Soil Science Society of America Journal 65:597-612 (2001)
© 2001 Soil Science Society of America
REVIEW AND ANALYSIS
An Assessment of Environmental and Solution Parameter Impact on Trace-Metal Sorption by Soils
Robert D. Harter*a and
Ravendra Naidub
a Dep. of Natural Resources, Univ. of New Hampshire, Durham, NH 03824
b CSIRO Division of Land and Water, Private Mail Bag No. 2, Adelaide, South Australia 5064
* Corresponding author (rharter{at}bigfoot.com)
 |
ABSTRACT
|
|---|
When studying metal sorption by soils, the potential influence of environmental and solution parameters on the experimental systems cannot be ignored. Characteristics of the soil mineral surfaces are the final determinative factors in whether a metal ion will be sorbed, but soil-solution composition affects both mineral surface properties and whether the metal ions will be in forms that can react with the surfaces. Examples of factors affecting sorption of metals by soil surfaces include ionic strength, cations, anions, and/or organic ligands present in solution, solution pH, and solution metal concentration. In addition, sorption will be affected by external factors such as pressure, temperature, soil/solution ratio, and the manner in which soils to be studied are sampled and stored before investigation. To date, there has been little attempt to standardize experimental protocol, so results obtained using varied systems in different laboratories cannot be readily compared. An initial suggestion that all sorption studies include at least one treatment meeting minimal standards of ionic strength (0.01), background electrolyte (NaNO3), pH (between 5.5 and 6.0), and temperature (25 ± 3°C) is presented as a first step toward enabling improved ability to make interlaboratory comparisons.
Abbreviations: DOC, dissolved organic carbon • DOM, dissolved organic matter • I, ionic strength • PDI, potential-determining ion • PZNC, Point of Zero Net Charge
|
INTRODUCTION
|
|---|
THE QUANTITY OF METAL that can be sorbed by a soil or soil constituent is a complex function of surface properties and a variety of environmental and solution parameters. To elucidate sorption processes, soil chemists have paid close attention to surface properties. This approach cannot be severely criticized since, after all, surfaces are the point of attachment and sorption mechanisms cannot be postulated without considering these surface properties. In establishing experimental protocol, however, the potential influence of environmental and solution parameters cannot be ignored. On the other hand, considering certain environmental factors is sometimes impossible due to the limited facilities for controlling external influences. Internal influences, that is, the effect of solution variables, have been more readily recognized and studied but are subject to experimental protocol decisions so have been unevenly investigated. Our goal in this article is to point out some of the factors that affect metal sorption and to indicate the extent of current knowledge on the impact of these factors. Most of the discussion will be oriented toward Cd2+, since among trace metals of environmental concern this ion is one of the more weakly bonded by the soil. Weakly bounded ions are more susceptible to leaching, being readily displaced or desorbed by competing cations such as Ca2+, and they tend to be more bioavailable. As a result, these metals are of greater environmental concern than are the more tightly bound metals. The general principles discussed, however, apply to all trace metals.
|
Nature of Soil Mineral Surfaces: Permanent vs. Variable Charge
|
|---|
Interactions between trace metals and soil particles occur at the solid–solution interface. Thus surface chemical properties of soils, particularly the potential in the plane of adsorption and the surface-charge density, may play a dominant role in controlling the nature of interactions between the metals and soil surfaces. Such interactions may vary, depending on both soil type and the soil-solution composition. Soils from the temperate region are generally characterized predominantly by the presence of minerals with permanent charge but variable potential. Although the surface potential of permanently charged minerals must be constant, this potential is seldom or ever available to solution. Rather, reactions at the surface of these minerals depend on the zeta potential, which varies with solution conditions due to differential filling of the Stern layer. In contrast, soils from tropical regions or from regions with andesitic and allophanic parent material are characterized by minerals with variable charge, but constant potential surfaces (i.e., surface charge varies with solution conditions to maintain constant potential). The variable charge in soils usually arises on the edges of lattice clay minerals, as well as on the surfaces of sesquioxides and amorphous materials such as allophane, imogolite, and organic matter (McBride, 1994; Sparks, 1995). The reactive groups responsible for variable charge are similar in all the inorganic colloids and different in organic matter. In these soils the surface chemical properties change in response to the composition of the ambient soil solution. For example, the surface-charge density of oxidic minerals in an oxisol and an inceptisol from a similar geographic area has been shown to respond similarly to change in pH and phosphate (Fig. 1)
. Changes in ambient-solution pH results in marked variations in the surface positive and negative charge densities of both inorganic and organic colloids. For inorganic colloids, the pH at which the surface positive and negative charge densities are equal has been defined as the Point of Zero Net Charge (PZNC) (Sposito, 1984). At pH values above the PZNC, the soil particle surface is net negatively charged; below the PZNC, the surface is net positively charged. Metal-sorption studies have shown that ligand ions can alter the PZNC, thereby either enhancing or reducing the amount of trace metal sorbed. In both temperate and tropical soils the net charge density usually derives from both permanent and variable-charge components, varying primarily in the relative contribution of each component.
