-XRF/XAS/XRD) should consider this possibility.

The addition of lime effectively decreased Zn availability (Fig. 1 and Table 3). We suggest that the increment in the pH values to 6.5–6.6 (Table 1) likely deprotonated the soil-solid surfaces and, consequently, improved the effective CEC (Table 1), thus improving the electrostatic affinity between soil particles and Zn ions. It is possible that the amount of Ca added to the system was not sufficient to compete with Zn for adsorption sites (Escrig and Morell 1998). On the other hand, Pb desorption was not affected in the same extension by lime treatment because its specific adsorption mechanism is less pH-dependent than Zn (Harter 1983; Bradl 2004), likely because Pb presents a higher electronegativity (Pb2+  = 1.87 arbitrary units) than Zn (Zn2+  = 1.65 arbitrary units) (Pan et al. 2022).

Phosphate amendment was able to decrease Pb desorption (Fig. 2), confirming our previous study, which demonstrated that phosphate and lime amendments decreased Pb mobility due to the formation of pyromorphite-like (i.e., Pb5(PO4)3Cl), a stable and insoluble mineral, as observed on Pb-XANES data (Gomes et al. 2022). An additional possibility is the establishment of ternary complex among “mineral surface–phosphate–metal” was formed, which could strengthen the stability of ion adsorption complexes through chemical bonds between metal and ligand surface (Elzinga and Kretzschmar 2013; Ren et al. 2015; Liu et al. 2016). On the other hand, the full dose of phosphate amendment increased Zn desorption (Fig. 1). This is likely due to the presence of ammonium (NH4H2PO4) in the P source, which after NH4+ oxidation to nitrate causes a decrease in pH values (Bouman et al. 1995; Hu et al. 2014). This increase in soil acidity increased the H + activity in the soil solution, consequently displacing Zn ions from the surface of soil colloids and increasing their mobility (Fig. 1).

Biochar and biosolids have high CECs and alkalinity properties (Table 2), which explain the Zn immobilization in their applications. Phosphate, biochar, and 5 and 10% of biosolids promoted a consistent decrease in Pb desorption (Fig. 2 and Table 3). We suggest that the forest soil studied here had a strong buffer capacity, likely due to the higher clay content (77%) than mine-waste impacted soil (13%, Gomes et al. (2022)), which is a crucial parameter for immobilizing soluble organics released by biochar and biosolids (Jardine et al. 1989; Kalbitz et al. 2000; Barreto et al. 2021). However, at a 20% rate, the addition of biosolids increased Pb desorption, probable surpassing the soil’s buffer capacity, increasing the metal mobility by soluble organo-Pb complexes. Evidence of the higher buffer capacity of the forest soil is supported by the results of sequential extraction fractionation (Fig. 3 and Fig. 4SM). For Zn, only the F1 fraction changed as a result of the amendments and rates applied, while the treatments kept similar features of the original control soil (Fig. 3). For Pb, the F2 fraction after phosphate addition showed a significant difference when compared to the control soil. Beyond clay content, the high content of soil organic matter present in the original soil also had a strong bearing effect on Zn (Fan et al. 2016) and Pb (Strawn and Sparks 2000). Silicate minerals and soil organic matter could prevent strong variations in the metal distribution, as observed over sequential extractions.

The LCF of XANES data suggests Zn-kerolite as the major species in the non-amended soil (57%), and ca. 30% after the treatments at the higher rates (Fig. 4). Zn-kerolite had been identified in previous studies (Voegelin et al. 2008, 2011; Jacquat et al. 2009) as a proxy for Zn-phyllosilicates. We suppose that this Zn-silicate fraction is part of the original mine dust that originally contaminated the area because the Vazante mine primarily contained Zn-silicates such as willemite (Zn2SiO4) in the ore (Monteiro et al. 1999; Babinski et al. 2005). Nachtegaal and Sparks (2004) observed the Zn complex developing at the goethite-coated kaolinite surface at pH 7.0, which is a system comparable to our real soil system, and after aging for 45 days, they concluded that precipitation at the kaolinite surface was thermodynamically preferred over adsorption to the goethite coating. Over the years of aging, any Zn released by mine dust would have been re-precipitated as Zn-rich phyllosilicate, such as Zn-kerolite species (Panfili et al. 2005; Nachtegaal et al. 2005). It is noteworthy that the stability of Zn-kerolite and other Zn-bearing phases was discussed by Jacquat et al. (2008), who observed that synthetic Zn-kerolite added to pristine soil was removed by 1 M NH4NO3 followed by 1 M NH4-acetate at pH 6.0, which is in agreement with the sequential extraction procedure (Fig. 3). Therefore, the changes in soil chemical equilibrium after soil amendment application likely led to a Zn-kerolite replacement for Zn-Fe species, especially after phosphate and biosolid amendments.

