3.2.1. Adsorption and Retention Curves
Figure 1 shows the adsorption and desorption curves for Ni in the all ten soil samples (as an example), while for Cu, Zn, Pb and Cr are shown in
Figures S1–S4. In all of them, dotted curves show the experimental data while continuous lines show the Freundlich modelled curved. Although most of data fitted well both Freundlich and Langmuir models (except that from S6 and S7 samples), in general, data from adsorption and desorption experiments better fit Freundlich (
Table 2) than Langmuir model (
Table S1) as was also found by other authors [
21]. Most of the determination coefficients (R
2) derived from the adjustments to Freundlich models are higher than 0.90 or even 0.95 (
Table 2). In most of the 50 fitted curves n Freundlich parameter for adsorption was lower than 0.5 (41 cases) or around this value (2 cases), showing high heterogeneity in the adsorption sites. With the exception of S2 for Pb (n = 0.74) values significantly higher than 0.5 were found for S6 and S7 for Cr and Cu. For Pb adsorption data didn’t fit to Freundlich equation due extremely high adsorption (probably due to precipitation processes). Results from S6 and S7 samples for Cr and Cu show linearity of the adsorption processes or homogeneous sorption sites but also probably revealing irreversible reactions or precipitation for Cu, a probable process due to neutral pH values [
29,
30]. For most of the cases, the parameter n slightly varies from adsorption to desorption processes, or even remains the same (for Cr experiments) showing consistent pattern for n among adsorption and desorption curves and revealing heterogeneous sites for adsorption and retention after desorption (
Table 2).
The corresponding K
F values derived from Freundlich data fitting are also shown in
Table 2. The K
F values for adsorption of Cu (8407.2 and 2539.8 L
n mg
1−n kg
−1), Zn (1332.3 and 700.4 L
n mg
1−n kg
−1) and Ni (867.1 and 537.1 L
n mg
1−n kg
−1) in S7 and S6 soils, respectively, are the highest in agreement with the high adsorption. These results are consistent with those reported by Elbana et al. [
20] after 1 day adsorption experiments for soils with pH higher than 7. Pb adsorption does not fit Freundlich model due to the high adsorbed amounts (equilibrium dissolution concentration is close to zero values). This strong Pb adsorption is indicative of irreversibility process and probably precipitation [
20]. The affinities of S6 and S7 for Cr adsorption are the lowest of the studied soils as with high soil pH values the sorption of anions is usually low due to the low positive charge of the minerals among other processes that may occur simultaneously [
19].
After desorption experiments, KF values are also very high for S7 and S6 samples (Cu: 19,664.5 and 9244.2 Ln mg1−n kg−1; Zn: 2416.0 and 876.3 Ln mg1−n kg−1; and Ni: 1866.7 and 976.0 Ln mg1−n kg−1).
When comparing K
F values from the more acidic soils, values are lower than those just mentioned before and soil affinities are clearly depending on both metal on concern and treatment applied (adsorption and desorption). K
F values for Cu adsorption range from 204.6 (S3) to 491.4 L
n mg
1−n kg
−1 (S10), those for Zn increase from 60.8 (S3) to 148.3 L
n mg
1−n kg
−1 (S2); Ni affinities vary from 46.6 (S3) to 157.4 L
n mg
1−n kg
−1 (S2) and for Pb from 673.6 (S3) to 2100.9 L
n mg
1−n kg
−1 (S2). Elbana et al. [
20] reported compatible results for Cu, Zn, Ni and Pb adsorption in acid soils. Those from Zn and Ni adsorption shown in Antoniadis and Tsadilas [
14] are also compatible. Focussing on Cr adsorption, K
F from S7 is the lowest (3.3 L
n mg
1−n kg
−1) while the highest is in S1 (186.2 L
n mg
1−n kg
−1) and they are compatible with those obtained for different pH sorbents by Otero et al. [
31] and Choppala et al. [
32].
Affinities of soils for Pb, Cu, Ni and Cr increase after desorption experiments as KF desorption values are always higher than the corresponding KF adsorption values. During desorption process the easily reversible adsorbed metal is released while that remaining at soil surface is probably under high selective or specific adsorption reason why soils affinities increase. In addition, soil order affinities for metal retention suffer some changes after desorption process. The Cu retention affinities range from 455.3 (S3) to 800.5 Ln mg1−n kg−1 (S10), those of Ni increase from 95.2 (S3) to 268.2 Ln mg1−n kg−1 (S2), the KF for Pb retention varies from 1100.7 (S3) to 5082.4 Ln mg1−n kg−1 (S2) and, again, the lowest retention KF for Cr results is in S7 (38.3 Ln mg1−n kg−1) while the highest is in S1 (442.2 Ln mg1−n kg−1).
