2.2 Soil Properties Affecting Trace Metals Sorption and Desorption
2.2.4 Assessment of the Role of Different Soil Components
At present, it is very difficult to experimentally identify the contribution of individual sorbents to the control of trace metal sorption/desorption soil particles. The partition coefficient Kp, which can be obtained from the slope of a sorption isotherm, has been widely used to characterize the equilibrium relationship of trace metals between solution and solid phases. The empirical relationships have been derived by relating Kp to key soil and solution parameters (e.g. total metal concentration, SOM content, pH, etc.) through multiple regression analysis (Janssen et al., 1997;
Impellitteri et al., 2002; McBride et al., 1997; Sauvé et al., 1997, 2000a, 2000b, 2003). For example, the distribution of trace metals between solid and solution phases was fitted based on pH, total metal content and organic matter (Sauvé et al., 2000b).
Distribution of the metals for forty-one soils with different properties collected from
around the world correlated within an order of magnitude with soil pH, 0.01 M HCl extraction equilibrium pH, percent organic matter, total recoverable metal, and 0.01 M HCl extract metal (Impellitteri et al., 2002).
Zachara et al. (1992) reviewed the experimental techniques used to assess the importance of different sorbents, including (i) application of sequential extractions, (ii) development of statistical relationships between the sorption and soil properties, (iii) observation of similarities in sorption on single sorbents and whole soils, and (iv) comparison of sorption on natural and treated soils. They concluded all these
techniques had their limitations. For example, the treated soils may expose new sorbing sites which are accessible in natural soils. Thus, besides these traditional techniques, it is necessary to develop new methods to quantify the contributions of different sorbents in soil particles to metal sorption and desorption. Recently, two approaches have been widely used for assessing the role of different sorbents for metal binding, including assemblage equilibrium models and spectroscopic techniques.
(1) Assemblage equilibrium models. Many efforts have been expended to develop the predictive equilibrium model for the reaction of metals with humic
substance/soils. Considering the heterogeneity of humic substances, Perdue and Lytle (1983) proposed a continuous ligand distribution model for proton and metal binding by humic substance in which the ligand concentrations are normally distributed with respect to their logK values. Recently, some equilibrium models have been
successfully developed and used to describe metal partition between soil particles and
competitive consistent adsorption (NICA) based model (Venema et al., 1996; Weng et al., 2001; Weng et al, 2002), and (iii) Stockholm humic model (SHM) (Gustafsson et al, 2003).
WHAM is capable of calculating the equilibrium chemical speciation in surface and ground waters, sediments, and soils (Tipping, 1994), especially when the chemical speciation is dominated by organic matter. WHAM is a combination of several submodels, including the Humic Ion-Binding Model V/VI for the proton and metal binding to humic substance, the surface complexation model for proton and metal binding to oxides, the electrostatic model for cation exchange on clays, and models of inorganic solution chemistry. Model V is a computer model which describes the ions reactions with humic substances, mainly through complexation which is modified by the electrostatic reactions. The nonspecific binding of ions is also considered although, in most of cases, it is minimal compared with the
complexation. The reaction sites in humic substances can be divided into two groups:
carboxyl groups and phenolic groups. The intrinsic pK for the carbonxyl groups is always less than 7 and the phenolic groups have higher pK. The metal-proton
replacement reaction is built into the model through metal exchange with the proton.
The bidentate sites can be formed through the monodentate sites. Tipping (1998) proposed an improved version, Model VI, in which the formation of tridentate sites were permitted. The tridentate sites, which are the strong sites with low density, can be very important at low metal loadings in the humic substances. In the oxide surface complexation model, four oxides are considered including iron oxide (FeOx),
manganese oxide (MnOx), aluminum oxide (AlOx), and silica (SiOx), which is similar to the surface complexation model proposed by Dzombak and Morel (1990). For the metal-clay reaction, a clay cation-exchanger is considered and only non-specific interaction is included.
The NICA based multi-surface model simulates metal binding to SOM, DOM, clay, and iron hydroxides using adsorption and cation exchange models, including NICA-Donnan, Donnan, diffuse double layer (DDL), charge distribution multi-site complexation (CD-MUSIC) models (Weng et al., 2001). The NICA- Donnan model was used to calculate metal binding to SOM which considered the proton and metal ions competitive binding to humic substances, binding sites
heterogeneity, ion specific nonideality, variable stoichiometry, cation exchange, and electrostatic effect (Weng et al., 2001). Two types of iron oxides, amorphous and crystalline, were considered using hydrous ferric oxide (HFO) and goethite. The metal binding to HFO was calculated by a two site surface complexation DDL model and the binding to goethite was described using CD-MUSIC. Illite was taken as the representative of clay mineral, and the reaction with metals was described using a electrostatic Donnan model.
Both Model V/VI and NICA-Donnan models consider the heterogeneity of humic substance but with different approaches. Briefly, the Model V/VI assumes a series of discrete binding sites with simple and conventional chemical reactions whereas the NICA-Donnan model uses a continuous distribution of binding sites.
