Representative soil samples were collected from the yards of 3 Pb-paint contaminated homes from Jersey City and another three homes from San Antonio. Surface soil samples (0–6 inches) were collected using grid sampling. Owning to the sizes of yards, 12 samples were collected from each of the homes in Jersey City, whereas 44, 50, and 54 samples were collected from the three homesites in San Antonio.

In the laboratory, soil samples were air-dried and passed through a 2 mm sieve before characterization. Pb concentrations in all soil samples were measured and recorded on-site, using a portable X-ray Fluorescence Analyzer (p-XRF, Thermo Fisher Scientific, Niton XL3t) and in the laboratory using Inductively Coupled Plasma-Optical Emission Spectroscopy (ICP-OES, Agilent Technologies 5100) after acid digestion following USEPA Method 3050B (USEPA, 1996). Afterwards, the soil samples collected from individual houses were combined to make a composite soil sample for other analysis and EDDS extraction tests. Total concentrations of major metals (Al, Fe, As, Cu, Zn, Cr, Cd, and Ni) were determined by ICP-OES after the acid digestion. The pH and electrical conductivity (EC) of soil samples were assayed using standard protocols described in Sparks et al. (2020), and the organic matter content in samples was determined using the loss on ignition (LOI) method described in Schulte and Hopkins (1996). All extractions and analyses were done in triplicate. The Quality Assurance/Quality Control protocols in Soil Sampling Quality Assurance User's Guide (Barth et al., 1989) and Acid Digestion of Sediments, Sludges, and Soils (USEPA, 1996) were followed for soil sampling and analysis in this study.

To optimize the concentration of EDDS needed to solubilize the bound Pb to plant-available forms, EDDS solution at different concentrations (i.e., 2, 5, 10, 15, 20, and 30 mmol/L) was applied to 1 g of a dried composite soil sample from each site. Plant-available forms of Pb, which were water-soluble and exchangeable Pb, were quantified after EDDS application.

Sequential extraction of Pb contaminated soil (composite soil samples) before and after EDDS treatment was done using Tessier et al. (1979) method with a few modifications. Six geochemical fractions of Pb (i.e., water-soluble F-1, exchangeable F-2, carbonate bound F-3, oxides bound F-4, organic bound F-5, and residual silicate bound F-6) in soil samples were determined.

EDDS extractions were done using 2 and 30 mmol/L solutions. Soil samples before and after EDDS extraction were subjected to sequential extraction. Briefly, (1) 1 g oven-dried soil sample was diluted with 15 mL deionized water, shaken at 250 rpm for 2 h, and centrifuged at 3,500 g for 30 minutes to obtain the F-1 fraction. (2) The residual soil was extracted using 8 mL of 1 M MgCl₂ for 1 hour at room temperature for the exchangeable Pb fraction (F-2). (3) Extraction of the F-3 fraction from the residual soil from step 2 was done using 8 mL of 1 M CH₃COONa at pH 5.0 (adjusted using CH₃COOH) for 5 h. (4) The oxide-bound Pb (fraction F-4) was extracted from the residual soil obtained from the previous step with 20 mL of 0.04 M NH₂OH·HCl in 25% (v/v) CH₃COOH at 96°C for 6 h. (5) To extract the organic bound Pb (fraction F-5), 3 mL of 0.02 M HNO₃ and 8 mL of 30% H₂O₂ (pH 2.0, adjusted with HNO₃) were added to the soil from step 4 and the samples were incubated at 85°C for 5 h. After cooling, 5 mL of 3.2 M CH₃COONH₄ in 20% (v/v) HNO₃ was added. (6) The residual silicate bound Pb (F-6) was then obtained by digestion of the residual soil sample from step 5 with concentrated HNO₃ at 105°C. After each step, the supernatant was analyzed for Pb using ICP-OES. The soil samples from each step were weighed and used for the next step in the sequential extraction procedure.

The data were statistically analyzed using the JMP statistical software package (JMP 14, SAS Institute Inc.). A Tukey-Kramer HSD test was used to test whether group mean pairs were significantly different at α = 0.05.

