1.

Introduction
Discharge and disposal of waste products contaminated with heavy metals have resulted in the contamination of valuable land resources and groundwater. Because heavy metals do not degrade and are toxic to biological systems, heavy metals will continue to be an environmental concern for a long time unless they are taken out from the ecosystem. Lead (Pb) is one of the many concerned heavy metals. The major sources of lead contamination of soil include lead mining and smelting activities, disposal of lead-based paints, and lead battery reclamation. At areas closed to rifle ranges, lead contamination may result from the pellets and bullets from firearms. The concentration of lead in uncontaminated soil is between 10 and 200 mg/kg (Lindsay, 1979). Urban soils show higher lead concentration than rural soils mainly because of motor vehicle emissions from using leaded gasoline (Davies, 1988). Elliott and Brown (1989) found that the lead concentration in the soils near an automobile battery recycling facility may be as high as 21% (w/w). Austin et al. (1993) reported that the lead concentration in the soil at an old smelter site may be as high as 30 000 mg/kg. Various techniques have been introduced to remediate metal-contaminated soils. One of these techniques is to separate the metals from soil by using chelating agents to form soluble metalchelate complexes. Chelating agents such as ethylenediaminetetraacetic acid (EDTA) have been shown to form strong metalligand coordination compounds and are highly effective in remediating leadcontaminated soils (Norvell, 1984; Elliott and Brown, 1989; Elliott et al., 1989; Brown and Elliott, 1992; Peters and Shem, 1992; Cline et al., 1993; Kim and Ong, 1998; Kim and Ong, 2000). Ideally, the minimum EDTA molar amount needed to extract lead from contaminated soil should be the same as the molar amount of lead in the soil. However, EDTA is a non-specific chelating agent and it reacts with other metals present in soil. Fig. 1 shows the change in conditional stability constants with pH for various metalEDTA complexes by assuming equal molar of metals present. As shown in Fig. 1, ferric ions may have a stronger affinity for EDTA depending on the pH of the soil or solution. In some soils, the molar amount of metals such as ferric and calcium ions may be larger than that of lead resulting in the formation of metalEDTA complexes rather than PbEDTA complexes. Therefore, in most studies, EDTA molar concentrations higher than the molar concentration of lead in soil are used to achieve maximum lead extraction from lead-contaminated soils.

Fig. 1. Effect of pH on conditional stability constants of metalEDTA complexes (adapted from Kim and Ong, 1999). (The general conditional stability equation is given by Ps=(CT,M)(CT,A) where Ps is the conditional stability constant, CT,M and CT,A are the total concentrations of the metal and the anion in all forms.) Figure options

Even though EDTA has been shown to be a suitable chelating agent for the remediation of lead-contaminated soils, not much information is available on the impact of the other metals present in the soil on the extraction of target metals such as lead with EDTA. The objectives of this study were to investigate the effects of major cations present in soils such as iron, aluminum, calcium, zinc, copper, magnesium, and manganese on the extraction efficiency of lead using EDTA. Several lead-contaminated soils from Superfund sites were used and variables such as solution:soil ratio and EDTAPb stoichiometric ratio were also investigated.

2. Materials and method

2.1. Sample characterization and preparation

Soil samples from lead-contaminated sites were used for this research. Three soil samples (identified as Soils A, B and C) were collected from Superfund sites in New Mexico. One soil sample (Soil D) was taken from a rifle range in Florida while Soil E was an oxidized glacial till from Iowa that was artificially contaminated with lead. Soil A came from a lead smelter area while Soil B came from an abandoned lead mining area. Soil C was obtained from a former battery recycling and smelter facility area. All the soil samples were air dried, screened through Number 25 (0.707 mm) sieve to remove large particles including lead pellets and organic debris before they were placed in a plastic containers with screw caps. The oxidized glacial till was contaminated with lead by mixing batches of 200 g of till with lead nitrate solution. The lead-contaminated soil was air dried, ground and sieved with a Number 25 sieve. The target lead concentration of the artificially contaminated soil was 2500 mg/kg. Each soil sample was characterized by measuring the soil pH, specific surface area, cation exchange capacity (CEC) and soil organic carbon (SOC). Soil pH was measured using an Accumet model 25 pH/Ion meter with a glass pH-indicating electrode and a calomel reference electrode. Soilwater ratio used for pH measurement was 1:2. The specific surface area of the each soil sample was measured using the ethylene glycol monoether (EGME) method (Carter et al., 1966) after the soil was treated with H2O2 to remove organic matter (Kunze and Dixon, 1986). The CEC of each soil was determined by saturating the exchangeable sites with sodium followed by substitution with magnesium ions (Polemio and Rhoades, 1977). SOC was measured using the method suggested by Nelson and Sommers (1982). Cations present in the soil such as lead, iron, aluminum, magnesium, manganese, calcium, copper and zinc were determined using Smith Hieftje 12 atomic absorption spectrophotometer after the soil samples were acid digested by boiling with concentrated HNO3. The total mass of available cations present in soil was assumed to be equal to the sum of the acid extracted cations (lead, iron, aluminum, magnesium, manganese, calcium, copper and zinc) which were assumed to be the dominant cations in soil. For iron ions, oxalate extractable iron concentration was also measured. The oxalate extractable iron generally reflects the amorphous iron present in the soil (McKeague and Day, 1966).

2.2. Extraction procedure

Three sets of extraction experiments were conducted to assess the effects of (i) solution:soil ratio, (ii) stoichiometric molar ratios of EDTAPb, and (iii) major cations on lead extraction efficiencies. A typical extraction procedure consists of placing 1 g of soil with a measured volume of EDTA solution in a 50 ml polypropylene centrifuge tube. Diluted HNO3 or NaOH solution was added, if needed, for pH adjustment. The slurry was then shaken with a Burrell wrist-action shaker for up to 15 days. Kinetic experiments previously conducted showed that 24 hours were sufficient to extract lead from the soil matrix (Kim and Ong, 2000). Studies by other researchers also indicate that 24 hours were more than sufficient to reach steady state conditions (Elliott and Brown, 1989; Peters and Shem, 1992; Cline et al., 1993). However, as shown later for one or two soils, the extraction period needs to be extended. The samples were then centrifuged for 30 min at 3000 rpm followed by filtration using 0.45 m membrane filter paper to remove particulates in the solution and the pH of the filtrate was measured. An adequate volume of the filtrate was set aside and preserved with 5% (v/v) nitric acid in a 100 ml volumetric flask. The concentrations of major cations were analyzed using an atomic absorption spectrophotometer. Separate experiments using Nanopure water at different pH values, adjusted with diluted HNO 3 or NaOH solution, were conducted to measure the solubility of the cations in water. All experiments were conducted in duplicate under room temperature. The solution:soil mass ratio on lead extraction efficiencies were assessed by using two soils, Soil D and Soil E (rifle range and artificially contaminated). The volumes of EDTA solution used were 3, 5, and 10 ml giving solution:soil mass ratios of 3, 5, and 10, respectively while the mass of soil used was 1 g. To study the effects of EDTAPb stoichiometric ratio, applied EDTA concentration varied from 0.0001 to 0.2 M giving an EDTAPb stoichiometric ratio of 0.1 to 100. All five soils were used to study the impact of EDTAPb stoichiometric ratio on lead extraction efficiency. The solution:soil ratio used was 10:1. The extraction time was 24 hours for both sets of the above experiments. For the study on the effects of major cations on lead extraction efficiency, the concentrations of EDTA used were 0.005, 0.002 and 0.003 M for Soil C (battery recycling), Soil D (rifle range) and Soil A (lead smelter) giving an EDTAPb stoichiometric ratio of 0.78, 1.0 and 1.0, respectively. In this set of experiments, the extraction periods were 24 hours and 7 days and the solution:soil ratio was 10:1.

3. Results and discussions

Selected physicalchemical properties and the major cation concentrations in the five soils are presented inTable 1. All the soil samples except for Soil B (lead mine) had a soil pH higher than 8. Soil B (lead mine) was acidic with a pH of 2.7. In addition, Soil B (lead mine) had a high amorphous iron content (31 700 mg/kg) and calcium content (103 900 mg/kg) while the other soils had relatively low amorphous iron content (between 300 and 500 mg/kg) and a calcium content of 700020 000 mg/kg. Soil D (rifle range) is a sandy soil with a low specific surface area.
Table 1. Soil properties and major cations concentrations for the soil samples Sample Pb Amorphous Fe 544 31 720 316 328 Mn Al (ppm) 25 940 707 14 330 2440 Ca Mg Cu Zn Specific surface area (m2/g) 9.7 37.3 4.4 15.4 0.7 CEC (meq/100 g) 11.6 17.8 4.3 17.5 5.8 % Organic carbon 0.75 2.26 0.10 2.52 0.18 pH

Artificiala Soil A Soil B Soil C Soil D a

2413 4180 1247 13 260 6238

384 91 2820 21

19 580 103 900 12 410 7450

6780 115 5270 436

129 557 30 279

1860 1078 86 70

8.16 8.55 2.68 8.13 8.47

Major metal cations were not measured.

