Adsorption / Desorption Patterns of Selenium for Acid and Alkaline Soils of Xerothermic Environments

: Selenium adsorption / desorption behavior was examined for eight Greek top soils with di ﬀ erent properties, aiming to describe the geochemistry of the elements in the selected soils in terms of bioavailability and contamination risk by leaching. Four soils were acid and four alkaline, and metal oxides content greatly di ﬀ ered between the two groups of soils. The concentrations of Se(IV) used for the performed adsorption batch experiments ranged from 1 to 50 mg / L, while the soil to solution ratio was 1 g / 0.03 L. Acid soils adsorbed signiﬁcantly higher amounts of the added Se(IV) than alkaline soils. Freundlich and Langmuir equations adequately described the adsorption of Se(IV) in the studied soils, and the parameters of both isotherms signiﬁcantly correlated with soil properties. In particular, both K F and q m values signiﬁcantly positively correlated with ammonium oxalate extractable Fe and with dithionite extractable Al and Mn, suggesting that amorphous Fe oxides and Al and Mn oxides greatly a ﬀ ect exogenous Se(IV) adsorption in the eight soils. These two parameters were also signiﬁcantly negatively correlated with soil electrical conductivity (EC) values, indicating that increased soluble salts concentration suppresses Se(IV) adsorption. No signiﬁcant relation between adsorbed Se(IV) and soil organic content was recorded. A weak salt (0.25 M KCl) was used at the same soil to solution ratio to extract the amount of the adsorbed Se(IV) that is easily exchangeable and thus highly available in the soil ecosystem. A much higher Se(IV) desorption from alkaline soils was observed, pointing to the stronger retention of added Se(IV) by the acid soils. This result implies that in acid soils surface complexes on metal oxides may have been formed restricting Se desorption.


Introduction
Selenium (Se) is an essential micronutrient for humans and animals, but can lead to toxicity when taken in excessive amounts. Plants are the main source of dietary Se, but the essentiality of Se for plants is still controversial, although the beneficial effects of low doses of Se on plants have been reported in several studies [1][2][3]. The concentration of Se in plants is directly related to the concentration and the bioavailability of the element in the soil and the plant species [4]. Selenium reactivity in soils depends not only on its total content but also on its chemical form [5,6]. The mobility and plant-availability of Se in soil is controlled by numerous chemical and biochemical processes, as follows: sorption, desorption, microbial activity, the formation of organic and inorganic complexes, precipitation, and dissolution and methylation to volatile compounds [6,7]. Depending on the oxidation state, Se is present in soil as selenide (Se 2 − ), elemental selenium (Se 0 ), selenite (SeO 3 2− ), selenate (SeO 2− ) and organic Se. The main factors controlling Se solubility and availability in soils are considered to be pH, oxidation-reduction potential (Eh), metallic oxy-hydroxides and clays, organic matter, microorganisms, and the presence Environments 2020, 7, 72 3 of 12 The desorbing solution of 0.25 M KCl was prepared by dissolving the proper amount of KCl salt in deionized water. This solution was also stored in airtight sterile glass containers.

Batch Experiments
For every soil six falcon tubes were used. One gram of soil was introduced to each falcon tube and 30 mL of the appropriate stock solution was added, resulting in a 1:30 w/v soil:solution ratio. Afterwards the falcon tubes were placed in an incubator with an adjusted steady temperature of 22 ± 1 • C and gently shaken at 120 rpm for 24 h on an end-to-end shaker. Then, the falcon tubes were centrifuged for 5 min at 3500 rpm and the supernatants were filtered through a Whatman paper No 42. Absorbed Se(IV) was calculated by the difference between the initial and the equilibrium solutions Se(IV) concentrations. Moreover, since pH plays an important role in Se behavior in the soil environment, the pH values of the initial Se(IV) solutions and of the equilibrium solutions were also recorded.
To desorb adsorbed Se(IV), 30 mL of 0.25 M KCl extractant solution was added in the falcon tubes containing the soil samples. Falcon tubes were placed again in an incubator with an adjusted steady temperature of 22 ± 1 • C, and gently shaken at 120 rpm for 24 h on an end-to-end shaker, centrifuged, and filtered through a Whatman paper No. 42, following the same procedure as described above. Desorbed Se(IV) was determined in the equilibrium solutions at the end of the process.