Metal sorption in soils has been related to both nonspecific and specific interactions involving both charged surface and neutral sites on mineral surfaces. Trace-metal interactions are largely nonspecific in nature with a small fraction adsorbing via chemisorption. Since the surface charge of variably charged minerals is created by the adsorption of potential-determining ions (H+ and OH-) onto the surface, the net charge is determined by that ion sorbed in excess. It therefore follows that adsorption of trace-metal ions by variable-charge soils can be influenced by both the nature and composition of the background electrolyte. Indeed, Naidu et al. (1994b) demonstrated that chemisorption of inorganic-ligand ions such as SO2-4 and PO3-4 modified the surface chemical properties of variable-charge soils, thereby enhancing Cd2+ adsorption, whereas we have found that SO4-2 sorbed on permanently charged soil resulted in little impact on Cd2+ adsorption. Clearly, a precise understanding of the trace-metal interaction in soils requires use of an experimental protocol that has minimal impact on the surface chemical properties of the soil particles.
|
Composition of Soil Solution
|
|---|
The soil solution has been defined by Adams (1974) as "the aqueous component of a soil at field-moisture content" and by the SSSA (1997) as "the aqueous liquid phase of the soil and its solutes." Regardless of the definition, as Adams (1974) states, it can be "conveniently viewed as a dilute solution of electrolytes at equilibrium with definable solid phase and gas phase components of the soil." It is the medium within which all soil reactions occur and therefore it is dynamic in nature. Information on soil-solution composition is critical for understanding the chemical processes that control soil weathering, bioavailability, mobility, and geochemical cycling of elements. Soil-solution composition varies significantly with both soil type and changes in the soil environment. The effect of soil type on soil-solution composition is illustrated in Table 1. The ionic strength of the strongly weathered soils such as tropical oxisols is usually expected to be <5 mmol L-1, although the Gillman and Bell (1978) data clearly demonstrates that oxisol ionic strengths can exceed 10 mmol L-1. On the other hand, less weathered entisols and inceptisols of temperate regions more commonly have ionic strengths of 10 mmol L-1 or greater. Although such generalities are made, ionic strength as well as cation and anion concentrations in any individual soil can vary substantially (Table 1), depending on both soil formation and current conditions. Certainly acid soils and alkaline or calcareous soils within any given soil group will have quite different ionic strengths. Based on the data for a range of soils (dominated by those from the temperate region), many laboratory studies have used electrolytes with ionic strength = 0.03 M as the model for soil-solution studies. In fact, the average ionic strength of all Table 1 data is only slightly lower at 0.0225 M (22.47 mmol L-1). This value is, however, heavily weighted by three arid region soils containing apparent excess salt buildup. If these three soils are excluded, the average ionic strength drops to 11 mmol L-1. Table 1 contains a rather limited data set, and a wider-scale empirical evaluation of the relationship between soil-solution properties and soil classification might be a worthwhile undertaking. Based on the data in Table 1, however, it would appear that an ionic strength of 0.01 M might be a better overall concentration for simulation of field soil-solution conditions until such a study can be completed. The common 0.03 M background ionic strength should be used only when simulating cultivated or saline-soil conditions where ionic strength is expected to be higher.