Figure 2.1 showed how WHAM considered the metal reactions with different sorbents in soils and ligands in solutions.
Another model, SHM, which is a discrete-ligand model, is similar to Model V/VI in many respects considering metal and humic substance reactions although it involves a different electrostatic model (Gustafsson et al., 2003;
Gustafsson and Van Schaik, 2003). It has been calibrated and used in soil systems recently (Gustafsson et al., 2003). Besides some limitations of all these models, they provide very useful tools to predict the partition equilibrium of metals between soils/humic substances and solutions at different chemistry conditions.
Using this type of component additivity models, SOM has been reported as the major component among soil components accounting for metal binding (Weng et al., 2001; Gustafsson et al., 2003; Tipping et al., 2003). Weng et al. (2004)
calculated the Ni sorption on each soil components using the NICA based multi- surface model. At relatively lower pH (<pH 6.5) most of Ni was bound by SOM. At the pH close to neutral, iron (hydro)oxides are more important than SOM. The clay fractions are not important for the whole pH range. Ponizovsky et al. (2006) also demonstrated that the SOM is the major soil component responsible for the Cu partition between soils and solutions. Furthermore, it has been reported that only a portion of SOM, which is the so-called “active organic matter”, is responsible for metal binding (Tipping et al., 2003; Gustafsson et al., 2003).
Figure 2.1 Schematic picture of metal reactions in soils and solutions in WHAM. (from Lofts and Tipping, 2000).
Besides the success of these equilibrium models, it should be kept in mind these models still have their limitations. These models originally were calibrated with the generic humic substances, and the generic humic substances may be different from SOM which comes from different degradation pathways. WHAM does not consider the precipitation of trace metals at high pH, which restricts its being used to model the high pH soils, e.g. the calcareous soils. For the metal reactions with mineral phases, both model approaches are relative simple in this stage. For example, the interactions between organic matter and metal oxides and clay are not considered in the modeling and all sorbents independently bind metals. The formation of metal-LDH has been reported in soils (as reviewed previously) but none of these models have considered it.
These models should be further improved in order to be closer to the field conditions.
(2) Spectroscopic techniques. Another promising technique are spectroscopic techniques, which may be the only direct way to identify the mechanisms controlling metal sorption and desorption on soil particles. Some
techniques, such as X-ray absorption fine structure (XAFS), X-ray diffraction (XRD), electron microprobe (EM), and microfocused XAFS have been used to identify or quantify the metal species in soil particles (Manceau et al., 1996; Manceau et al., 2000; Roberts et al., 2002). Pb species in soil particles are highly dependent on the contamination history (Manceau et al., 1996). For example, in soil contaminated by alkyl-tetravalent lead compound, lead was found to be divalent and complexed to salicylate and catechol-type functional groups of humic substances. Lead sulfate and silica-bound lead are the predominant forms in the vicinity of battery reclamation area.
In a study using XAFS by Strawn and Sparks (2000), adsorbed lead in soils was found to bind with SOM. Another spectroscopic study has demonstrated the SOM is the major sorbent for Cu sorption from solution to soils (Flogeac et al., 2004), which was coated on the mineral phases in soil particles. Zn species were difficult to identify in soil particles because Zn may be present as one of several forms. Manceau et al.
(2000) found Zn was removed from solution by the formation of Zn-containing phyllosilicates and by adsorption to FeOx and MnOx. Zn speciation analysis in another two soils showed different Zn forms in soil particles (Roberts et al., 2002). In the strong acidic, black, and organic matter-rich topsoil containing 6200 mg/kg Zn, bulk XAFS revealed about 2/3 Zn was bound in franklinite and 1/3 bound in sphalerite, and only minor amount of Zn was adsorbed to Fe and Mn oxides. In the yellowish, loamy subsoil containing less Zn (890 mg/kg), Zn was mostly bound to Al- groups, and then to Fe and Mn oxides. Only minor amount of Zn was detected to be bound to organic matter. Cu has been recognized forming strong complex with organic matter.
These spectroscopic techniques provide direct information on the metal speciation in the soil particles. However, these techniques always require high metal concentrations in the soils which may be present in highly contaminated conditions.
The spectroscopic data analysis requires the linear least-square fitting of reference data in order to identify the metal speciation in the soils. Furthermore, when comparing the modeling results and spectroscopic measurement, it is important to pay attention to the
soils are always very high in order to obtain reliable spectroscopic results. The soils experienced different contamination history which may be very different from the simple soil and metal solution systems. For the equilibrium model predictions at this stage, they simulate the reactions between soils and metal solutions which may be only suitable for limited conditions in the field.
Overall, the importance of the SOM controlling metals partitioning between soils and solutions have been emphasized by both equilibrium modeling and spectroscopic studies. The importance of mineral phases may vary depending on the metals and reaction conditions. For example, for Cu, SOM is always the major
sorbent in soils, and for Zn, mineral phases may be very important in some conditions.