Isotherm Equations
Langmuir and Freundlich adsorption isotherms were produced based on the equilibrium adsorption data. However, the Langmuir model assumes that biosorption takes place at specific homogeneous sites on the adsorbent by monolayer coverage, while the Freundlich model is empirical and assumes sorption on a heterogeneous surface.
The linear form of the Langmuir model is [31] C e q e = 1 q m C e + 1 b L q m (1) where C e is the equilibrium concentration of ion in the solution (mg/L), q e is the amount of ion adsorbed per gram of adsorbent at equilibrium (mg/g), q m is the monolayer biosorption capacity (mg/g) and b L is the affinity constant related to the binding strength of adsorption (L/mg). The values of q m and b L can be determined from the linear plot of C e /q e versus C e . The linear form of the Freundlich model is [32] ln q e = ln K F + 1 n ln C e (2) where C e is the equilibrium concentration of ion in the solution (mg/L), q e is the amount of ion adsorbed per gram of adsorbent at equilibrium (mg/g), K F is a constant relating to the biosorption capacity (mg/g) (L/mg) 1/n and 1/n is an empirical parameter relating to the biosorption intensity. The values of K F and 1/n can be determined by plotting ln q e versus ln C e .

Distribution Coefficient (K d )
The distribution coefficient (K d ) (L/kg) was calculated according to the following formula: where C e is the equilibrium concentration of ion in the solution (mg/L) and q e is the amount of ion adsorbed per kg of adsorbent at equilibrium (mg/kg).

Analytical Determinations
Selenium, iron, manganese and aluminum concentrations were determined by using an atomic absorption spectrophotometry, Varian-spectraAA-300system. For the determination of Se at low concentrations, a Varian model VGA77 hydride generator was used. Available phosphorus concentrations were determined by a Shimadzu UV-1700 spectrophotometer. Every 10 samples a control sample was analyzed, and at the end of the measurements procedure 30% of the samples were reanalyzed to test reproducibility.

Soil Properties
The physicochemical properties of the studied soils and the total Se concentrations are summarized in Table 1. Most of the soils are characterized as medium to fine textured with low organic carbon content, as expected for Mediterranean agricultural soils, and with very low total Se concentrations, less than 0.28 mg kg −1 , pointing to Se deficiency [33]. Ammonium oxalate and dithionite extractable Fe, Al and Mn are expressed as % oxides content (g 100 g −1 soil) and presented as Feo, Alo, Mno and Fed, Ald and Mnd, respectively. Metal oxide concentrations greatly varied, ranging between 0.08 and 0.40% and 0.72 and 6.32% for Feo and Fed, between 0.55 and 1.03% and 0.06 and 0.26% for Alo and Ald and between 0.01 and 0.10% and 0.02 and 0.15% for Mno and Mnd. The pH range of both alkaline and acid soils was very narrow-7.4 to 7.8 for alkaline soils and 5.5 to 6.0 for acid soils. Electrical conductivity values in the alkaline soils were significantly higher than in the acid soils (p < 0.001, n = 4), but were not restrictive for the growth of crops. Most soils were marginally to moderately supplied with available phosphorus.

