*3.1. Soil Characterization*

The initial pH of the soil was close to neutral value (7.21) and the LOI index was 7.01%. The study of the particle size distribution resulted in classifying the soil as a loam, confirming its suitability for plant growth and agricultural activities. Among all of the measured PTEs, only Cu and Zn exceeded the threshold values (TVs) established by the European Regulation (Table 1) [34].



The PTEs reported in Table 1 are very common soil pollutants as they can be found in some pesticides and fertilizers [35]. High concentrations of Cu and Zn can represent a serious risk for human health since both metals could become toxic [36], making soil treatment necessary. Results of the sequential extraction showed that the percentage of Cu and Zn in the bioavailable fraction (sum of the first three fractions) was 81% and 70%, respectively (Figure 5). These values further confirmed the risk of Cu and Zn migration from the soil to agricultural products.

### *3.2. Optimization of the Soil Flushing Process*

Figures 2 and 3 report the results of Cu, Fe, Mn, and Zn removal efficiency obtained during the flushing tests. Results from Figure 2 displayed that the EBCT influenced the flushing process only at high EDDS concentrations, resulting in a different removal efficiency. These differences were accentuated mainly for the lower values of bv (Figure 2). Actually, the slopes of the curves related to Cu and Zn removal were equal when the bv was higher than 2. On the contrary, no effects on the removal efficiency at different treatment times were found for tests characterized by significantly high EDDS concentrations (Figure 3). In this case, in fact, it was possible to achieve significant PTEs removal efficiency for all of the investigated treatment times. At a high EDDS concentration (3.6 mM), Cu and Zn extraction rates were characterized by a fast kinetic only in the first period of the treatment, whereas in the remaining time the kinetics were significantly slower. This observed effect was likely due to the following two phenomena: (i) an immediate extraction of the two metals from the carbonate fraction where they are in the di-valent form, i.e., Cu2+ and Zn2+; (ii) a subsequent oxidation of the metals from the mono- and zero- to the di-valent form, as a consequence of the humic acid reduction [33,37]. Actually, metals in zero-valent form cannot be complexed as Me-EDDS [33]. On the contrary, the extraction rates of Fe and Mn increased during the overall treatment time. In detail, a lower removal was observed at the beginning of the flushing treatment followed by a gradual increase at the end of the process. Such a result was attributed to the higher affinity of EDDS for Cu and Zn compared to Fe and Mn, which mainly affected the removal efficiencies at the beginning of the flushing treatment. Subsequently, the effect of the high concentration of EDDS prevailed on the previous factors, resulting in a relevant Fe and Mn removal efficiency. Moreover, the pattern was also related to the occurrence of different metals mobilization depending on the redox potential conditions of the soil. In fact, it is generally reported that Cu and Zn have a higher mobility rate under oxidizing conditions while Fe and Mn are characterized by higher mobility at lower values of soil oxidizing potential [38]. During the tests, after the first few days the soil in the column became water-saturated. This caused a reduction of the oxidizing potential and the conversion of Fe and Mn to their more leachable reduced forms [39].

During tests conducted with [EDDS] = 0.36 mM solution at different EBCTs, the extraction process was not influenced by the different values of bv. Results in Figure 2, in fact, show an overlap of the extraction curves.

This result was due to the occurrence of a total EDDS complexation with metals, as a consequence of the low concentration of EDDS in the solution. Hence, the EDDS concentration represented a limiting factor of the treatment velocity. As the EDDS was fully complexed, the metal exchange mechanisms acquired a higher importance in the process. Indeed, in all tests with lower EDDS concentration (i.e., [EDDS] = 0.36 mM), Cu was the only metal with an excellent initial extraction rate. Results showed that for bv values up to 4 the amount of leached Cu was 77 mg·kg−1, corresponding to approximately 49 ± 5 μmol (Figure 2). This value was higher than 70% of the corresponding EDDS moles injected (72 μmol). Such a result could be ascribable to the adsorption of a small amount of EDDS onto the soil and the formation of Cu-EDDS complexes with the remaining EDDS [40]. Then, Zn-EDDS complexes started to form only for bv values higher than 4, as a result of the different values of the stability constant (Kst) of the two PTEs-EDDS. In fact, the Kst of Cu-EDDS is higher than that of Zn-EDDS [41]. For fixed values of treatment time, the EBCT affected the extraction rate of metals since lower EBCT values corresponded to a higher amount of the injected EDDS solution, and resulted in higher metal extraction rates (Figure 3). As the EDDS was totally complexed by Cu and Zn, the cations Fe and Mn

### *Int. J. Environ. Res. Public Health* **2018**, *15*, 543

could not form complexes with the flushing agent, and therefore could not be extracted. The pH and the LOI index of the soil were evaluated at the end of the tests. The soil pH displayed a value equal to 7.05 ± 0.2, assessing for the investigated soil buffering capacity in minimizing pH changes [42]. Likewise, the LOI value was observed at the end of the test confirming no soil characteristic alteration due to the EDDS involvement.