In addition to the inorganic phase, many soil solutions contain varying contents of dissolved organic carbon (DOC). For example, Dalva and Moore (1991) reported concentrations from 5 to 60 mg L-1 (mean 36.3) in organic soil horizons to 5 to 15 mg L-1 (mean 8.1) in mineral soil horizons, and Fotovat and Naidu (1997) found that the DOC concentration in alkaline sodic soils can exceed 200 mg L-1. The organic acids that contribute to DOC have a wide range of dissociation constants and therefore their interactions with trace-metal ions vary significantly as well. Both dissolved organic and inorganic ligands have been demonstrated to impact trace-metal chemistry of soils (Wallace and Lunt, 1956; Davis and Leckie, 1978; Chubin and Street, 1981; Elliot and Denneny, 1982; Naidu et al., 1994a; Fotovat and Naidu, 1997). The extent to which such ligand ions influence aqueous-phase chemistry of trace metals depends on both soil-solution composition and the nature of interactions with the metal ions. Where the metal–ligand interactions lead to thermodynamically stable complexes, adsorption reactions can be retarded in soils. Detailed impact of ligand ions on metal sorption in soils is discussed below.
The effect of soil-solution composition and ionic strength on metal sorption will be felt through their influence on the potential near a charged surface. As the ionic strength and ionic charge increase, the thickness of the diffuse double layer decreases. For variable-charge surfaces, the decrease in potential can be expressed by
 | (1) |
where 
o is the electric potential at the surface, k is the Boltzman constant, T is the absolute temperature, e is the electronic charge, 
is the valence of the potential-determining ion (PDI), c is the concentration of these ions in solution, and co is the concentration at the point of zero charge. The precise manner in which the double layer responds to changes in ambient-solution composition cannot be treated in detail here and readers are directed to soil chemistry textbooks such as McBride (1994) or Sparks (1995). Depending on the PZNC and the pH, increasing ionic strength may shift the potential in the plane of adsorption in either a negative or positive direction. The effect of ionic strength on the potential in the plane of adsorption and subsequent sorption of trace metals has been discussed by Barrow (1986) and Naidu et al. (1994a).
Since variable surface charge arises from the chemisorption of PDIs, H+, or OH- in most cases, the surface potential, o, is considered to be constant unless the solution concentration of PDI is changed. Equation [1] illustrates changes in the nature of electrolytes and electrolyte concentrations that cause modifications in the PDI, thus influencing the surface (or zeta) potential of soil particles. Therefore, at a constant pH, an increase in electrolyte concentration will increase the surface charge of variable-charge minerals to maintain a constant potential, or decrease the zeta potential of permanently charged surfaces. In practice, however, the pH of the reactor system in batch or column studies are rarely kept constant to the original field pH. As a result, introduction of solutions with composition different from the soil solution can lead to marked changes in the surface properties. Naidu et al. (1997) have demonstrated that changes in the ionic strength and composition of soil solution influence the surface chemical properties of soils and that such changes alter the soil's capacity to sorb metal ions. For instance, presence of Ca2+ reduced Cd2+ adsorption through its effect on soil surface potential, electric double-layer thickness, and competition with Cd2+ for sorption sites. Fotovat and Naidu (1997) confirmed that changes in the soil/solution ratio can influence the aqueous-phase chemistry of trace metals, Cu, and Zn in soils. They demonstrated that a decrease in the ionic strength of soil solution, as occurring with increase in the soil/solution ratio, can lead to marked changes in the ion-pair, free hydrated metal concentrations, and complexation reactions. They found that the slope of a line relating log Zn2+ to pH changed with changes in the soil/solution ratio as follows:
 | (2) |
 | (3) |
 | (4) |
Despite these known influences of ionic charge and ionic strength on trace-metal reactions, studies involving trace metals (as well as other sorption studies) continue to use background electrolytes containing a wide range of cations, anions, and ionic strength (Table 2) atypical of soil-solution composition. Later in this article we will show that both adsorption and desorption reactions are sensitive to changes in background electrolyte composition.