Selenium Adsorption
Acid soils showed a much higher retention of added Se(IV) than alkaline soils, in accordance with many studies [6,[34][35][36]. In particular, Se(IV) adsorption ranged between 8.52 and 234 mg kg −1 The distribution coefficient (Kd) is a measure of the occupation of available sorption sites in relation to the concentration of the added element. Depending on added Se(IV) concentrations, the Se(IV) Kd values were within the ranges 2.6-36.7 and 3.5-1091.5 L/kg for alkaline and acid soils, respectively. Over the whole range of added Se(IV) concentrations, the Kd values of acid soils were considerably higher than those of the alkaline soils ( Figure 2). The observed Kd values for the acid soils were noticeably higher than those reported by Soderlund et al. [36] for selenite adsorption on mineral soils (0.4-240 L/kg), while the highest Kd values are close to those determined by Sheppard et al. [37] for indigenous selenium (800-1500 L/kg). A decreasing trend of Kd values is commonly observed as the concentration of the element in solution increases, indicating that proportionally less of the added element is adsorbed by the soil colloids. Indeed, for all studied soils, Kd decreased as the Se(IV) solution concentration increased (Figure 2), and the higher to lower Kd ratio ranged between 4.6 and 9.1 for alkaline soils, whereas the corresponding range for acid soils was 10-90.2. The distribution coefficient (K d ) is a measure of the occupation of available sorption sites in relation to the concentration of the added element. Depending on added Se(IV) concentrations, the Se(IV) K d values were within the ranges 2.6-36.7 and 3.5-1091.5 L/kg for alkaline and acid soils, respectively. Over the whole range of added Se(IV) concentrations, the K d values of acid soils were considerably higher than those of the alkaline soils ( Figure 2). The observed K d values for the acid soils were noticeably higher than those reported by Soderlund et al. [36] for selenite adsorption on mineral soils (0.4-240 L/kg), while the highest K d values are close to those determined by Sheppard et al. [37] for indigenous selenium (800-1500 L/kg). A decreasing trend of K d values is commonly observed as the concentration of the element in solution increases, indicating that proportionally less of the added element is adsorbed by the soil colloids. Indeed, for all studied soils, K d decreased as the Se(IV) solution concentration increased (Figure 2), and the higher to lower K d ratio ranged between 4.6 and 9.1 for alkaline soils, whereas the corresponding range for acid soils was 10-90.2. The distribution coefficient (Kd) is a measure of the occupation of available sorption sites in relation to the concentration of the added element. Depending on added Se(IV) concentrations, the Se(IV) Kd values were within the ranges 2.6-36.7 and 3.5-1091.5 L/kg for alkaline and acid soils, respectively. Over the whole range of added Se(IV) concentrations, the Kd values of acid soils were considerably higher than those of the alkaline soils ( Figure 2). The observed Kd values for the acid soils were noticeably higher than those reported by Soderlund et al. [36] for selenite adsorption on mineral soils (0.4-240 L/kg), while the highest Kd values are close to those determined by Sheppard et al. [37] for indigenous selenium (800-1500 L/kg). A decreasing trend of Kd values is commonly observed as the concentration of the element in solution increases, indicating that proportionally less of the added element is adsorbed by the soil colloids. Indeed, for all studied soils, Kd decreased as the Se(IV) solution concentration increased (Figure 2), and the higher to lower Kd ratio ranged between 4.6 and 9.1 for alkaline soils, whereas the corresponding range for acid soils was 10-90.2.

Selenium Desorption
In the present study, 0.25 M KCl was used to extract adsorbed Se(IV). As is stated by Dhillon and Dhillon [34] and Zhu et al. [38], chloride ion can replace non-specifically adsorbed Se through ion exchange and mass action mechanisms. The desorption pattern was almost identical for all soils, i.e., increasing the initial Se(IV) solution concentration resulted in increasing the Se amounts desorbed from the soils (Figure 3). For all initial Se(IV) concentrations, less Se desorbed from acid soils, a trend more pronounced for initial solution concentrations up to 40 mg Se(IV)/L. Depending on the initial Se(IV) solution concentration, desorbed Se ranged between 2.6 and 117.6 and 0.2 and 84 mg kg −1 for alkaline and acid soils respectively (Figure 3).