In the case of Zn, due to the comparatively higher amounts desorbed during desorption experiments the opposite happens. The higher released amounts of Zn during the desorption process indicate an important process to be investigated, especially in acidic soils [
33] as part of adsorbed Zn is susceptible to leaching. When comparing with the corresponding adsorption K
F values there is a decrease of the soil retention affinities except for S8. Therefore, S6, S7 and S8 are the only samples where K
F of Zn retention is higher to the corresponding K
F for Zn adsorption. In the rest of soil samples retention K
F range from 9.6 (S4) to 35.6 L
n mg
1−n kg
−1 (S10).
According to KF values derived from adsorption and desorption data, especially from the acidic soils, it is possible to deduce a probably selectivity or preference of the soils for each metal. In contrast, in the slightly neutral soils the preferential selectivity only comes out for anion or cation adsorption.
3.2.3. Adsorption Versus Retention Concentrations
The adsorbed and retained (still adsorbed in soil fraction after desorption experiments) amounts for each added concentration are represented in
Figure 2.
If soils do not release the previously adsorbed cation after desorption experiments, the retained concentration will be equal (or very similar) to the adsorbed one and this will be noticed in the graphs (
Figure 2) as that representation will overlap the thick grey line (retention: adsorption, 1:1). If retained concentrations are lower than previously adsorbed (i.e., the soil releases the previously adsorbed cations during desorption experiments) the corresponding representation in the graph will not overlap the grey strip showing up at the lower right of the graph. After representing all studied cases (10 soils and 5 metals) when comparing the concentration retained of each metal versus the adsorbed one several trends where identified. Generally, soils S6 and S7 (pH close to neutral) retained almost all the previously adsorbed amount of metal. The colour lines fitting the grey strip indicate that there is almost not released concentration after desorption experiments. The colour lines (green, yellow, grey, blue and red) indicate the ratio (retained/adsorbed) for each metal concentration (Cu, Zn, Pb, Ni and Cr, respectively) for soil S6 (normal lines) and soil S7 (dotted lines). Circles and squares show the highest adsorbed and retained concentration of each metal in S6 and S7, respectively, which in most of cases are comparable to the parameter β derived from Langmuir model data adjustment (
Table S1). This parameter describes the maximum adsorption or retention amounts when the fitting is good. S6 and S7 are in most of cases the soils with the highest β values for Cu, Ni, Zn and Pb adsorption and retention (
Table S1) and also those with the highest K
F values from Freundlich model (
Table 2) as it was already indicated. S6 adsorbs and retains a relative less amount of Ni and Zn than S7 and the concentrations of Cr adsorbed and retained in both soils are the lowest as β (
Table S1) and K
F (
Table 1) also indicate. In addition, when the highest concentration of Cr is added, part of the adsorbed concentration is released after desorption experiments in both soils (the red lines do not fit the grey strip,
Figure 2).
Focusing on the other and more acidic soils (S1–S5 and S8–S10), the trends identified are more related to the metal studied. In all cases, almost all the adsorbed Cu and Pb remain retained after desorption experiments as the corresponding lines fit on the grey strip (
Figure 2). All soils adsorb and retain higher amounts of Pb than Cu as β from Langmuir model also shows (
Table S1). Almost all Pb added remains retained in the soils (coloured squared overlap in the graph) but not the case for Cu and soil affinities are noticeable as the colored circles do not overlap. S3, followed by S4 and S5 adsorbed and retained the lowest Cu concentrations while no differences are identified among Cu concentrations fixed by S2 and S8–S10. In these cases, desorption remains low suggesting that is controlled by the previous adsorption process [
36].
Similar interpretations can be derived after diluted concentrations of Ni and Cr are added but the opposite happens after adding the highest concentrations as it shows the lower right part of the corresponding graph (
Figure 2). For Zn cases, although the adsorbed amounts are similar to those of Ni and Cr in the corresponding soils, after desorption experiments the concentration of Zn retained is much lower for most of the concentrations added suggesting nonspecific adsorption during adsorption experiments vs. specific retention of Zn after desorption ones. In addition, both β and K
F parameters derived from Langmuir and Freundlich models also indicate lower retained amounts and lower soil affinities for Zn after desorption experiments (
Table S1 and
Table 1, respectively). Due to in all cases they are high affinities at low concentrations (
Figure 2) the concentration level chosen for adsorption experiments will contribute to not overestimate the affinity coefficient [
20].