Table options

Lead extraction efficiencies for different solution:soil ratios and EDTAPb stoichiometric ratios are present inFig. 2 for Soil D (rifle range) and Soil E (artificially contaminated). Extraction time for these experiments was 24 hours. The figure shows that the extraction with solution:soil ratio as low as 3:1 on a mass basis were similar to that of a solution:soil ratio of 10:1 for both contaminated soils. The extraction efficiencies were found to be dependent on the quantity of EDTA present. Since the wastewater generated from the extraction process should be treated before disposal, reducing the volume of wastewater would reduce the treatment costs and provide additional savings for the application of soil washing technology. It is interesting to note that the extraction efficiency for Soil D (rifle range) was gradual for higher EDTAPb stoichiometric ratio while for Soil E (artificially contaminated), the change in extraction efficiencies seemed to be quite steep for a small change in EDTAPb stoichiometric ratio. As shown by the results, caution should be used when studying metal extraction efficiency with artificially prepared soil since metal extraction by EDTA tends to be much easier than actual lead-contaminated soils even though artificially contaminated soil may facilitate consistency in the soil samples.

Fig. 2. Stoichiometric and volume ratio effects on lead extraction for Soil D (rifle range) and Soil E (artificially contaminated) (24-hour extraction period without pH adjustment). Figure options

Fig. 3 shows that the results of lead extraction from lead-contaminated soils and the artificially contaminated soil for EDTAPb stoichiometric ratios of between 0.1 and 100 over a 24-hour extraction period without pH adjustment. This figure demonstrates that lead extraction efficiency for each soil was different and that lead extraction was a function of the stoichiometric ratio of the applied EDTA concentration to the total lead concentration in the soil sample. However, if sufficiently large amount of EDTA was applied, all

the lead may be extracted for certain soils. Fig. 3 shows that for a unit stoichiometric ratio of EDTA to total lead concentration, approximately 75% of lead were extracted from artificially contaminated soil were extracted, 55% of lead from Soil C (battery recycling), 40% from Soil D (rifle range), 10% from Soil A (lead smelter) while none of the lead was extracted from Soil B (lead mine). The different lead extraction efficiencies for different soils at unit stoichiometric ratio may be due to the different soil and solution properties such lead species present in the soil sample and the cations present in the soil. EDTA is a non-specific chelating agent and therefore may react with metal ions other than lead. Because each metal ion has different reactivity with EDTA, the competition between lead ion and other metal ions is dependent on the dissolved concentration of the specific metal ion, dissolved anions, pH and the stability constant between the specific metal ion and EDTA.

Fig. 3. Impact of EDTA:lead stoichiometric ratio on lead extraction (10 ml of EDTA solution to 1 g of soil) (24-hour extraction period without pH adjustment). Figure options

As shown in Fig. 3, EDTA concentrations above unit stoichiometric requirement was needed for most soils to maximize lead extraction. For example, an EDTAPb stoichiometric ratio of at least 7 is needed to achieve the maximum lead extraction efficiency for Soil E (artificially contaminated) while a EDTAPb stoichiometric ratio of at least 20 was needed for the Soil A (lead smelter). Of the four actual lead-contaminated soils, lead extraction from Soil C (battery recycling) seemed to be easier than the other three leadcontaminated soils. In the case of Soil B (lead mine), lead was not extracted at all even for very high EDTAPb stoichiometric ratio (up to 30) over a 24-hour extraction period. The possible reasons for this lack of lead extraction are that the lead species present in the soil have very low solubility and that the lead was occluded within the different oxides in the soil matrix. It is also possible that the low pH of the soil may result in the precipitation of some of the applied EDTA or that the EDTA may have reacted with other metal ions present in the soil. To further investigate the possible causes, the molar amount of various extracted metals from Soil B (lead mine) using Nanopure water and 0.005 M EDTA (EDTAPb ratio=8.3) was investigated as presented in Fig. 4. The pHs of the solutions were 3.15 for the 0.005 M EDTA solution and 2.70 for the water. This figure shows that significant amount of iron was extracted with 0.005 M EDTA solutionthe amount of iron extracted was approximately equal to 90% of the applied molar amount of EDTA. For calcium the extracted molar amount with water and EDTA was about three molar times higher than the applied EDTA. For other metals such as zinc, copper, and magnesium, the dissolved amounts for both solutions were similar which means that the dissolved metal ions may be due to the low pH of the solution. Lead was not extracted at all with both water and 0.005 M EDTA solution. It is probable based on the results obtained that the presence of iron at low pH (see Fig. 1) may have played a role in suppressing lead extraction at low values. In a separate experiment (24-hour extraction), lead was not extracted at all from Soil B (lead mine) using 0.2 M EDTA solution (EDTAPb ratio=330) at an adjusted solution pH of 8.7 where the effects of iron with EDTA is minimized. Significant amount of calcium was found to be extracted. To investigate this further, 0.2 M EDTA solution was applied to Soil B (lead mine) for up to 15 days at a pH of 4.3. Approximately 10% of lead was extracted in 15 days (seeFig. 5). However, iron extraction from Soil B (lead mine) seemed to mirror that of lead extraction, both reaching a constant percent extracted after 13 days. As more iron was extracted, more lead was extracted. A possible reason for this observation is that lead ions may be strongly adsorbed or occluded in the iron preventing lead ions from complexing with the applied EDTA. Therefore, as the iron dissolved, more lead becomes available for complexation with EDTA.

Fig. 4. Molar amounts of metals extracted with water and 0.005 M EDTA solution at pH 3 (for Soil B (lead mine)) (stoichiometric ratio of EDTA Pb=8.3, 24-hour extraction period). Figure options

Fig. 5. Lead and iron extraction from Soil B (lead mine) with 0.2 M EDTA solution over time (stoichiometric ratio of EDTA Pb=330, 15 days extraction period). Figure options

To assess the effect of lead species on the lead extraction with EDTA, an effort was conducted to analyze the species of lead in Soil B (lead mine) using X-ray diffraction method. However, information on the lead species present could not be obtained due to detection limitation of analytical equipment. It is probable that the lead species present in Soil B (lead mine) had low solubility such as lead sulfide (Clevenger et al., 1991). The extraction results of Soil B (lead mine) suggest that lead extraction might be inhibited by a combination of factors such as type of lead species present, the location of lead within the matrix and possible competition with other metal ions presented in the soil. Fig. 6, Fig. 7 and Fig. 8 show the lead, iron, and calcium extraction efficiencies for Soil C (battery recycling), Soil D (rifle range) and Soil A (lead smelter). The EDTA concentrations used were 0.005, 0.002 and 0.003 M giving a EDTAPb stoichiometric ratio of 0.78, 1.0 and 1.0 for Soil C (battery recycling), Soil D (rifle range) and Soil A (lead smelter). Extraction was conducted with water and the EDTA solution for 1-day and 7-day extraction periods.

Fig. 6. Lead, iron and calcium extraction efficiencies for Soil C (battery recycling) with water and 0.005 M EDTA solution for different reaction times (solid lines are best fit lines). Figure options

Fig. 7. Lead, iron and calcium extraction efficiencies for Soil D (rifle range) with water and 0.002 M EDTA solution for different reaction times (solid lines are best fit lines). Figure options

Fig. 8. Lead, iron and calcium extraction efficiencies for Soil A (lead smelter) with water and 0.003 M EDTA solution for different reaction times (solid lines are best fit lines). Figure options

For Soil C (battery recycling), the percent of lead extracted around pH 6 after 7 days was 78% which was similar to the EDTAPb stoichiometric ratio applied (see Fig. 6). This may imply that for Soil C (battery recycling) most of the applied EDTA appeared to be complexed with lead around pH 6. However, much less lead was extracted for 1-day extraction period for pH greater than 6. As pH increased, kinetics seemed to be a factor controlling the extraction of lead from the soil. For pH value less than 6, the extraction efficiencies of lead were slightly lower but were similar for 1-day or 7-day extraction period. The lower pH may solubilize other ions such as iron and calcium which then may compete with lead for EDTA ligand sites. Note that the extraction of iron and calcium was similar for 1-day or 7-day extraction period. It is interesting to note that the total molar concentration of calcium dissolved at pH less than 8 was much larger than the applied EDTA molar concentration. Even at pH higher than 10, the dissolved molar concentrations of calcium ion was approximately 30% of the molar concentration of applied EDTA. However, as shown in Fig. 1, calcium and iron may not be a factor at these high pH values. The metal ion that may compete with lead for EDTA ligand sites at high pH is copper (Sommers and Lindsay, 1979). But copper concentrations in the soil tested were very low (see Table 1) in comparison to the lead concentrations. For Soil D (rifle range) at the 1-day extraction period and pH value less than 6, the percent of lead extracted was between 60% and 80%. After 7 days more lead was extracted showing that steady state conditions were not achieved after one day. The extraction efficiency of lead decreased gradually up to pH value of 8 where about 43% of lead was extracted after 7 days. An interesting observation is that the percent of lead extracted in water at low pH was similar to that of EDTA extraction. This implies that the majority of the lead extracted at low pH may be due to the pH of the solution. On the other hand the percent of iron extracted with water at low pH was much lower than that for lead. This may imply that the EDTA present may have assisted the extraction of iron more than that of lead. Unlike Soil C (battery recycling) where extraction of iron was completed in one day, more iron was extracted after 7 days just as for the lead. This result may imply that some of the lead may be occluded in the iron. As in the case of Soil A (lead smelter), only 50% of lead in soil were extracted in 7 days for pH values less than 6 for unit EDTAPb stoichiometric ratio. Soil A (lead smelter) had a significant amount of zinc which may compete with lead (see Fig. 1). Fig. 9 shows that the percent of zinc extracted was dependent on the pH. It is probable that part of zinc may inhibit the lead extraction at low pH values.