Selenium Desorption
In the present study, 0.25 M KCl was used to extract adsorbed Se(IV). As is stated by Dhillon and Dhillon [34] and Zhu et al. [38], chloride ion can replace non-specifically adsorbed Se through ion exchange and mass action mechanisms. The desorption pattern was almost identical for all soils, i.e., increasing the initial Se(IV) solution concentration resulted in increasing the Se amounts desorbed from the soils (Figure 3). For all initial Se(IV) concentrations, less Se desorbed from acid soils, a trend more pronounced for initial solution concentrations up to 40 mg Se(IV)/L. Depending on the initial Se(IV) solution concentration, desorbed Se ranged between 2.6 and 117.6 and 0.2 and 84 mg kg −1 for alkaline and acid soils respectively (Figure 3).

Equilibrium Solutions pH
For all soils the acid initial solutions, pH led to acidic equilibrium solutions pH (Figure 4). In particular, the equilibrium solutions' pH values for alkaline soils showed a decrease between one and three units as the concentration of added Se(IV) increased, while for acid soils the corresponding decrease was sharp for a 10 mg/L initial Se(IV) concentration, remaining almost constant thereafter for higher Se(IV) concentrations. Alkaline soils 3 and 4 showed higher resistances to pH changes than alkaline soils 1 and 2, probably due to the higher buffering capacity attributed to the higher carbonates content (Table 1)

Equilibrium Solutions pH
For all soils the acid initial solutions, pH led to acidic equilibrium solutions pH ( Figure 4). In particular, the equilibrium solutions' pH values for alkaline soils showed a decrease between one and three units as the concentration of added Se(IV) increased, while for acid soils the corresponding decrease was sharp for a 10 mg/L initial Se(IV) concentration, remaining almost constant thereafter for higher Se(IV) concentrations. Alkaline soils 3 and 4 showed higher resistances to pH changes than alkaline soils 1 and 2, probably due to the higher buffering capacity attributed to the higher carbonates content (Table 1). Environments 2020, 7, x FOR PEER REVIEW 7 of 12

Selenium Adsorption
The experimental data fitted well with Freundlich and Langmuir isotherms, in agreement with Dhillon and Dhillon's results [35] ( Table 2). The calculated adsorption maxima (qm) from the Langmuir isotherm were higher for acid soils, as was in most cases the value of the bonding constant (bL), indicating the stronger Se(IV) retention by the acid soils.

Selenium Adsorption
The experimental data fitted well with Freundlich and Langmuir isotherms, in agreement with Dhillon and Dhillon's results [35] ( Table 2). The calculated adsorption maxima (q m ) from the Langmuir isotherm were higher for acid soils, as was in most cases the value of the bonding constant (b L ), indicating the stronger Se(IV) retention by the acid soils. Table 2. Parameters of the Langmuir and Freundlich models for Se(IV) sorption in the eight soils. Contact time 24 h, agitation rate 125 rpm, sorbent/solution ratio 1 g/0.03 L, Se(IV) concentrations at start time from 1 to 50 mg/L, temperature 22 • C.