Due to the increased availability of phosphate and biosolids, we suggested that the Zn-Fe feature is a result of ternary complex formation, as previously observed in soil (Bolland et al. 1977; Agbenin 1998) and also in the purified system for ferrihydrite (Liu et al. 2016; Van Eynde et al. 2022). The correlation of this observation with desorption data of soil samples amended with phosphate (i.e., the highest rate) suggests that this new Zn-Fe phase is prone to solubilization; however, the data of biosolids does not totally support this assumption. On the other hand, biochar and lime preserved a high proportion of Zn-Al species, which are not associated with high (or easy) Zn desorption. The alkalinity produced by lime decomposition and biochar immobilization of H+ from the media (Xiao and Pignatello 2015; Ahmed et al. 2018; Zhang et al. 2018) likely contributed to the stability of Zn-Al species.

The high C content in the forest soil presented in this study (4 g kg−1 of total C, Table 1), might offer a stable and insoluble PTE-organic complex (Ahmad et al. 2016; Li et al. 2022). This stabilization is less prone in our previous study (1.5 g kg−1 of total C). Thus, soil management that decreases soil C content might increase PTE mobility. Beyond C, the heavily contaminated soil/sediment in the mine site presented a lower buffer capacity for Pb remobilization after biochar and biosolid addition, likely because of the lower proportion of Fe and Al oxyhydroxides that are strong sinks of soluble organic-mineral complexes. Thus, we outlined that the physiochemical properties such as mineralogy composition and soil C content, beyond the level PTE contamination, should be observed to predict the efficiency of remediation practices.

Environmental implications and conclusions

The use of soil amendments has been shown to enhance the adsorption of PTE, thus decreasing its easily mobile fraction. The selection of the most suitable soil amendment is initially predicted and/or tested in simplified systems, for example, observing the behavior of single contaminant in purified materials, such as clays and, or Fe/Al oxyhydroxides. However, the situation is more complex at real contaminated sites, where several possibilities of chemical reactions might happen, and often, there is more than one PTE that needs to be immobilized. In the case of the forest soil evaluated here, the results of PTE desorption by stirred flow indicate that only around 11% of the total Pb or Zn amount could be removed after long-term desorption. Consequently, 89% was immobile, which could be seen as a metal soil compartment of low risk for environmental implications.

Our observations suggest that phosphate addition kept Pb less available, in agreement with our previous study; however, it increased Zn desorption. Thus, it is necessary to be aware of the possible incompatibility of soil recommendations based on heavily contaminated mine sediments (i.e., contaminations in a few gram per kilogram) and recommendations destined for contaminated soil (i.e., contaminations in milligram per kilogram). Based on our studies, the addition of lime seems to be a suitable option as it decreases Zn and Pb releases. However, the life span of the lime reaction is limited by the acidification of the soil (Guo et al. 2010). It hinders a good prediction of soil remediation success over the long term. Aligning our studies, the most suitable strategy might be combining the soil amendments, for instance, initially applying phosphate to immobilize Pb by pyromorphite precipitation (Gomes et al. 2022), and then applying lime focused to reduce Zn solubility (Fig. 1). This alternative should be considered in future studies.

We conclude that lime is a suitable option for most soil remediation cases. We endorse the necessity to observe these amendments, or a combination of them, in long-term field trials to avoid overestimating laboratory results and to confirm the amendment efficiency against PTE remobilization by weather (e.g., precipitation) and vegetation life cycle (i.e., root exudation, litter decay, and biomass decomposition). Thus, it leads to a better decision on multiple PTE remediation for long-term scenarios.