Fig. 9. Zinc extraction efficiencies for Soil A (lead smelter) with water and 0.003 M EDTA solution for different reaction times (solid lines are best fit lines). Figure options

The experimental results obtained seemed to suggest that the addition of EDTA changed the solution properties resulting in an increase in the dissolution of major metals. The dissolution of metals would result in a corresponding increase in anion concentrations. It is probable that the released anions may form soluble ion pair complexes with the dissolved metals. For example, calcium, magnesium and manganese can easily form ion pair complexes with phosphate, carbonate and sulfate ions in solution (Bohn et al., 1985). Another possible reason for high calcium and magnesium extraction by EDTA is that the iron oxides and hydroxides compounds which were dissolved by EDTA have high adsorption capacities for these metal ions. Therefore, the concentrations of the metals in solution increased correspondingly when the iron oxides were solubilized.

4. Conclusion

Based on the experimental results, it was shown that a solution:soil ratio as low as 3:1 had similar extraction results as that of a solution:soil ratio of 10:1 and that the extraction efficiencies were dependent on the molar concentration of EDTA present. Using different lead-contaminated soils, results of the experiments showed that lead extraction efficiencies for soils were different for a given stoichiometric ratio but if sufficiently large amount of EDTA was applied, most of the lead may be extracted for certain types of soils. To maximize extraction, the EDTAPb stoichiometric ratio needed varied from as low as 7 for artificially contaminated soil to as much as 20 for contaminated soils from a lead smelting facility. For one of the soils tested (Soil B (lead mine)), very low amounts of lead (<10%) were extracted even with an EDTAPb stoichiometric ratio of 300. The molar concentrations of all the extracted metals due to the addition of EDTA were more than the molar concentration of applied EDTA. For pH less than 6, dissolved ferric ions appeared to compete with EDTA, therefore, reducing the lead extraction efficiency. In addition, for pH value less than 6, other metal ions such as zinc and copper may compete with lead. Although the nature of the metal species present in the contaminated soils was not known, the experimental results appeared to indicate that besides competition other mechanisms such as occlusion of lead in the iron oxides and the type of lead species may play a role.

Abstract
Recycling chelant is a precondition for cost-effective EDTA-based soil remediation. Extraction with EDTA removed 67.5% of Pb from the contaminated soil and yielded washing solution with 1535 mg L1 Pb and 33.4 mM EDTA. Electrochemical treatment of the washing solution using Al anode, current density 96 mA cm2 and pH 10 removed 90% of Pb from the solution (by electrodeposition on the stainless steel cathode) while the concentration of EDTA in the treated solution remained the same. The obtained data indicate that the Pb in the EDTA complex was replaced by electro-corroded Al after electro-reduction of the EDTA and subsequently removed from the solution. Additional soil extraction with the treated washing solution resulted in total removal of 87% of Pb from the contaminated soil. The recycled EDTA retained the Pb extraction potential through several steps of soil extraction and washing solution treatment, although part of the EDTA was lost by soil absorption.

Keywords

Lead; Contaminated soil; Remediation; EDTA; Electrochemical treatment

1. Introduction
Recent decades have witnessed the rapid development of various technologies for the remediation of heavy metals contaminated soils. One of the most promising methods is soil washing with aqueous solutions of chelants. The advantages of chelants include a high efficiency of metal extraction, good solubility of the metal complexes formed and minor impact on the soil s physical, chemical and biological properties compared to acid soil extraction (Lim et al., 2004). For Pb contaminated soils, ethylenediamine tetraacetic acid (EDTA) has often been shown to be the most effective chelant (Lestan et al., 2008). In practice, however, the use of EDTA in full-scale is prohibited by a large volumes of waste washing solution generated, which must be treated before disposal. Effective treatment methods for waste washing solution, particularly recycling and reuse of EDTA, are needed. Although several EDTA recycling procedures have been demonstrated on a laboratory scale, there is currently no practical and commercially available method. Ager and Marshall (2001) used zero-valent bimetallic mixtures (Mg0Pd0, Mg0Ag0) to precipitate Pb from the solution while liberating EDTA in alkaline pH. Metals liberated from the EDTA complex were cemented to the surfaces of the excess magnesium or removed from the solution as insoluble hydroxides. The method is efficient but could be economically prohibitive. Hong et al. (1999) separated Pb from EDTA with Na2S, resulting in almost complete recovery of metals through precipitation in the form of insoluble metal sulphides. This method has found limited application due to the hazardous nature of the reagents and the sludge produced. Kim and Ong (1999)recycled chelant from Pb-EDTA solution by substituting Pb with Fe3+ in acidic conditions, followed by precipitation of the released Pb with phosphate near neutral pH. Fe3+ ions were then precipitated as hydroxides at high pH using NaOH, thus liberating EDTA. The process does not use expensive or hazardous reagents but it is complicated, with several operations involved and a slow kinetics of some reactions. Electrochemical technologies are simple and efficient methods for the treatment of many wastewaters, characterised by a compact size of the equipment, simplicity of operation, and low capital and operating costs (Chen, 2004). Johnson et al. (1972) reported that using a Pt anode in a conventional electrolytic cell oxidized EDTA into CO2, formaldehyde and ethylendiamine, and could thus potentially be used for treating waste soil washing solutions. However, simultaneous recovery of metals and chelant was not possible with this system. A two chamber cell separated by a cation-selective membrane was therefore proposed to allow liberation of metals from the complex and to prevent oxidation of EDTA at the anode. Metals, including Pb, were reduced and deposited onto the cathode and the EDTA was simultaneously recycled (Allen and Chen, 1993 and Juang and Wang, 2000). This method, however, is prone to operational problems such as membrane fouling and degradation. We recently proposed electrocoagulation for the removal of metals and EDTA from a washing solution obtained after extraction of primarily Pb contaminated soil (Pociecha and Lestan, 2009). In electrocoagulation, Al (or Fe) ions are generated from the sacrificial anode. The reactions for the electrochemical system at the Al anode are as follows (Eqs. (1) and (2)):

equation(1)
Turn MathJaxon

equation(2)
Turn MathJaxon

The nascent Al3+ ions are very effective coagulants and form large networks of Al-O-Al-OH flocks, with a large surface area and considerable absorption capacity (Shen et al., 2003). In our study Pb was almost entirely removed from the soil washing solution while, to our surprise, some EDTA remained in the washing solution. This early result indicated separation of Pb from EDTA. In the current work, the feasibility of electrochemical separation of EDTA and Pb in waste soil washing solution using an Al anode and a single-chamber electrolytic cell was studied.

2. Materials and methods

2.1. Soil properties
Soil contaminated with 3980 60 mg kg1 of Pb was collected from the 030 cm surface layer of a vegetable garden in the Meica Valley, Slovenia. The Meica Valley has been exposed to more than three hundred years of active lead mining and smelting. For standard pedological analysis, the pH in soils was measured in a 1/2.5 (w/v) ratio of soil and 0.01 M CaCl2 water solution suspension. Soil samples were analyzed for organic matter (as C content) by modified WalkleyBlack titrations (ISO 14235, 1998), cation exchange capacity (CEC) by the ammonium acetate method (Rhoades, 1982) and soil texture by the pipette method (Fiedler et al., 1964). The following values were obtained: pH 6.57, C content 8.2%, CEC 20.7 mval 100 g1of soil, sand 51.0%, silt 42.5%, clay 6.5%. The soil texture was sandy loam.

2.2. Soil washing

The extraction of soil with EDTA solutions was performed in two scales. To obtain the washing solution for the electrochemical treatment, we placed 0.5 kg of air-dried soil and 875 mL of aqueous solution (1:1.75 soil: washing solution ratio) of 75 mmol of EDTA (disodium salt) per kg of soil (43 mM EDTA), pH 4.3, in 1.5 L flasks. Soil was extracted on a rotating shaker (3040 GFL, Germany) for 72 h at 16 RPM. Approximately 400 mL of the washing solution was decanted from each flask after the soil was allowed to settle for 24 h. The decanted washing solution was filtered (filter paper was wide-pored, grade 388, density was 80 g m2). The same procedure was used to extract the soil with the recycled EDTA solution, except that centrifugation at 2880 g for 5 min and not decantation was used to separate the soil from the washing solution, to minimise solution loss. Fine particles were removed from the solution by filtration as described above.

2.3. Electrochemical treatment of the soil washing solution

The electrolytic cell consisted of an Al anode placed between two stainless steel cathodes (distance = 10 mm), the overall anode surface 63 cm2 and the surface area ratio between the cathodes and anode 1:1. The electrodes were placed in 500 mL of magnetically stirred soil washing solution in a 1.0 L flask. Current density was kept at 96 mA cm2, and the cell voltage measured with a DC power supply (Elektronik Invent, Ljubljana, Slovenia). The electrode cell was cooled using a cooling mantle and tap water (flow rate 250 mL min1) to keep the temperature of the treated washing solution below 35 C. The contact time of the electrochemical treatment was calculated as the ratio of the electrode cell volume to the volume of the washing solution and multiplied by the operation time (initially 30 min of operation time equalled 3.78 min of contact time). During the electrochemical treatment, the pH of the washing solution was regulated to pH 6, 8 and 10 by drop-vice addition of 5 M NaOH and HCl. Samples (20 mL) of washing solution were collected periodically and the pH and EC measured immediately. Samples were afterwards centrifuged at 2880 g for 10 min and the supernatant stored in the cold for further analysis of Pb and EDTA. The pellet (mainly Al hydroxide precipitate) was suspended in 200 mL of deionised water acidified with 37% HCl to pH 1.5. The resulting solution with some finely suspended

precipitate of EDTA, which is insoluble in acidic media, was stored in the cold for Pb determination. At the end of the electrochemical treatment, the cathodes were etched with 30 mL of 65% HNO3 to dissolve and later measure the concentration of electro-deposited Pb. The Al anode was weighed before and after treatment of the washing solution to determine the amount of electro-corroded Al. During the electrolysis, the surface of the Al anode was passivised by an oxide/hydroxide layer, which increased the potential between the electrodes (Mouedhen et al., 2008). In order to break down this passive layer and reduce the power consumption, we applied small amounts of Cl (as NaCl) when the voltage increased above 8 V (Chen, 2004). To prepare the recycled EDTA solution for soil extractions, we electrochemically treated the washing solution at pH 10 for 24 min (contact time) and separated the recycled EDTA solution from the Al hydroxide precipitate by centrifugation at 2880 g for 30 min.