**Figure 2.** Breakthrough curves at different Empty-Bed Contact Times (EBCTs) with respect to bed volume. [EDDS] = 3.6 mM: EBCT = 33 h; EBCT = 27 h; EBCT = 21 h; [EDDS] = 0.36 mM: EBCT = 33 h; EBCT = 27 h; EBCT = 21 h—(**a**) Cu; (**b**) Zn; (**c**) Mn; (**d**) Fe.

**Figure 3.** Breakthrough curves at different EBCTs with respect to treatment time. [EDDS] = 3.6 mM: EBCT = 33 h; EBCT = 27 h; EBCT = 21 h; [EDDS] = 0.36 mM: EBCT = 33 h; EBCT = 27 h; EBCT = 21 h—(**a**) Cu; (**b**) Zn; (**c**) Mn; (**d**) Fe.

### *3.3. Soil Washing Process and Comparison of the Removal Efficiencies*

The results obtained from the soil washing tests, with constant LSR and different EDDS concentrations, showed different removal trends for Cu and Zn (Figure 4). In all tests, a substantial increase in the removal efficiency occurred in the first 48 h, followed by a non-significant increase in the removal efficiency at 96 h. This was also confirmed by statistical analysis since no statistically significant differences between the removal efficiency at 48 and 96 h were observed (*p* > 0.05). These results were in accordance with previous investigations [33,43].

**Figure 4.** Cu and Zn removal by soil washing treatment at different times. 48 h; 96 h—(**a**) Cu; (**b**) Zn.

Results of tests conducted with varying the LSR showed significant differences in terms of the removal efficiency, especially for Zn, as also observed from the statistically significant differences (*p* < 0.05). It is worth noting that higher LSR values, achieved by increasing the liquid phase and keeping constant the EDDS solution molarity, induced a consequent increase of EDDS moles in the solution. Such a trend may be attributed to the following processes: (i) the presence of a certain amount of free EDDS non-complexed with metals; and (ii) the occurrence of metal exchange phenomena among the PTEs-EDDS complexes, which promotes the formation of Cu-EDDS or Zn-EDDS complexes [44].

In agreemen<sup>t</sup> with Tsang et al. [45], tests conducted with an EDDS concentration deficiency led to Fe and Mn removal efficiencies lower than 1%. At higher EDDS concentration, Fe and Mn removals slightly increased, but the overall removed amount of these elements was negligible compared to the initial total amount in the soil [46].

Comparing the results of the soil flushing and soil washing tests, it could be concluded that a significantly higher Cu removal efficiency was obtained with the soil flushing treatment considering the same amount of EDDS solution used. On the contrary, only a slightly higher Cu removal efficiency was achieved with the soil flushing treatment compared to the soil washing when the same treatment time was considered (Figure S1).

### *3.4. Fractionation of PTEs/Main Cations in Different Soil Layers after Leaching*

Interesting results were obtained from the sequential extraction procedure by comparing pre- and post-treatment metal distributions (Figure 5). Cu was initially bound to the organic substance and metal oxides/hydroxides complexes, while the amount of ions in the cation exchange sites was not relevant. A higher removal efficiency was observed for the Cu fraction bound to the organic substance and absorbed onto the metal oxides, as shown by the values of Cu extracted in the second and third steps. As regards Zn, instead, the highest removal efficiency was observed in the exchangeable and reducible fractions (steps 1 and 2). Small amounts of Cu and Zn were still found in the exchangeable and weak acid soluble fractions (step 1) after the treatment. This latter result was attributed to the EDDS amount adsorbed onto the soils after leaching [40,47].

Finally, Fe was extracted almost exclusively from the reducible fraction (step 2), while Mn was mobilized mainly from the reducible fraction (step 2) and partially deposited on the cation exchange sites along the soil column. The results showed (Figure 5) that the Cu and Zn concentrations at the end of treatment were lower than the threshold levels (Table 1) and their removal occurred mainly from the bioavailable fraction. This treatment displayed interesting outcomes since it allowed the reduction of the leaching of PTEs. In fact, the main removal occurred from the acid soluble fraction of the soil, which is generally characterized by fast metals mobilization [48] and represents a serious contamination risk for the environment [49].