3.2.4. Pearson Correlation: Soil Properties vs. KF
In order to avoid precipitation influence (soil pH influence) and to better understand the role of the soil components and properties in metal adsorption and retention affinities, two different Pearson correlation analysis were performed between K
F values (obtained from both adsorption and desorption process) and soil components or properties. In the first one, all soil samples were included but in the second one the values corresponding to S6 and S7 samples were removed due to previously explained results and as according to Visual Minteq 3.1 program (software provided by Gustafsson [
37] calculations, precipitation mainly of Pb and Cu took place during adsorption and desorption experiments in these samples (pH of dissolution equilibrium was always higher than 6.5).
Table 3 shows the significant correlation coefficients between soil properties and K
F values from adsorption (ADS) and desorption data (DES). In both cases, when “all soils” or without S6 and S7 samples (No. S6&S7), Pearson correlation analysis showed no significant correlation with TC, TN and K
e, this is reason why they are not shown in
Table 3.
When all soils are included in the analysis, properties like pH (both measured in water and KCl) and eCEC together with contents of Ca
e, show high and significant correlation values with most of metal adsorption and desorption affinities (K
F). Although these results are in agreement with previous studies [
7,
20,
38,
39] they are mainly derived from S6 and S7 soil properties as they have the highest sorption capacities for most of metals and also very high pH and eCEC together with the high proportion of Ca
e. In fact, when these soils are not included in the analysis other soil properties show up depending on metal or treatment, except for Cr due to the similar order of magnitude of K
F values among soils.
In this context, adsorption of Cu is only correlated (positively) with organic matter content while retained amount does not correlate with any of the soil parameters evaluated. When S6 and S7 are not included, the pH is only correlated (negatively) with adsorption and retention of Cr. The exchangeable complex seems to play a specific role on Ni retention as it correlates with both eCEC and Mg
e and Al
e contents. These contents also influence Pb adsorption and retention soil affinities. According to Bradl [
3] and Tiller and Hodgson [
40], the adsorption of Zn to clay is characterized as dominantly reversible although part of the Zn retained is lattice entrapped in an irreversible nonexchangeable form, these results probably explain that only retention of Zn correlates with clay contents (
Table 3). In addition, Yang et al. [
33] argue that the higher desorption of Zn, compared to that of Cu or Cd, is probably due to Zn forms very weak complexes with organic matter that are easily released during desorption process.
Soil pH is a very important parameter determining the soil affinities and capacities for heavy metal retention. If soils are acidic or slightly acidic, heavy metal retention will be more influenced by soil components and other properties as organic matter and clay contents and eCEC.
3.2.5. Sorption Reversibility. Principal Component Analysis
Released concentrations during desorption experiments probably depend on the metal studied, especially in the more acidic soils as it was previously described. For understanding this process, the sorption irreversibility was studied by principal component analysis (PCA) in order to reduce the variables involved during the process. The ratio K
F retention/K
F adsorption (K
FDES/K
FADS,
Table 2) is commonly used as an indicator of the adsorption irreversibility or hysteresis. Zn is the more mobile metal in all cases as most of K
FDES/K
FADS indexes are lower than 0.4 (
Table 2) while the hysteresis indices for Cu, Ni, Pb and Cr vary depending on soils.
In order to synthetize the influence of soil properties and understand the underlying factors on specific adsorption (irreversibility), the hysteresis indices were included in the PCA together with soil properties. S6 and S7 results were not included in the principal component analysis as mainly precipitation and not real soil properties influence takes place. Principal component analysis was run in correlation mode and varimax rotation for maximizing the loadings of the variables on the components of the PCA [
41].
Five components explain 92.4% of the variance of the soil metal adsorption irreversibility (
Table 4). The first component (PC1-Pb) explains 32.72% of the variance and shows large positive loadings for hysteresis of Pb (His-Pb) and Mg
e, Al
e, DOC, eCEC as well as large negative loadings for soil pH. The high positive loadings of exchangeable cations, even those of Ca
e, indicate the importance of the cation exchange on soil Pb adsorption irreversibility. According to Vidal et al. [
1] retention of Pb is based mainly in specific sorption and Rosen and Chen [
42] indicated that exchangeable fraction of soils adsorbed Pb, specially at high metal loading rates. In agreement to eCEC values (
Table 1), S1 and S2 soil samples show the largest factor scores for the extracted PC1-Pb while the lowest are in S4 and S5 samples (
Table 5) showing the important role that cation exchange capacity plays in Pb adsorption and its irreversibility after desorption takes place.
The second component (PC2-Cu) explains 22.32% of the variance and shows large negative loadings for the hysteresis of Cu as well as large positive loadings for OM, K and Ca
e (
Table 4). Organic matter is known as a high affinity component for Cu adsorption even in acid soils [
1] but after desorption process, part of the Cu adsorbed in OM in exchangeable positions is released and S10, the soil with highest OM contents (
Table 1) shows the highest factor scores for PC2-Cu (
Table 5). The high affinities of Cu-organic matter are well known as it accumulates preferably in surface horizons [
43] but the soluble organic complexes can also influence the release of the Cu previously adsorbed. Therefore, Cu is adsorbed in the organic fraction of the soils but its solubilisation is also enhanced due to the mobilization of organometallic complexes [
42].