2.4. Treatment of the soil washing solution with dosing Al-salt

A weight of 4110 mg of AlCl3 was dosed into 100 mL of the soil washing solution with pH 10 and gently stirred for 22.68, 45.36 and 113.4 min, which corresponds to 1, 2 and 3-times the total contact time of the electrochemical treatment, respectively. The amount of chemically dosed Al was the same as the molar amount of Al electro-corroded from the anode during electrochemical treatment. During the coagulation treatment with Al dosing, the pH of the washing solution was kept at pH 10, using 5 M NaOH. The precipitate was removed from the treated solution by centrifugation at 2880 g for 30 min, and the concentrations of Pb and EDTA in the supernatant measured. Afterwards, the pH of the chemically treated washing solution was adjusted to 4.3 and the solution reused for soil Pb extraction.

2.5. EDTA determination

The concentration of EDTA was determined spectrophotometrically according to the procedure of Hamano et al. (1993). The method involves the reaction of EDTA in washing solution with Fe3+ under acidic conditions to produce the Fe-EDTA chelate (transcomplexation), followed by the removal of excess of Fe3+ by chelate extraction in the aqueous phase using chloroform and N-benzoylN-phenylhydroxylamine and the formation of a chromophore with 4,7-diphenyl-1,10-phenanthroline-disulfonic acid. Using a spectrophotometer, absorbance was measured at 535 nm against a blank solution with the 4,7-diphenyl-1,10-phenanthrolinedisulfonic acid replaced with an equal volume of distilled water. The limit of EDTA quantification was 20 mg L1.

2.6. Pb determination
Air-dried soil samples (1 g) were ground in an agate mill, sieved through a 160 m mesh and digested in a glass beaker on a hotplate with 28 mL of aqua regia solution (HCl and HNO3 in a 3:1 ratio (v/v)) for 2 h at 110 C. Condensation of evaporating fumes was achieved via circulation of cool tap water through the glass tubes placed on top of the glass beakers. After cooling, digested samples were filtered through Whatman no. 4 filter paper and diluted with deionised water up to 100 mL. The pseudo-total concentration of Pb was determined by flame (acetylene/air) AAS with a deuterium background correction (Varian, AA240FS). The Pb in the solutions was determined by AAS directly. A standard reference material used in inter-laboratory comparisons (Wepal 2004.3/4, Wageningen University, Wageningen, Netherlands) was used in the digestion and analysis as part of the QA/QC protocol. The limit of quantification for Pb was 0.01 mg L1. Reagent blank and analytical duplicates were also used where appropriate in order to ensure accuracy and precision in the analysis.

2.7. Statistics
The Duncan multiple range test (Statgraphics 4.0 for Windows) was used to determine the statistical significance ( P < 0.05) between different treatments.

3. Results and discussion

3.1. Soil washing
Soil extraction with 75 mM EDTA per kg of dry soil removed 67.5% of Pb. The molar ratio between the Pb initially present in the soil and the EDTA in the washing solution was 1:3.9. It is known that even strong chelants such as EDTA cannot remove heavy metals

from the soil entirely, even at high molar ratios of EDTA vs. heavy metal concentration applied (Nowack et al., 2006). However, the Pb residual in soil after stringent soil washing with EDTA is encapsulated in soil minerals or strongly bound to the non-labile soil fractions and therefore essentially non-leachable and non-bioavailable (Udovic et al., 2009). The concentrations of Pb and EDTA in the soil washing solution before treatment in the electrolytic cell were 1535 and 12 444 mg L1 (33.4 mM), respectively. The pH of the washing solution before treatment was 7.91.

3.2. Electrochemical treatment of soil washing solution

Soil washing solutions were treated at various pH (6, 8 and 10). The pH of the solution tended to increase with treatment time, since the electrochemical system generated enough OH at the electrode to counteract the H+ released by the formation of Al hydroxides as a net final product (Canizares et al., 2006). Solution treated at pH 10 consequently required very little pH adjustment. The voltage between the electrodes also tended to increase with treatment time, regardless of the pH of the washing solution. The main reason for the voltage increase was the passivisation of the Al anode surface by formation of an insulating film of Al oxide (Mouedhen et al., 2008). In order to break down the passive film and thus to reduce the cell voltage surge and increase of power consumption, small amounts of Cl (as NaCl) were applied (Chen, 2004), to keep the voltage close to initial 8 V. The amount of Al consumed from the Al anode was 14.6 2.5, 12.0 0.4 and 9.4 0.4 g L1 of solutions treated at pH 6, 8 and 10, respectively. The amount of electrocorroded Al decreases with increasing pH. A higher aluminium current efficiency at higher alkaline conditions than at neutral is generally found in electrochemical systems (Chen, 2004). The electro-conductivity of the washing solution increased from an initial 6.1 to up to 10.0 mS cm1 (solution with pH 6). This increase followed the increasing concentration of charged Al hydroxide (electrolyte) (data not shown) during the electrochemical process. pH is an important operating factor influencing the performance of electrochemical processes (Chen, 2004). The effect of different pH of electrochemical treatment on the dynamics of Pb removal and precipitation from the washing solution and on the mass balance of Pb is shown in Fig. 1 and Table 1. During electrochemical treatment, metals complexed to EDTA could be removed from the soil washing solution by absorption on Al hydroxide flocks (electrocoagulation). Metals (M) could also be released from the EDTA complex after reduction reactions at the cathode (Juang and Wang, 2000), Eqs. (3) and (4). equation(3)
Turn MathJaxon

equation(4)
Turn MathJaxon

Fig. 1. Removal of Pb from the washing solution (A) and accumulation of Pb in the precipitate (B). Washing solutions were electrochemically treated at pH 6, 8 and 10. Error bars represent standard deviation from the mean value ( n = 3). Figure options Table 1. Balance of Pb after electrochemical treatment of the soil washing solution at different pH. Standard deviation from the mean value (n = 3) was calculated. Treated washing solution Pb balance (%) In solution pH 6 pH 8 pH 10 30 6 13 2 11 3 Precipitated 20 7 71 15 2 Electrodeposited 33 5 68 5 62 5 83 10 88 7 88 7 Table options

Metals liberated from the EDTA complex could then be removed from the solution by direct electrodeposition on the cathode, precipitation as insoluble hydroxides, or absorption and co-precipitation on Al hydroxide flocks according to the following reaction (Eq. 5): equation(5)
Turn MathJaxon

Table 1 indicates that, in all electrochemical treatments, the majority of Pb was removed from solution by electrodeposition on the cathode. However, at pH 8 and 10, a significantly higher amount of Pb was removed this way than at pH 6. This could again be

explained by a higher current efficiency of Al anode electrochemical systems at higher pH (Chen, 2004). The phenomenon presumably also explains the faster rate of Pb removal from the washing solution treated at pH 10 (Fig. 1). After treatment at pH 10, the EDTA remained almost entirely preserved quantitatively in the washing solution, while at pH 6 and 8, approximately one half of the initial EDTA was removed (Fig. 2). EDTA was presumably removed from the washing solution by electrocoagulation adsorption on Al hydroxide flocks, although some EDTA degradation by anodic oxidation (Johnson et al., 1972) might also occur. At pH 6 and 8, negatively charged EDTA complexes (M-EDTA2-) are probably partly absorbed on various monomeric and polymeric Al hydroxides species formed during the electrocoagulation process (Nowack and Sigg, 1996). These species, such as Al(OH)2+, Al(OH)2+, Al2(OH)24+, Al6(OH)153+, Al7(OH)174+, Al8(OH)204+, Al13O4(OH)247+ or Al13(OH)345+, have a long lasting positive charge before they finally transform into amorphous Al(OH)3 according to complex precipitation kinetics (Mouedhen et al., 2008 and Rebhun and Lurie, 1993).