View this table:
[in this window]
[in a new window]
|
Table 2. Some examples of the background electrolyte composition and soil/solution ratio used in adsorption studies.
|
|
|
Solution Factors Affecting Sorption
|
|---|
At least five soil-solution parameters have been shown to influence the adsorption of metals by soils. These include ionic strength, competing ions, pH, counter ions, and loading rate. Failure to take into account these factors can result in considerable confusion relating results published in the literature and in implementing results at the field level.
Ionic Strength
Many studies have not made provision for background electrolyte composition that would be comparable to soil solutions. Soil solutions seldom have an ionic strength (I) approaching zero, which has been used in some sorption studies (Table 2). Whether results from such studies adequately model soil reactions will depend greatly on the soil itself. Some of our results indicate that variably charged soils and permanently charged soils respond quite differently to variation in I. Increasing I decreased Cd2+ sorption in both a permanently charged and variably charged soil, but the effect of I(NaNO3) on Cd2+ sorption was particularly dramatic in a variably charged oxisol (Fig. 2)
. In the oxisol, the Cd2+ sorption isotherm changed from an "L" type to an "H" type as I was decreased from 0.015 to 0.001. It is apparent that Na competes effectively with Cd2+ for nonspecific sorption sites when the Na/Cd ratio approximates 3000 (I = 0.03 M), but less effectively at a Na/Cd ratio of 100 (I = 0.001 M). Sorption of Cd2+ continues to be observed at high Na/Cd ratios, however, indicating that a certain fraction of Cd is probably sorbed onto specific sites. Thus metal sorption will be dependent on I and composition of the background electrolyte and the nature of reactions involved; specific sorption of metals by oxides appears to occur via inner-sphere complexation. Since Ca2+ can frequently compete for these inner-sphere sites, whereas Na+ cannot, the extent to which I affects metal sorption by soils dominated by oxide materials will depend on whether the soil solution is dominated by Ca2+ or Na+ ions. Studying Ni2+ sorption by pyrophyllite, Scheidegger et al. (1996) concluded that sorption can be divided into two pH regions: At pH < 7.0, Ni2+ sorption was inversely correlated with I (0.01–0.1 M), whereas at pH > 7.0, Ni2+ sorption became slow and unaffected by I. At high pH, soils generally have very high affinity for metals and therefore the number of sites available for adsorption reactions is high, while the reverse is true at acidic pH values. In other words, with increasing pH, there is a marked increase in surface charge and consequently the number of sites available for sorption. Thus they felt the effect of ionic strength on sorption may simply be related to the number of sites available for binding.
In a similar study of two oxisols, a xeralf, a xerert, an andept, and a fragiaqualf, Naidu et al. (1994a) found that the effect of I and pH on Cd2+ sorption depended only on the PZNC. In the six soils, which contained varying ratios of permanent to variable charge, increasing I decreased the amount of Cd sorbed at pH values above the PZNC, whereas at pH values below the PZNC, increasing I increased Cd sorption. They concluded that sorption was primarily a function of surface charge, but that adsorption below the PZNC, when the net surface charge was positive, indicated specific sorption.
Electrolyte
The effect of index cations on metal sorption can be manifested through both direct competition for adsorption sites and through its effect on diffuse double-layer chemistry. Bowden et al. (1973) reported that the potential in the plane of adsorption is related to the valence of the ion through its effect on surface-charge density. Increasing the valence of the cation makes the potential in the plane of adsorption less negative at pH values greater than PZNC, thereby reducing cation adsorption. The two most commonly used index cations for adsorption studies include Ca2+ (Harter, 1992) and Na+ (Naidu et al., 1994b) ions, and sorption studies show that adsorption of metal cations is decreased in the presence of Ca ion compared with Na (Fig. 3)
. However, this trend is reversed for trace-metal anions such as chromates (Fig. 4)
and arsenates. Such observations have implications to both the bioavailability and mobility of trace metals. Thus the effects of index cations can be substantial and failure to take these effects into account can lead to conclusions that have little relevance to real situations. The effect of index cations on metal sorption also varies with changes in the electrolyte pH. For instance, since more Cr was sorbed in the presence of Ca2+ (Fig. 4), there will usually be a greater sorption decrease with increasing pH in the presence of Na+ compared with Ca2+. This is related to Na+ only competing for outer-sphere sorption sites, whereas H+ ions will compete at the inner sphere. In other words, these differences between the effect of Ca2+ and Na+ on adsorption of Cr are due to the changes in the potential in the plane of adsorption through specific adsorption of the index cations. Lagerwerff and Brower (1972) reported that the sorption of Cd2+ depends on the nature of cation species in the background electrolyte and increased in the order Al < Ca < K < Na. Boekhold et al. (1993) reported a similar trend for adsorption of Cd2+ in Na+ and Ca2+ solutions, and showed the effect to be strictly ionic competition for bonding sites. A similar effect of Na+ and Ca2+ on Cd2+ sorption was reported by Temminghoff et al. (1995). More recently, Kookana and Naidu (1998) investigated the effect of soil-solution composition on Cd2+ transport through variable-charge soils. They used a multicomponent isotherm similar to that of Temminghoff et al. (1995) to discern the effect of ionic strength, pH, and Ca. They found that the amount of Cd2+ sorbed was a linear function of [Cd2+]x[H+]y[Ca2+]z.