The third component (PC3-Zn) explains 17% of the variance and only shows large loadings for hysteresis of Zn and clay contents (
Table 4). S8 is the only acidic soil sample with retention K
F values higher than adsorption ones due to the high clay content (
Table 1) as the factor score of S8 for PC3-Zn shows (
Table 5). Veli and Alyüz [
44] already shown the efficiency of clay adsorption for Cu but specially for Zn removal from metal concentrated solutions. Kabata-Pendias and Pendias [
43] have also indicated that clays are capable of holding Zn quite strongly.
The fourth (PC4-Cr) and fifth (PC5-Ni) components explain 10.72% and 9.64% of the variance and show large loadings for hysteresis of Cr and Ni, respectively. The close to zero loadings for OM are in agreement with results from Bloomfield in Kabata-Pendias and Pendias [
43] as they indicate the bonding of Ni to the organic ligands could be not particularly strong. They are also related to the hysteresis of Pb and Cu as they show the moderately large loadings (
Table 4).
Soil adsorption irreversibility measures soil affinities for the previously adsorbed metal and can be used as an indicative of more specific adsorption as higher irreversibility values indicate less nonspecific adsorption and lower susceptibility for leaching. Although higher eCEC and clay and organic matters are responsible of higher specific adsorption, in the case of Cu, organic matter may also account for adsorption reversibility.
3.2.6. Competitive Adsorption and Desorption Experiments
Pair metal competitive experiments were conducted in S1, S6 and S7 samples.
Figure 3 shows the adsorption and desorption curves of Ni when one other metal is added during adsorption experiments as well that curve derived from single solution experiments (black lines). The corresponding competitive curves of Cr are shown in
Figure 4 and those of Cu, Zn and Pb are shown in the
Supplementary Materials (
Figures S5–S7, respectively).
The competitive adsorption and desorption results derived from S1, S6 and S7 samples show that compared to single experiments, competition resulted in decreased metal adsorption as for Echeverría et al. [
38] and Antoniadis and Tsadilas [
14]. Generally, individual adsorption and retention of Cu, Zn, Ni and Pb is always higher than competitive one except if Cr is in competition (see
Figure 3 and
Figure 4 as an example). In the latter case, competitive adsorbed or retained amounts of metal are even higher than in individual experiments, especially in S1 and S6 samples. Single adsorption and retention of Cr is lower than the obtained during competitive experiments in most of the cases (
Figure 4).
In order to synthetize these results, competitive adsorption and desorption data were also compared to Freundlich model. In addition, the Murali-Aylmore model was also applied, and the corresponding derived parameters from good fittings are shown in
Table 6.
In most of cases the competitive adsorption of Pb in samples S6 and S7 did not fit to the models applied mainly due to the low equilibrium dissolution concentration. This means that in the partitioning, most of added Pb is moved to the soil solid phase. For S1 samples, both Freundlich and Murali-Aylmore models [
45] indicates that Pb is preferably sorbed than any of the metals added in competition. K
F-
Pb(m2) is always higher than K
F-m2(Pb) and a
m2(Pb) is always higher than a
Pb(m2) (
Table 6).
Except if Pb is present as competing ion, soil affinities for Cu adsorption are higher than those of Ni, Zn or Cr in S1, S6 and S7 samples. If Cr is the competing ion, the affinities for Cu adsorption are even higher than when no competition takes place. The promoted adsorption of Cr when other competing ion is present (
Figure 4) is well described by the model as K
F of competitive adsorption of Cr are similar to those of individual sorption in S1 samples and n values from Freundlich model obtained in S6 and S7 samples are higher than 1.24. The parameter a
Cr(m2) show large negative values indicative of promoted adsorption. In fact, after desorption occurs, the specific adsorption sites play important roles during competitive experiments as the affinities of desorption K
F of competitive experiments (
Table S2) are even higher than those of individual experiments. In addition to the ones already identified during adsorption experiments, adsorption of other elements is also promoted if Zn, Ni and mainly Pb are present as competing ion in S1, S6 and S7 samples. The inhibition in adsorption of a metal in presence of other heavy metals is an usual finding [
46]. However, the synergistic effect between heavy metals in adsorption is not frequently reported. The competition/synergic adsorptions may be related with the cationic or anionic nature of the heavy metals. When two metals in competition are cations the competition seems clear [
47]. However, when a cationic metal is in solution with an anionic metal, synergistic effect may occur as suggest results of this study. Adsorption synergistic effects between cations and anions were previously found between Cu-P [
48,
49].