The chemical properties of the soil samples collected from lead paint-contaminated homes in Jersey City and San Antonio are shown in Table 1. The Jersey City soils were slightly acidic with pH values ranging from 5.32 to 6.14, while the San Antonio soils were slightly alkaline with pH values ranging from 7.48 to 7.74. San Antonio soils contained more salts (indicated by higher EC values) and slightly more organic matter compared to Jersey City soil. Total Al and Fe concentrations are shown in Table 1. Since the concentrations of other metals (e.g., As, Cu, Zn, Cr, Cd, and Ni, listed in Supplementary Material) were much lower than Fe and Al, we primarily considered the effects of Al and Fe on Pb fractionation in this study. San Antonio soils had higher levels of Al compared to Fe, whereas the Jersey City soils had higher levels of Fe compared to Al. Figure 1 shows that the overall Pb concentrations in Jersey City soil were lower than those of San Antonio soils. The maximum detected Pb concentration in Jersey City soil was 2,696 mg/kg, while that value for San Antonio soil was 14,721 mg/kg. Pb concentrations in soil samples varied spatially in each yard but showed a generally declining trend with increasing distance from the wall (Figure 2). This Pb distribution trend suggested that the major source of Pb in residential soil was from the exterior Pb-based paint (Litt et al., 2002; Clark and Knudsen, 2013), and more Pb was accumulated in the soil that was close to the residential wall after the deterioration of the paint.

The performance of p-XRF for in situ soil Pb measurement was evaluated by comparing p-XRF data with ICP-OES data. The average Pb concentrations of triplicated samples measured by both methods are shown in Figure 3. Strong positive correlations are indicated by R² ranging from 0.81 to 0.97 for Jersey City soil and from 0.93 to 0.94 for San Antonio soil. These R² values were consistent and show a better correlation compared to the reported value of 0.89 in Maliki et al. (2017). Previous studies have used p-XRF analysis to assess metal contamination in a variety of soils. Reports indicate that the accuracy of p-XRF measurements is variable, based on soil moisture, organic matter content, the source of the metals, as well as the types of metals being analyzed (Caporale et al., 2018; Ravansari et al., 2020). Moreover, Caporale et al. (2018) showed that the regression fits of the p-XRF data to those obtained after extraction and mass spectrometric analysis are site-specific, and hence prediction models need to be developed for soil samples from a particular site. In the current study, the fittings for all sites showed good correlations, except for JC1 which had a relatively lower R² at 0.81, indicating the differences in soil properties between Jersey City and San Antonio did not significantly affect the accuracy of p-XRF measurements. The regression slops for four of those sites (i.e., JC3, SA1, SA2, and SA3) were similar at around 0.8, and a better correlation at JC2 showed a slope of 1.1. Thus, our data show that p-XRF is a useful tool for rapid on-site measurements for field studies.

The plant-available form of Pb in the soil is the key and limiting factor to facilitate phytoextraction. Using vetiver grass as an example, with a massive root system, it was reported to accumulate Pb in root at approximately 20,000 mg/kg dry weight without showing phytotoxic symptoms in hydroponic systems (Chen et al., 2012). However, when planted in Pb-contaminated soil, the average Pb accumulation in vetiver grass root was reported to be below 60 mg/kg (Attinti et al., 2017). Attinti et al., 2017 found the solubility of Pb in soil could be enhanced by the addition of EDDS, and the accumulation of Pb in vetiver grass root increased by over 70%. Thus, the transformation of Pb from bound forms to plant-available forms (i.e., water-soluble form F-1 and exchangeable form F-2) after EDDS treatment could be employed to evaluate the potentials of EDDS in enhancing phytoextraction of Pb.

Figure 4 shows that the plant-available forms of Pb consisted of 0.5 to 3.9% of the total Pb in composite Jersey City soil samples. After extracting with EDDS at 2 mmol/L, there was a large increase in plant-available Pb that ranged from 21.9 to 34.2%. Although the ratios of plant-available Pb increased with increasing EDDS concentrations (2 mmol/L to 30 mmol/L), the plant-available Pb levels mobilized when the soils were treated with EDDS concentrations below 20 mmol/L were much lower than at 20 mmol/L. Significant differences were also seen at 30 mmol/L EDDS extraction compared to that of 20 mmol/L. The maximum plant-available Pb mobilized after EDDS treatment accounted for 44.8 to 61.2%, depending on the location, after EDDS extraction at 30 mmol/L.

Compared to the Jersey City soils, lower plant-available Pb concentrations were observed in San Antonio soils, ranging from only 0.1 to 1.4% without EDDS extraction (Figure 4). After extraction using 2 mmol/L EDDS, a 36% increase in plant-available Pb was observed (Figure 4). Higher concentrations of EDDS increased the average percentages of plant-available Pb over total Pb in soil; however, in contrast to Jersey City soils, there were generally no significant differences in Pb mobilization beyond 10 mmol/L EDDS concentrations. With EDDS extraction in this study, up to 68.9% of total Pb in San Antonio soil samples was converted to plant-available forms. The increasing trend in mobilized Pb in Figure 4 with increasing concentrations of the chelating agent is consistent with literature that used chelating agents in Pb remobilization (Soltan, 2006).