Langmuir Constants Freundlich Constants
Soil q m (mg/g) b L (L/mg) R 2 p-Value K F (mg/g) (L/mg) 1 The parameters of both isotherms, i.e., K F and 1/n from the Freundlich isotherm, and q m and b L from the Langmuir isotherm, showed significant correlations with soil constituents. Both K F and q m significantly positively correlated with ammonium oxalate extractable Fe and with dithionite extractable Al and Mn (p < 0.01, Table 3), underpinning the crucial role of amorphous Fe, Al and Mn oxides in the exogenous Se(IV) behavior of the studied soils. The ability of Fe (especially amorphous), Al and Mn oxides to control Se geochemical behavior has been highlighted in many studies, supporting thus the leading significance of metal oxides in regulating Se mobility in soils [6,10,34,[39][40][41]. K F and q m were also significantly negatively correlated with EC (p < 0.05, Table 3) and negatively but not significantly with bonding constant (b L ). These relations suggest that an increased soluble salts concentration suppresses both Se(IV) adsorption and strength of Se(IV) retention in soils, and leads to the increased availability of freshly added Se(IV) in the soil environment. This finding is also reported in the review of Natacha et al. [10] and in references therein. Furthermore, the bonding constant (b L ) of the Langmuir isotherm significantly positively correlated with the Feo/Fed values of acid soils and with the eqCaCO 3 content of alkaline soils (Table 3), pointing to the fact that in acid soils the fresh Se(IV) retention strength increases when amorphous Fe oxides constitute a larger part of free the Fe oxides, whereas in alkaline soils carbonates may possibly affect Se(IV) sorption. No significant correlation between the organic matter content and the initial or the adsorbed Se(IV) content was observed, a conclusion commonly reached by many researchers. Coppin et al. [42] did not find a direct relation between adsorbed Se and organic material, and suggest that Se may be indirectly sorbed on organic particles by forming associations with surface Fe oxides and clays. Additionally, Soderlund et al. [36] reported the limited importance of organic matter on Se retention compared to Fe and Al phases, even when the latter are incorporated in organic substances. Though clay is considered to affect Se sorption in soils [6,43], no significant correlations emerged between the clay content of the soils and the parameters of the Langmuir and Freundlich isotherms, or the distribution coefficient.
In Table 3, the correlation coefficients for Feo, Ald and Mnd and mean K d (calculated from K d values for each initial added Se concentration) relations are presented. The significant correlations between K d values, ammonium oxalate extractable Fe and dithionite extractable Al and Mn (p < 0.05) further support that metal oxides govern Se(IV) sorption in the studied soils. The point of zero charge (PZC) of most Fe-oxides was shown to deviate slightly, ranging usually between pH 7 and 9, while the Environments 2020, 7, 72 8 of 12 pH pzc values for various Al oxides reported in the literature vary widely, with a median of 8.6. [44,45]. In the pH range of equilibrium solutions, the Fe and Al oxides are positively charged and can adsorb negatively charged Se species. At low pH values, Mn oxides may have offered additional positively charged sites, since the PZC for most Mn oxides usually occurs at pH < 5 [46,47], leading to the increased adsorption capacity of acid soils. Nakamaru et al. [48], by using 75 Se as a tracer, found that the K d values for selenite adsorption in Japanese soils were highly correlated with the active Al (Alo) and Fe(Feo) content of the soils. Premarantha et al. [49] reached the same conclusion for acid soils from rice-growing areas in Sri Lanka. However, Zhe Li et al. [50] did not observe any significant relation between K d and Alo and/or Feo concentrations in 18 soils from China, and report only a strong negative correlation between K d and soil pH values, indicating the stronger adsorption of selenite in acid soils. According to Table 3, the EC of soils was also significantly negatively correlated with mean K d values (p < 0.05). Interestingly, Se availability was not only regulated by the absolute poorly crystallized iron oxides, but also by the relative Feo content in the free iron oxides, as can be deduced from the significant correlation between mean K d and Feo/Fed values (p < 0.05, Table 3). Considering that the Feo/Fed ratio is used as an indicator for soil development, this result leads to the speculation that the stage of soil development can influence added Se(IV) behavior in the soil environment, and ultimately in the food chain. Nevertheless, the soils of the present study may have been formed from different parent materials, and such observations could be case specific, but may also be regarded as an indication for further research.