Fig. 2. Concentration of EDTA in the washing solutions during electrochemical treatment at pH 6, 8 and 10. Error bars represent standard deviation from the mean value (n = 3). Figure options

Amorphous Al(OH)3 is a typical amphoteric metal hydroxide, which leads in alkaline conditions to the reaction shown in Eq. (6) (Chen, 2004). equation(6)
Turn MathJaxon

The formation of negatively charged Al hydroxides explains why, at pH 10, negatively charged EDTA was also not be removed from the washing solution by electrocoagulation (Fig. 2). Electrochemical treatment at pH 10 efficiently removed the majority of the Pb from the washing solution (Fig. 1), while the EDTA remained almost completely preserved in the washing solution (Fig. 2). As explained below, these data indicate the replacement of Pb from the complex with EDTA, removal of liberated Pb from the solution, and formation of Al-EDTA complex (trans-complexation). EDTA is a hexaprotic system. The degree of EDTA protonation and complexation with metals depends on the pH of the washing solution and the nature of the metal ions present. Al has a lower complex formation stability constant (Ks) than Pb (log Ks are 16.3 and 18.0 (at 20 C and ionic strength = 0.1) for Pb and Al, respectively, Martell and Smith, 2003). However, Al ions formed in abundant concentrations during electro-corrosion of the Al anode and Treacy et al. (2000) showed that the stability of Al-EDTA complex was higher in a solution with pH 9 than in solutions with pH 7 and 4. On the other hand, the stability of Pb-EDTA complex decreases in solutions with pH > 8 (Chang et al., 2007). The trans-complexation hypothesis, however, still needs to be proven. Our data indicate that using an Al anode enables electrochemical treatment in a conventional, simple, one-compartment electrolytic cell, without significant EDTA degradation. Presumably Al is oxidized at the anode (Eq. (1)) preferentially to EDTA oxidation, due to the high Al reactivity (electro-positivity). The standard electrode (oxidation) potential of the Al/Al3+ couple is 1.66 V (Evangelou, 1998). This is the first report on using this type of electrochemical system for EDTA recycling from Pb soil washing solution. The significance of pH for effective treatment was demonstrated.

3.3. Recycling and reuse of the treated EDTA soil washing solution
The efficiency of EDTA recycled from a washing solution electrochemically treated at pH 10 to extract Pb from the soil is shown in Fig. 3. After adjustment to pH 4.3 (pH of the fresh EDTA washing solution), the treated washing solution retained almost 90% of the Pb extraction potential (from original soil) compared to freshly prepared EDTA solution of the same molarity and pH. In this experiment (soil was extracted in two separate batches), fresh EDTA solution removed 63% of the Pb from the original soil and further extraction using the treated solution on the once extracted soil removed an additional 24% of Pb (Fig. 3). In total, using EDTA recycling, almost 90% of Pb was removed from the soil. Interestingly, the potential of the treated washing solution to extract Pb from previously (once) extracted soil was even higher than that of fresh EDTA solution, although the difference was not statistically significant (P < 0.05), Fig. 3.

Fig. 3. Removal of Pb from the original and previously extracted soil using fresh EDTA solution, EDTA solution after soil extraction but not treated, and electrochemically treated EDTA solution. Soil washing solutions were treated at pH 10 and, for soil extraction, the solutions pH was adjusted to 4.3. EDTA concentration was equal in all solutions (30 mM). Error bars represent standard deviation from the mean value (n = 3); letters (a, b, c) denote statistically different Pb removal potential within the categories of original and extracted soil, according to the Duncan test (p < 0.05). Figure options

Since we used a high molar ratio of EDTA against soil Pb (as usual in soil washing; Nowack et al., 2006) only part of the EDTA in the washing solution was complexed to Pb (approximately 22% calculated from data on Pb and EDTA concentration in the washing solution, section 3.1.), some EDTA was presumably left in the original form or in various stages of protonation. Spent non-treated washing solution was therefore expected to retain some Pb extraction potential, as indeed shown in Fig. 3. Table 2 shows the potential of recycled EDTA for Pb extraction from the original soil through several steps of soil extraction and washing solution treatment. First, fresh EDTA solution was used (in the 1st ext./treat. step) following by the use of electrochemically treated washing solutions (2nd and 3rd ext./treat. step) for soil extraction. Washing solution treated once retained 86% and solution treated twice 69% of the Pb extraction efficiency of fresh EDTA solution (calculated from data on percentages of Pb removed from the soil presented in Table 2). The decrease of Pb extraction potential can be explained by the loss of EDTA from the solution (Table 2), mainly due to EDTA absorption into the soil solid phase (Nowack and Sigg, 1996). Some EDTA (<10%) was also lost during the solution treatment phase, in which EDTA was either precipitated or anodically oxidized. Alternatively, EDTA could also be degraded by chlorine (Cl2) and hypochlorite (HOCl), which are strong oxidants and could be generated anodically following additions of NaCl into the electrolytic chamber to break down the anodic passive film (Chen, 2004).
Table 2. Potential of fresh and electrochemically treated EDTA solutions for Pb extraction from original (previously not-extracted) soil during three consecutive steps of soil extraction and washing solution treatment. Washing solutions were electrochemically treated at pH 10 and, for soil extraction, the pH was adjusted to 4.3. The percentage of lost EDTA after soil extraction (Lost EDTA extraction) and after electrochemical treatment of the solution (Lost EDTA treatment) was calculated. Soil extraction/Solution treatment 1. ext./treat. 2. ext./treat. Pb removed (%) 71 61 Initial EDTA concentration (mM) 43 29 EDTA conc. after extraction (mM) 32 23 EDTA conc. after treatment (mM) 29 22 Lost EDTA extraction (mM) (%) 26 19 Lost EDTA treatment (%) 9.9 6.4

Soil extraction/Solution treatment 3. ext./treat.

Pb removed (%) 49

Initial EDTA concentration (mM) 22

EDTA conc. after extraction (mM) 15

EDTA conc. after treatment (mM) /

Lost EDTA extraction (mM) (%) 32

Lost EDTA treatment (%)

Table options

3.4. Chemical dosing of Al

Electrochemical treatment of wastewaters with a sacrificial Al anode is an alternative to more commonly used chemical coagulation of pollutants by dosing Al-salts. To compare these two methods, the same amount of Al (as AlCl3) was dosed in the washing solution (pH 10) as the amount of Al electro-corroded during the corresponding electrochemical treatment. Chemical dosing (Table 3) removed very little Pb from the washing solution, much less than electrochemical treatment (Fig. 1, Table 1), but did remove some EDTA. The reason is that chemical coagulation with AlCl3 removed Pb with EDTA complexes, while electrochemical treatment liberated Pb and separated the EDTA. Chemical dosing did not lead to EDTA recycling; the Pb extraction potential of the chemically treated solution (with pH adjusted to 4.3) was even lower than the extraction efficiency of non-treated washing solution with the same pH (27 3% removed Pb,Fig. 3).
Table 3. Pb and EDTA removal from the washing solution after chemical dosing of AlCl3 at pH 10, and Pb extraction efficiency of chemically treated soil washing solution (after pH was adjusted to 4.3). The treatment time of chemical dosing was selected as a 13 multiple of the electrochemical treatment time (tel). Standard deviation from the mean value (n = 3) was calculated. Treatment time (min) 23 (1 tel) 45 (2 tel) 113 (3 tel) Pb removed (%) 35 45 93 EDTA removed (%) 34 9 27 4 26 3 Pb extraction potential (%) 26 2 23 6 21 0 Table options

3.5. Cost and safety considerations

An accurate evaluation of the costs associated with soil remediation would require a pilot-scale experiment. However, the cost of Al and electricity consumption, Al-hydroxide sludge disposal and treatment of the final spent washing solution (which represent the major part of the material costs) can be extrapolated from the obtained data. If we assume two re-cycles of EDTA washing solution treatment and reuse, then (including compensation for lost EDTA, Table 2) extraction of 1 ton of soil (with 75 mmol kg1 of fresh and recycled EDTA) would require 13.9 kg of EDTA. At a price of 1.3 per kg1 EDTA (information obtained from a major European EDTA producer) this translates into 18.1 . Treatment of the washing solution and EDTA recycling at a constant current density of 96 mA cm2 and an average voltage of 8 V would require approximately 115 kW h and, at an approximate cost of 0.1 per kW h, this translates into 11.5 . During the treatment/EDTA recycling, approximately 5 kg of Al would be expected to electro-corrode from the anode. The current prices of Al in the open market are below 1.6 kg1, which equals 8 for spent Al. During the process, approximately 20 kg of liquid Al hydroxide sludge was formed, calculated per ton of treated soil. The sludge can be deposited after treatment, i.e., after solidification (and stabilisation of metals) with cement. The disposal cost of solid hazardous waste transportation and disposal was assessed to be approximately 200 per ton (Meunier et al., 2006), which adds an additional 4 to the total cost. Since Al hydroxide is the major component of aluminium ore bauxite (together with AlO(OH) and Al2O3), it could perhaps be reused in the Hall-Hroult process to recycle aluminium and avoid disposal costs. The spent washing solution contains EDTA and Pb, which need to be completely removed before safe discharge. In a previous paper, we proposed an electrochemical advanced oxidation process using a boron-doped diamond anode for EDTA degradation and removal of Pb from treated solution by (electro) precipitation (Finzgar and Lestan, 2008). The electricity cost for this operation was estimated to be 10.3 ton1 of soil. The total estimated cost of major material expenses for the treatment of 1 ton of contaminated soil thus amounts to approximately 52 . This cost does not include capital investment in the equipment, which, in terms of electrochemical technologies, is characterised as relatively low (Chen, 2004). The cost seems favourable compared to the current cost of soil washing, which can go up to 350 per ton (Summergill and Scott, 2005). During the proposed remediation technique, some deposition of Al from the EDTA complex into the soil is expected. Aluminium is known to reduce plant growth on acid soils, in which Al3+ cations disturb root growth and function. However, Al constitutes aluminosilicate minerals and sesquioxides and is naturally present in the soil. It is harmless to plants, immobile and non-toxic in pHneutral soils (Andersson, 1988).