Higher levels of plant-available forms of Pb in soil from Jersey City compared to those of San Antonio soils may be explained by the difference in soil pH values. Generally, the mobility and bioavailability of Pb increase with the decrease of pH (Kushwaha et al., 2018). Sauve et al. (1998) reported a strong negative correlation between dissolved Pb concentrations and pH under acidic conditions, and a near-constant and lowest concentration of dissolved Pb was shown for pH values between 7 and 8. Lower soil pH may also facilitate the conversion of other forms of Pb, such as carbonate bound Pb to soluble forms (Schock, 1980), which enhanced the fraction of plant-available Pb in Jersey City soil samples (Kushwaha et al., 2018). The pH differences between Jersey City and San Antonio soil samples could also explain the trends for the conversion of Pb forms; the plant-available Pb ratios for Jersey City soil did not change much at lower EDDS concentrations ranging from 2 to 10 mmol/L, while those ratios increased significantly for San Antonio soil. This could be explained by better Pb extraction performance by EDDS under alkaline conditions, especially when there was an EDDS deficiency situation (Yan et al., 2010).

By tracking the changes in the bound forms of Pb in Jersey City and San Antonio soils using sequential extraction before and after EDDS treatment, the impact of EDDS in mobilizing the various fractions of soil Pb could be estimated. For this experiment, EDDS treatment was performed at 2 mmol/L and 30 mmol/L concentrations. In general, a decreasing trend was seen in the various Pb fractions after EDDS treatment, in particular in the F-3, F-5, and F-6 fractions (Figure 5). Lead in F-1 and F-2 fractions were negligible for all soil samples before EDDS treatment, which corresponds to the low plant-available Pb seen in Figure 4. The majority of Pb existed in the residual silicate bound form (F-6), indicating that most of the Pb were recalcitrant with low solubility, which would be difficult to remediate (Elless et al., 2007). Compared to the Jersey City soil samples, there was more Pb within the F-3 and F-5 fractions of the San Antonio soil samples, which can be explained by the higher pH and the higher organic matter content in these soils respectively (Table 1).

Soil Pb mobilization by chelating agents such as EDTA and EDDS has been previously reported. In both acidic and alkaline soil conditions, chelating agents have been reported to primarily reduce organic-bound Pb (F-5) among all Pb fractions while enhancing the water-soluble and exchangeable Pb fractions (Li et al., 2005; Yan et al., 2010; Chen et al., 2012). Li et al. (2005) also indicated a decline in oxide-bound Pb (F-4) and residual silicate bound Pb (F-6) Pb fractions using EDTA as the chelating agent. Our results show that while the Pb in the F-4 fraction increased with the 2 mmol/L EDDS treatment, at the higher dose of 30 mmol/L, there was a decrease in Pb concentration in the same fraction. It is likely that with EDDS application at 2 mmol/L, the dissolution of certain compounds (e.g., organic matter) induced by EDDS released Pb which could then be re-adsorbed to the metal oxides, leading to the increase in F-4 bound Pb (Figure 5). This assumption is further supported by the presence of high concentrations of Fe and Al in soil samples from both Jersey City and San Antonio, which can induce Pb-binding (Table 1) (Yan et al., 2010). Cao et al. (2008) illustrated that higher EDDS dosage (10 mol EDDS:1 mol Pb) could reduce ratios of metal oxides bound Pb compared to low dosage (1 mol EDDS:1 mol Pb), which may explain the decrease in the F-4 fraction after EDDS application at 30 mmol/L.

The silicate-bound Pb fraction (F-6) has been reported as the most difficult fraction to mobilize, while other Pb fractions (i.e., F-3, F-4, and F-5) could be transformed into plant-available forms. However, our results show that that 56.1–77.0% of the Pb in the F-6 fraction was mobilized by 30 mmol/L Pb in Jersey City soils, while the percentages for San Antonio soils were 67.2–86.8% (Figure 5). Since the Pb concentration in the F-6 fraction generally decreased with an increase in EDDS concentrations as shown in Figure 5, the above percentages could be even higher if the EDDS concentration is further increased. However, in real-world applications, mobilization of Pb needs to be balanced with plant uptake to prevent the leaching of Pb into surface and groundwater. An alternative would be to use a coated chelating agent, which slows down its release. In doing so, Pb mobilization could be balanced by plant uptake during phytoremediation (Li et al., 2005).