**Figure 5.** Values of metals in the four steps of the sequential extraction on the contaminated soil (CS) and on the four layers of the treated soil before and after SF1 tests. Exchangeable and weak acid soluble fraction, Reducible fraction, Oxidizable fraction; Residual Fraction (**a**) Cu; (**b**) Zn; (**c**) Mn; (**d**) Fe.

### *3.5. Metals Transportation through the Columns*

Figure 6 reported the results of metal profiles at different column layers obtained from SF2. Data were plotted at different bv values corresponding to different amounts of injected flushing solution. Metal concentration in the soil was evaluated by mass balance through the measure of their concentration in the spent solutions sampled at the bottom of each layer.

The various metals exhibited different behaviors due to the occurrence of metal exchange phenomena or metal-EDDS complex adsorption. These phenomena were more significant for Zn than Cu.

The Cu concentration initially increased with the depth, therefore the concentration measured in the deepest layer was higher than that in the top layer. Nonetheless, after the injection of the first 0.25 bv of the flushing solution, the gradient tended to be gradually less pronounced, along with the reduction of the residual concentration in each layer. The same trend was observed for Zn only after the injection of the first 0.5 bv of the flushing solution. For this bv value, the distribution of Zn concentration with the depth was not monotonic at the beginning of the treatment, due to Zn release from the top layer and its adsorption in the following two layers. Upon increasing the bv values, the removal of Zn also occurred in the second layer but at a lower extent than in the top layer, and the released metal amount was adsorbed in the two successive layers. Finally, Zn removal also occurred in the third layer after 1 bv, and in the last layer after 2 bv injection. At the end, the Zn distribution was uniform, as was the Cu distribution.

The obtained distribution of Cu with the depth was in contrast with the findings of Hauser et al. [21]. This result was ascribable to the different operative conditions set for the flushing tests. In the work of Hauser et al. [21], leaching tests were carried out with a non-continuous flow rate and non-water saturated soil. In such conditions the reduction from Cu2+ to elemental Cu occurred and the complex with EDDS did not form [33].

On the other hand, the distribution of Zn with respect to the depth was similar to that obtained in previous studies [21,50]. This result was attributed to the non-occurrence of Zn2+ reduction to elemental Zn. In fact, it is reported [50] that Zn can be released as Zn2+ and Zn(OH)+ according to the following reactions (r1–r2):

r1) Zn2+ + HEDDS3− ↔ ZnEDDS2− + H+ r2) Zn(OH)+ + HEDDS3− ↔ ZnEDDS2− + H2O

and H+ formation can cause ZnO dissolution (r3–r3):

r3) ZnO + 2H+ ↔ Zn2+ + H2O r4)ZnO+H+ ↔ ZnOH+

Moreover, Zn oxy-hydroxides can be chelated from EDDS as follows (r5):

r5) ZnO + HEDDS3− ↔ ZnEDDS2− + OH<sup>−</sup>.

As regards Fe and Mn distribution throughout the column depth, it was confirmed that the removal took place starting from 2 bv. A uniform removal profile was observed along the depth during the first period of the treatment. After 10 bv a relevant removal of Fe and Mn occurred in the top layer in addition to the release of the metals from the successive layers. However, the amount of Fe and Mn removed was negligible compared to their initial concentration in the soil. This confirmed that pollutants can be removed without damaging the soil if an appropriate technique is selected.

**Figure 6.** Cu, Zn, Mn, and Fe transport across soil column layers with respect to different values of bv: 0 bv; 0.25 bv; 0.5 bv; 1 bv; 2 bv; 5 bv; 19 bv—(**a**) Cu; (**b**) Zn; (**c**) Mn; (**d**) Fe.

### *3.6. Release of PTEs in the Water after the Treatment*

Results from the SF3 tests are reported in Figure 7. As can be seen, there was no release of Cu, Zn, or Mn in the water once the injection of EDDS was stopped (i.e., after 4 bv of solution injection), confirming the effectiveness of the treatment. This indicates that the PTEs were likely bound in forms that exhibit a low potential release in the environment and low bioavailability for the living organisms [51]. The only exception was represented by the first 0.5 bv of removal (4–4.5 bv), characterized by the presence of a low metals concentration, due to the retention of a certain amount of EDDS in the column. This occurrence led to the recommendation of soil flushing with water as a final step of the remediation treatment. Conversely, the release of Fe persisted all along the treatment until the same removal efficiency was achieved, as observed in the control test.

**Figure 7.** Cu, Zn, Fe, and Zn cumulative removal from the soil as a function of bv, during the SF3 tests. Full symbols: 0–4 bv [EDDS] = 3.6 mM, 4–18 bv [EDDS] = 0 mM— Cu; Zn; Mn; Fe. Empty symbols: 0–18 bv [EDDS] = 0 mM. Cu; Zn; Mn; Fe.