4. Conclusions
The following conclusions can be drawn from our study: Electrochemical treatment of soil washing solution obtained after EDTA extraction of Pb contaminated soil, using an Al anode in a conventional electrolytic cell at pH 10, efficiently separated EDTA and Pb. Pb was relatively efficiently removed from the treated washing solution (>85%, Fig. 1), mostly by electrodeposition on the cathode. Electrochemical treatment separated EDTA in an active form. We demonstrated that, after treatment, the EDTA solution retains almost all its Pb extraction potential. Less than 10% of EDTA was lost during electrochemical treatment. More EDTA was lost from the solution due to absorption onto the soil solid phases during soil extraction. Chemical dosing of Al was not effective in separating Pb and EDTA in the washing solution. We conclude, therefore, that electro-reduction of EDTA on a cathode (Eq. (3)) is essential for the exchange of Pb from the EDTA complex. Electrochemical treatment of the washing solution with an Al anode at alkaline pH has potential for cost-effective recycling and reuse of EDTA as a part of soil washing technologies.

Kinetic extractions to assess mobilization of Zn, Pb, Cu, and Cd in a metal-contaminated soil: EDTA vs. citrate

Jrme Labanowskia, Fabrice Monnab, Alain Bermondc, Philippe Cambiera, Christelle Fernandeza, Isabelle Lamya, Folkert van Oorta,
a

INRA, UR 251 Unit PESSAC, RD 10, F-78026 Versailles Cedex, France ARTeHIS, UMR 5594 CNRS, Univ. de Bourgogne Centre des Sciences de la Terre, Bat. Gabriel, F-21000 Dijon, France AgroParis Tech., Laboratoire de Chimie Analytique, 16 rue C. Bernard, 75231 Paris Cedex 05, France

Abstract
Kinetic EDTA and citrate extractions were used to mimic metal mobilization in a soil contaminated by metallurgical fallout. Modeling of metal removal rates vs. time distinguished two metal pools: readily labile (QM1) and less labile (QM2). In citrate extractions, total extractability (QM1 + QM2) of Zn and Cd was proportionally higher than for Pb and Cu. Proportions of Pb and Cu extracted with EDTA were three times higher than when using citrate. We observed similar QM1/QM2 ratios for Zn and Cu regardless of the extractant, suggesting comparable binding energies to soil constituents. However, for Pb and Cd, more heterogeneous binding energies were hypothesized to explain different kinetic extraction behaviors. Proportions of citrate-labile metals were found consistent with their short-term, in-situ mobility assessed in the studied soil, i.e., metal amount released in the soil solution or extracted by cultivated plants. Kinetic EDTA extractions were hypothesized to be more predictive for long-term metal migration with depth.

Keywords

Kinetic extraction modeling; Heavy metals; EDTA; Citrate; Metal mobility; Soils

1. Introduction
Fate of heavy metals in soils is of great environmental concern. They represent major risks regarding contamination of natural waters after release by metal-bearing soil constituents and migration via the soil solution downward to the water table (van Oort et al., 2006). Studying soil solution collected in situ from different horizons in metal polluted soils, Denaix et al. (2001) and Citeau et al. (2003) reported the presence of metals either as predominantly mobile, water-soluble forms (Zn, Cd) or as colloidal forms (Pb, Cu). Incorporation of metal pollutants in soils generally leads to changes in chemical speciation. Liberated metal ions are variously trapped by a wide range of reactive soil constituents, i.e. organic matter, iron and manganese oxides and hydroxides, phyllosilicates, phosphates, carbonates. A fraction of these metals can thereafter be (re)mobilized, either in dissolved or colloidal form before migrating downward. Mobilization is defined here as the potential capacity of metals to be transferred from the solid phase into the soil solution. It depends on the various links between metals and reactive sites of solid phase surfaces. Numerous operational chemical extraction methods, including one-step extractions (Chaignon et al., 2003 and Feng et al., 2005) and sequential extraction procedures (McGrath, 1996, Gleyzes et al., 2002 and Krishnamurti et al., 2002), have been developed to

estimate the extractable metal pools in relation to the metal-bearing soil phases. Although informative, such extraction procedures generally suffer from lacking selectivity with respect to the targeted chemical forms of metals. Alternatively, several authors have hypothesized that kinetic metal extraction data might better reflect dynamics of metals in soils (Gutzman and Langford, 1993, Bermond et al., 1998, Lo and Yang, 1999 and Bermond et al., 2005) and in sediments (Fangueiro et al., 2002, Fangueiro et al., 2005 and Gismera et al., 2004). For a given chemical agent, the kinetic extraction approach generates two kinds of data: (i) the proportion of metals extracted with respect to the total metal content of the soil sample; and (ii) the kinetic behavior of metals. It has been suggested that such kinetic extractions can be modeled by two first-order reactions, so-called TFOR model, which empirically define two different pools. Although chemical extractants used for experiments do not perfectly mimic natural conditions, the first pool of readily extracted metals, called labile by Fangueiro et al. (2005), and the second pool of more slowly extracted metals, called less labile, might be reasonably attributed to potentially mobile and/or bioavailable metal pools (Bermond et al., 2005). In this view, Degryse et al. (2006) showed that the proportion of labile metals was well-correlated with metal uptake in plants. Particularly for sandy textured soils, the distinction of two types of extractability with time is hypothesized to provide a good expression of the diversity of metal-soil constituent interactions, i.e., to various chemical bond stabilities. In soils with strong aggregation phenomena (i.e., clay soils), disaggregation phenomena due to chemical agents or physical stirring may, however, influence the extraction of discrete kinetic pools. Naidu and Harter (1998) stated that metals extracted by a mixture of organic acids are well-correlated with the mobile metal fraction in the soil solution. Organic acids are likely to be more representative of a mobile metal fraction that is available for biota. Low molecular weight organic acids, naturally exuded by plant roots or produced by microbial activity (Fox and Comerford, 1990), have been hypothesized to influence nutrient mobilization (van Hees et al., 2002) or translocation of metals in soil profiles (van Hees and Lundstrm, 2000 and Li et al., 2006). Ethylene diamine tetraacetic acid (EDTA) is a well-known strong chelating agent and has been widely used in agronomy for estimating the total extractable metal pool (Alvarez et al., 2006 and Manouchehri et al., 2006). EDTA was reported to remove metals organically bound, occluded in oxides, and associated with secondary clay minerals (Pay-Prez et al., 1993). In contrast, citrate has been reported to be one of the dominant organic acids in the soil solution (van Hees et al., 2002). Among the low-molecular-weight organic acids used to simulate metal mobilization, citrate presents a moderate metal complexation strength compared to EDTA, according to thermodynamic data from Sillen and Martell (1964)and Lindsay (1979) shown in Table 1.
Table 1. Formation constants (log Kf) for metal-EDTA and metal-citrate complexes at 0.1 M ionic strength Zn Me[EDTA]2 Me[citrate] Me2[citrate]22 Me[citrate]24 16.3 4.9 6.5 Pb 18.2 4.0 6.1 Cd 16.5 3.8 5.4 Cu 18.8 5.9 13.2 8.1

The log Kf values presented in this table are derived from those of Sillen and Martell (1964) and Lindsay (1979). They are recalculated to the same ionic strength. Table options

The present study aimed at estimating the labile fractions of different trace metals (Zn, Pb, Cu, and Cd) by applying kinetic EDTA and citrate extractions on samples from a moderately metal-contaminated agricultural soil (Labanowski et al., 2007). Comparison of EDTA and citrate extractions was used to quantify metal mobilization and to assess relationships between the energy of metal binding and the strength of the used chemical extractant. By expressing the data in terms of metal removal rates vs. time, we first checked the pertinence of fitting kinetic extractions with a TFOR model. Parameters provided by this modeling were then used to compare the different kinetic metal pools with results on metal mobility obtained from field studies (Citeau et al., 2003) and pot experiments (Mench et al., 2007) conducted on the same soil.

2. Materials and methods

2.1. Site conditions and soil sampling

The site selected for this study was an agricultural field located about 750 m from a former metallurgical complex in northern France (Elsass et al., 2002 and Labanowski et al., 2007). From 1901 to the late 1960s, the activity of a Zn smelter produced major atmospheric fallout (Thiry and van Oort, 1999), affecting agricultural land over an area of 3500 ha. The study site was a corn field representative of the most metal-contaminated agriculture areas, with reported total metal contents of 384489 mg Zn kg1, 89 124 mg Pb kg1, 16.618.9 mg Cu kg1, and 3.53.8 mg Cd kg1 (Chartier, 1998, Elsass et al., 2002 and van Oort et al., 2002). For understanding overall soil development, behavior, and metal distribution patterns, a large soil pit was dug and carefully sampled according to characteristic soil horizons as well as systematic sampling at 10-cm depth increments. For the present study, focusing on the plow layer, representative bulk samples of several kilograms between 0 and 28 cm were air-dried, sieved to <2 mm and homogenized before analyses.

2.2. Soil analyses

Characteristic physico-chemical parameters of the plow layer (particle size distribution, pH, CaCO3 content, CEC using cobaltihexamine) were determined by INRA's national soil analysis laboratory using standard methods (AFNOR, Agence Franaise de Normalisation, 1996). Total carbon content (Total C) was measured using NC 2500 C/H/N analyzer (NC Instruments). Total metal concentrations in the sample were determined by atomic absorption spectrophotometry (AAS), either flame or furnace (Varian SpectrAA 220) after HF/HNO3 (1:3, v:v) acid digestion under microwave heating (CEM MarsX) according to the digestion applications manual (CEM, 1998). HF (40%) and HNO3 (68%) were of analytical grade. Total metal concentrations were also measured on samples from subsurface horizons. Quality control included triplicate analyses, blanks and a standard reference soil (CRM 7001, CMI, Prague, Czech Republic). Measured concentrations of elements from the reference material always were <5% from certified values. Metal concentrations in blanks were below the detection limit.