Selenium Desorption
Selenium desorption, presented as the percentage of the adsorbed Se(IV) concentration found in the equilibrium desorption solutions, increased as the added Se(IV) amounts increased ( Figure 5). Much lower Se% desorption from the acid than from the alkaline soils was observed, indicating a stronger retention of fresh Se(IV) by the acid soils. In fact, the mean Se% desorption (the average of Se% values for each initial added Se concentration) from the acid soils was significantly lower than the mean Se% desorption from alkaline soils (p < 0.01). Acid soils provided more active sites for the adsorption of negatively charged Se(IV) forms, since when lowering the pH positive charges on soil colloids increase, i.e., there is a higher protonation of surface hydroxyl groups, such as Fe-OH and Al-OH functional groups [36]. However, the stronger retention of Se(IV) by acid soils over the whole concentration range implies the involvement of different sorption mechanisms by the two groups of soils. It is probable that surface complexes may have been formed between Se(IV) species and oxides that lowered the reversibility of sorption process in acid soils. As is shown in Figure 4, for acid soils the pH of equilibrating solutions was very low, supporting the claim that stronger acidic conditions may have occurred close to the surfaces of active soil colloids that could lead to the formation of Se species preferably sorbed on such sites [6,10,40]. On the contrary, Se on the active surfaces of alkaline soils may have been retained mostly as easily exchangeable, thus leading to higher Se desorption by KCl. Numerous studies support the claim that low soil pH favors the higher sorption of Se (independently of Se speciation in equilibrating solutions) [8,[51][52][53] but much less has been done on the evaluation of freshly added Se(IV)'s desorption behavior in acid and alkaline soils. The dominant role of metal oxides in the sorption-desorption behavior of Se(IV) under the conditions of the performed batch Environments 2020, 7, 72 9 of 12 experiments is also supported by the significant negative correlations between mean Se% desorption values and oxides concentrations (Table 3).
may have occurred close to the surfaces of active soil colloids that could lead to the formation of Se species preferably sorbed on such sites [6,10,40]. On the contrary, Se on the active surfaces of alkaline soils may have been retained mostly as easily exchangeable, thus leading to higher Se desorption by KCl. Numerous studies support the claim that low soil pH favors the higher sorption of Se (independently of Se speciation in equilibrating solutions) [8,[51][52][53] but much less has been done on the evaluation of freshly added Se(IV)'s desorption behavior in acid and alkaline soils. The dominant role of metal oxides in the sorption-desorption behavior of Se(IV) under the conditions of the performed batch experiments is also supported by the significant negative correlations between mean Se% desorption values and oxides concentrations (Table 3).

Conclusions
Both the adsorption and desorption processes of freshly added Se(IV) in acid and alkaline soils revealed distinct differences between the two groups of soils. Acid soils adsorbed significantly higher amounts of added Se(IV) than alkaline soils, and alkaline soils desorbed more Se. Fe, Al and Mn oxides, and particularly amorphous Fe oxides content, were the key parameters controlling the

Conclusions
Both the adsorption and desorption processes of freshly added Se(IV) in acid and alkaline soils revealed distinct differences between the two groups of soils. Acid soils adsorbed significantly higher amounts of added Se(IV) than alkaline soils, and alkaline soils desorbed more Se. Fe, Al and Mn oxides, and particularly amorphous Fe oxides content, were the key parameters controlling the sorption/desorption of Se(IV) in the studied soils. Indeed, increased Feo concentration led to higher Se(IV) sorption and to lower Se desorption from the studied soils. Soil pH and the equilibrium solutions' pH strongly influenced both sorption and desorption patterns, providing more positively charged sites on oxides surfaces, leading to higher Se(IV) sorption. Furthermore, metal oxide chemistry at low pH values favored the formation of stronger surface complexes, thus suppressing the Se desorption from acid soils by a weak salt. Overall, the results of this study showed that metal oxides content and pH determine Se geochemistry in soils. Considering that biofortification through plant uptake is also crop/plant-dependent, Se(IV) application in agricultural soils should be site-specific, since a high Se leaching hazard in alkaline soils with low metal oxides concentration may emerge, and low Se availability in acid soils with high metal oxides contents can appear.