2.3. Extraction kinetics

Extraction kinetics by EDTA or citrate were used on soil samples to study: (i) the maximal potential metal extraction capacity after 24 h; and (ii) the extraction rates of metals released as a function of time (between 0 and 24 h). For this, 1 g of soil was shaken with 10 ml of extracting solution for different periods (5, 15, and 30 min; 1, 2, 5, 11, 17, and 24 h) and then immediately filtered on a 0.45m cellulose membrane. For extractions >1 h, suspensions were centrifuged 5 min at 16,786 g before filtration. All EDTA and citrate extractions were made in three replicates and yielded 54 samples. Preliminary studies (not shown here) were conducted to optimize the experimental conditions. Finally, a 0.05 M Na2H2-EDTA and a 0.1 M (NH4)2H-citrate solution were found optimal for assessing the maximum extractability of the metals. Since many of the reactions with the extracting agents were pH dependent, all extracting solutions used in this work had a pH adjusted to 6.6 (soil pH) by adding adapted amounts of NaOH. Fangueiro et al. (2002) showed that a pH of about 6 guaranteed minimal variations of pH during extraction and hence, prevented competitive extraction by H +. Metal contents were determined as described above.

3. Results and discussion

3.1. Soil physico-chemical characteristics
The plow layer of the studied soil showed a slightly acidic pH and very low calcium carbonate content (Table 2). Other parameters, i.e., total carbon content (14.8 g kg1) and CEC (6.7 cmol kg1), were found to be typical for such sandy cultivated soils. Total metal concentrations reached 411 mg Zn kg1, 100 mg Pb kg1, and 3.6 mg Cd kg1. These values were much higher than those of the local geochemical background of the soils, estimated at 35, 16 and 0.2 mg kg1 for Zn, Pb, and Cd, respectively (Fernandez et al., 2007). The content of Cu (16.7 mg kg1) did not point to a marked industrial pollution. However long-term effects of fertilization and application of herbicides and fungicides may have contributed to the addition of anthropogenic Cu to the soil surface.
Table 2. Selected physico-chemical parameters of the studied soil (plow layer)

Particle size distribution (g kg1)

pH

CEC (cmol kg1)

Total Org. C (g kg1)

CaCO3 (g kg1)

Total metals (mg kg1)

Clay

Fine silt 76

Coarse silt 291

Fine sand 449

Coarse sand 86 6.6 6.7 14.8 2.5

Zn

Pb

Cu

Cd

98

411

100

16.7

3.6

Table options

Distribution of total Zn, Pb, Cu, and Cd concentrations between the Ap and the C horizon (Fig. 1) clearly showed their dominant accumulation in the plow layer. Below the Ap horizon, the concentrations of Zn and Cd decreased progressively indicating their migration until about 80-cm to depth, whereas Pb and Cu concentrations varied little. Such metal distribution patterns with depth are in agreement with comparable studies on other cultivated soils in the same area (Fernandez et al., 2007).

Fig. 1. Metal distribution patterns with depth in the studied soil profile (van Oort, unpublished data). Figure options

3.2. Proportion of metal extracted after 24 h

Examination of metal extraction evolutions with time showed that equilibrium was almost reached after 24 h (Fig. 2), except for Pb and to a lesser extent for Zn and Cd, when extracted with EDTA, which still showed some minor metal release at the end of the experiment. The percentages of extracted Zn, Pb, Cu, and Cd by EDTA were always higher than those extracted by citrate, although divergences were more marked for Pb and Cu than for Zn and Cd. These latter two elements can be considered as easily extractable metals, since both strong and moderately-strong chemical extracting agents were able to mobilize them in similar proportions, i.e., 4555% for Zn and about 7080% for Cd (Fig. 2). These results corroborate general statements that classify these two metals among the most mobile in soils, occurring frequently in appreciable amounts in the soil solution (Gray et al., 2003 and Citeau et al., 2003). For Pb and Cu, however, the extracted amounts differed, depending on the strength of the used extractant (Fig. 2). With EDTA, about 90% of Pb was removed from the soil. Such a value is in agreement with other works which frequently report lead extractions >60% with EDTA (Lestan et al., 2005). Yet Pb is widely considered to be a low-mobility or immobile metal in soils. This apparent contradiction clearly highlights the discrepancy between metal-mobilization studied in laboratory conditions and metal-mobility observed in field conditions. Solubilization of a metal with a chemical extractant therefore does not imply its mobility in the soil, but means that the metal is potentially mobile under certain physicochemical conditions.

Fig. 2.

Evolution of cumulative extracted amounts of Zn, Pb, Cu, and Cd metal vs. time with EDTA and citrate. Open circles: average of the results obtained from three replicates; error bars represent the 95% confidence level ( text for details). Figure options s.t( = 2; = 0.05, bilateral)/3). Thick grey line: modeled values (see

3.3. Kinetic aspects

In order to check the pertinence of the application of any first order model to kinetic extractions, the experimental results were expressed firstly in terms of metal removal rates per time unit. The amount of metal M extracted per weight unit of soil between ti and ti1, M (mg kg1), can be defined as: equation(1)

Turn MathJaxon

where CM(t), V and m represent the metal concentration at time t, the volume of extractant solution and the sample mass, respectively. Then, the average removal rate per time unit, AM(t), between ti and ti 1, for a given metal can be written as: equation(2)

Turn MathJaxon

Practically, the values AM(t) were computed for each lag time of the three independent replicates. The mean values and the 95% confidence intervals were calculated and plotted in a semi-log diagram (Fig. 3), which was another way for representing the data originally reported in Fig. 2. It appeared that the evolution of AM(t) generally occurs as two distinct linear segments with different, nonnull slopes and different intercepts on the Y-axis. Such clear differences of behaviour through time allow the variable AM(t) to be modelled by the sum of two distinct pools occurring simultaneously, each of them following an exponential decreasing function: equation(3)

AM(t)=AM1(0)M1t+AM2(0)M2t,
Turn MathJaxon

where AM1(0) and AM2(0) are the initial removal rates (at t = 0) of the metal M of the two pools and M1and M2 are their apparent rate constants.

Fig. 3.

Evolution of Zn, Pb, Cu, and Cd removal rate vs. time after with EDTA and citrate. Open circles: average of the results obtained from three replicates; error bars represent the 95% confidence level ( s.t( = 2; = 0.05, bilateral)/3). Thick grey line: modeled values (see text for details). Figure options

The extracted metal amounts gradually increased up to the end of the experiments with EDTA (Fig. 2) but quickly level off with citrate. Hence, it would have been tempting to model these latter using a single first order equation. As it can be seen in Fig. 3, and more particularly for extractions involving citrate, the representation using extraction rates vs. time reveals more clearly the presence of two metal pools than using cumulative amounts of extracted metals (Fig. 2). The metal removal rates appeared to be high at the start of the extraction, dominated by the parameter AM1(0) characteristic of the first pool, but the contribution of this first pool decreased quickly to become insignificant (exponential decrease governed by M1) (Fig. 3). Then, after a while, the contribution of the second pool became preponderant, with a clear slower extraction rate which still decreased with time (following M2). At time t, the total amount of metal extracted corresponds to the equation: equation(4)

Turn MathJaxon

and, after integration: equation(5)

Turn MathJaxon

Let us denote QM1 by AM1(0)/M1 and QM2 by AM2(0)/M2. Then Eq. (5) can be written as: equation(6)

QM(t)=QM1(1M1t)+QM2(1M2t)
Turn MathJaxon

Application of Eq. (6) to metal-extraction kinetics distinguishes two pools of metal, previously defined byFangueiro et al. (2005) as QM1 corresponding to the readily extractable metal fraction, called labile, associated with the apparent rate constant M1, and QM2 corresponding to the less readily extractable metal fraction, called less labile, associated with the apparent rate constant M2. In addition, the total non-extractable metal fraction, QM3, must also be considered. It corresponds to the difference between the total amount of metal in soil (Qtot) and the above mentioned pools: equation(7)

QM3=QTotQM1QM2
Turn MathJaxon

Calculations of kinetic parameters diverge to some extend from those proposed by Bermond et al. (1998) andFangueiro et al. (2005); both authors computing the parameters from the cumulative amount of metal extracted through time (Eq. (6)). In the present study, the parameters AM1(0), M1, and AM2(0), M2, were computed separately from both linear segments (Fig. 3). Each value was first weighted by using the inverse of its relative standard deviation, calculated on the basis of the three replicates. The parameters AM(0) andM were drawn from the Y-axis intercepts and the slopes, respectively by linear fitting to Eq. (3). From their standard deviations provided by Statistica 6.0 for Windows, a 95% confidence interval was calculated (given that two degrees of freedom were lost by each linear fitting). Then, the parameters QM were obtained by dividing AM(0) and M. Their 95% confidence intervals were calculated by combining individual errors ( Table 3 and Table 4).
Table 3. Initial extraction rates, AMi(0), kinetic parameters, Mi, and amounts, QMi, of metal extracted with EDTA per kg of soil for the two pools defined by the TFOR kinetic model AM1(0) (mg kg1 min1) Zn 21.525.5 M1(min1) QM1 (mg kg1) (% of total) 160192 AM2(0) (mg kg1 min1) 0.0580.068 M2 (min1, 103) 0.91.2 QM2 (mg kg1) (% of total) 5167 R2 QM1/QM2 Q3 (mg kg1) (% of total) 175

0.128 0.138

0.986

45% Pb 3.74.1 0.076 0.082 4652 0.0550.075 1.41.9

15% 3047 0.994 1.3

40% 10

50% Cu 0.720.79 0.118 0.126 5.76.5 0.0110.012 6.36.5

41% 1.71.9 0.964 3.5

10% 10

35% Cd 0.4170.423 0.155 0.157 2.62.7 5.4 104103 0.82.1

10% 0.20.8 0.941 7.5

55% 0.5

75%

10%

15%

Values correspond to confidence interval at a level of 95% (see text for details). Q3 has been evaluated by subtracting QM1and QM2 from the total metal amount per kg of soil. Table options Table 4. Initial extraction rates, AMi(0), kinetic parameters, Mi, and amounts, QMi, of metal extracted with citrate per kg of soil for the two pools defined by the TFOR kinetic model AM1(0) (mg kg1 min1) Zn 1822 M1(min1) QM1 (mg kg1) (% of total) 133169 AM2(0) (mg kg1 min1) 0.160.42 M2 (min1, 103) 3.67.0 QM2 (mg kg1) (% of total) 2277 R2 QM1/QM2 Q3 (mg kg1) (% of total) 200

0.127 0.142

0.981

3.5

35% Pb Almost instantaneous 3.84.8 5% 0.180.19 6.46.6

10% 2829 30% 0.994 0.2

55% 65 65%

AM1(0) (mg kg1 min1) Cu 0.280.42

M1(min1)

QM1 (mg kg1) (% of total) 1.72.7

AM2(0) (mg kg1 min1) 0.8 1021.2 102

M2 (min1, 103) 7.411

QM2 (mg kg1) (% of total) 0.61.4

R2

QM1/QM2

Q3 (mg kg1) (% of total) 13

0.138 0.167

0.954

15% Cd 0.190.22 0.134 0.151 1.31.6 5.9 1038.6 103 5.46.7

5% 0.91.4 0.991 1.4

80% 1

40%

30%

30%

Values correspond to confidence interval at a level of 95%. Q3 has been evaluated by subtracting QM1 and QM2 from the total metal amount per kg of soil. See text for details about the case of Pb. Table options

The metal amounts extracted by citrate after about 400800 min were very low so that the subtractionC(ti) C(ti 1) performed in Eq. (1) became inaccurate (Fig. 2). Therefore, the results for t = 1440 min were discarded for modelling citrate data. Besides, Pb extracted with citrate seemed to fit a single exponential decrease, even though the first extract was notably enriched by comparison to the general trend (Fig. 3). Consequently, this Pb case was treated as an instantaneous extraction pulse followed by a single exponential decrease. Finally, the total amounts of metal extracted with time have been simulated using Eq.(5) and compared to experimental observations ( Fig. 2 and Fig. 3) (we obtained 5 degrees of freedom for EDTA extractions because two times two degrees of freedom are lost by linear fittings, and 4 for citrate extractions because the last extract was discarded). R-square values were calculated on this basis. They provided an indication of how well the model fitted the data. All R2 were closed to unity, indicating that almost all of the variability has been accounted for with the variables specified in the model ( Table 3 and Table 4). Comparison between results of the two kinetic extraction methods indicated that both the labile and less labile metal pools were generally greater for EDTA than for citrate (Table 3 and Table 4). Only Cd does not fit this scenario since the less-labile fraction represented about 30% of the total Cd with citrate vs. about 10% with EDTA. A second general trend is that the labile metal fraction was generally greater than the less labile fraction for both extractants (QM1/QM2 > 1), except for Pb ( Table 3 and Table 4). The QM1/QM2 ratios of Zn and Cu were fairly similar for EDTA and citrate extractions, in spite of significantly different total amounts of metals extracted after 24 h. This suggests that the extracted fractions of Zn and Cu were bound to soil constituents with fairly homogeneous binding strengths for each metal. On the contrary, mobilization of Cd and Pb was markedly related to the extraction strength (i.e., the intensity of complexation measured by the log KF values, Table 1), notably in terms of proportion of labile and less labile pools. For Pb, the relative prevalence of the less-labile pool with citrate extractions (clearly highlighted by Fig. 3) suggests a slow and low mobilization under moderate extraction-strength conditions, consistent with generally low Pb mobility observed in soils. For Cd, a part of the fraction easily removed by EDTA becomes less labile with citrate ( Table 3 and Table 4): the configuration QM1: 75%, QM2: 10% with EDTA turns into QM1: 40%, QM2: 30%, maybe because, with citrate, more time is necessary to break links between Cd and soil constituents. Yet, at the end of the experiment, both EDTA and citrate have removed comparable amounts of Cd: 85% and 70%, respectively. In terms of kinetics, the labile pools of Zn, Cu, and Cd are extracted approximately at the same high apparent rate constant: M1 0.110.17 min1 whatever the extractants used and the total extracted metal amounts. This means that approximately 1020% of the remaining labile pool are removed per minute ( Table 3 and Table 4). Labile Pb seems to be less rapidly removed by EDTA (M1 0.070.08 min1). The calculation with citrate was not done for the reasons mentioned above. The M2 constants characterizing the less labile pool are much lower than the M1 constants. The M2 values of Zn, Pb, and Cd obtained with EDTA are comparable: 0.8 103 2.1 103 min1 while a similar homogeneity is observed for these metals, with citrate M2 = 3.6 103 11 103 min1. For both extractants, the M2 values of Cu are higher than those of other metals, suggesting that this Cu pool is associated with soil phases with low binding energy.

3.4. Operational use of kinetic extraction data in soil metal mobility studies
As already mentioned above, the two metal pools defined by the TFOR model have been attributed to mobile or bioavailable metal fractions (Bermond et al., 2005 and Degryse et al., 2006). Significance of kinetic chemical extractions for providing detailed insight on

metal speciation is still questionable yet, because different metal mobilization kinetics are observed following the reagent used. Studies connecting results from the kinetic approach with real-world investigations on metal mobility are cruelly lacking. In the present case, data issued by both approaches are interestingly available since complementary data on metal mobility and bioavailability were obtained for the soil of our study by experimental field work, and by laboratory pot experiments. Citeau and coworkers collected periodically over three years gravitational soil water by lysimeters placed at the bottom of the Ap horizon, and assumed to explore a soil volume of about 7 L (Citeau et al., 2001, Citeau et al., 2003 and Citeau, 2004). Results are reported in Table 5. Exposure of different cultivars to metal polluted soil was also studied (Chartier, 1998 and Mench et al., 2007). Green salad, carrot and rye grass were cultivated in 3 replicates using 10-kg pots filled with plow layer samples sieved to <5 mm. These cultivars were harvested at the usual growth stage (about 30 days). They were carefully washed with distilled water, dried at 60 C, ground and analyzed for total metal contents (Mench et al., 2007) (Table 5). Metals display approximately the same behavior in both experiments. Percentages of Zn and Cd extracted monthly from the Ap horizon by soil solutions or plants are remarkably close. They vary only within one order of magnitude or less and are about 1020 and 100200 times higher than those of Cu and Pb, respectively (Table 5). Interestingly, the same pattern is observed for the percentages of readily citrate-labile pools: Zn Cd > Cu > Pb (Table 4). Hence, labile metals extracted by citrate might be representative of short-term metal dynamics (solid-liquid transfer, plant-availability). In contrast, there is less resemblance of behavior between the readily labile pools using EDTA and real-world experiments, in particular for lead. This can be ascribed to the high extraction strength of EDTA (Table 1) which appears out of proportion with respect to natural soil conditions. As a consequence, EDTA rather provides an upper limit of metal mobility and can be a valuable parameter to predict long-term transfer in soils.
Table 5. Experimental data on metal mobility determined in the soil solution and by plant uptake Zn Average % of total metal extracted from the soil by the soil solution over 1 month* 0.25 Pb 0.0005 Cd 0.17 Cu 0.01

Average % of total metal extracted from the soil by plant over 1 month** Green salad Rye grass Carrot 0.71 1.39 0.26 0.002 0.008 0.001 1.54 1.05 0.25 0.09 0.42 0.05

*Mean values compiled from lysimeter data of metal amounts in, and volumes of gravitational water ( n = 31) leaching out of the Ap horizon (from Citeau, 2004); **mean values compiled from pot experiment data on metal-uptake by three cultivars with replicates (n = 3) (from Chartier, 1998 and Mench et al., 2007). Metal extraction was expressed as a percentage of total metal content in the bulk sample. Table options

4. Conclusion
The representation of EDTA and citrate kinetic extractions in terms of removal rate confirmed the existence of two kinetic metal pools, a labile and a less-labile pool, which can be modeled by two first order reactions. Zn and Cu showed similar kinetic behaviors (in terms of QM1/QM2 ratios) regardless of the extraction strength, contrarily to Cd and Pb which were markedly related to the used extractant. These results suggest that Zn and Cu have a more homogeneous binding strengths (with one or more soil constituents) compared to Cd and Pb.

QM1-citrate values were considered to be a good indicator for short-term metal mobility as they mimic metal mobilization in field and
experimental conditions and are consistent with the frequently observed mobility scheme: Zn Cd > Cu > Pb. EDTA extracted greater amounts of metals from the soil compared to citrate. It may therefore provide an indication for the maximum potential metal extractability, helpful for the prediction of long-term risks.

Acknowledgements

J. L. thanks the French National research Program ToxNuc-E and INRA for a 16-month post-doc fellowship. C.F. received a grant from the French ministry of Research and Technology (MRT). The authors gratefully acknowledge Sbastien Breuil for help in so il sampling and chemical extractions. We would like to address our special acknowledgements to the reviewer for his pertinent comments and suggestions, greatly improving the manuscript.