Counter Ion
As already noted (Fig. 1), inorganic ligands also modify the metal–soil reaction. For example, in the presence of phosphate, metal-sorption capacity of oxidic soils is much higher. Such observations are important because they may have implications for reducing the bioavailability of trace metals through application of oxides and P in combination. On the other hand, depending on the complexation constant, inorganic-ligand complexes can be formed in solution (Fig. 5)
and therefore compete with the soil surface for the metal ion. Clearly, unless the objective is to specifically evaluate effects of solution complexation on sorption of Cd by soil or sorption of Cd by saline soils, it would be unwise to conduct the investigation in a chloride or sulfate system. Formation of solution complexes will usually decrease sorption, but in certain cases sorption may be enhanced if the metal–ligand is subsequently sorbed by the surface. In either case, the possible formation of complexes must be considered in developing a research protocol.
View larger version (30K):
[in this window]
[in a new window]
|
Fig. 5. Cadmium, zinc, and copper species distribution as a function of background NaNO3, Na2SO4, or NaCl ionic strength, calculated using MINTEQA2 at pH 7.0 and 10 mmol L-1 M2+, and effect of pH on 10 mmol L-1 Cd2+, Cu2+, and Zn2+ in 0.03 M NaCl.
|
|
Each metal–ligand combination will have its own characteristic stable complexes in solution that must be evaluated when developing an experimental protocol. For example, Yin et al. (1996) noted that Hg(II) begins to complex with Cl- when the latter is present in concentrations greater than about 10-8 M, but addition of up to 10-2 M Cl- only decreased Hg(II) sorption in acid conditions, and then only in low organic matter soils. Apparently dissolved organic matter has a stronger affinity for Hg(II) than does Cl-, so in the presence of dissolved organic matter, Cl- in solution only minimally affects Hg(II) sorption. The effect of pH and Cl- on the adsorption of Hg(II) by finely divided silica was studied by McNaughton and James (1974), who found that in the absence of Cl-, adsorption began at about pH 2, reached a maximum at about pH 5, and then decreased. In the presence of Cl-, however, there was little adsorption at low pH, maximum adsorption was at pH values >5, and adsorption varied with the ionic strength of the electrolyte solution. Similar observations were reported by Forbes et al. (1974) and Kinniburgh and Jackson (1978). In 0.001 M Cl-, maximum adsorption was near pH 7, and in 0.1 M Cl- solution, Hg adsorption was apparently still increasing at pH 10. On the other hand, Zhang and Sparks (1996) found minimal differences between Cu sorption by a montmorillonite in the presence of chloride, perchlorate, nitrate, or sulfate ions, and we have found only minor differences between sorption of Zn by two permanently charged soils in the presence of chloride, nitrate, sulfate, and acetate. Boekhold et al. (1993) noted that metal sorption in the presence of different anions will be a function of free-metal activity, rather than total metal concentration in solution. As expected from the cited observations, neither Cu nor Zn tend to form complexes with inorganic anions other than OH- (Fig. 5).
The effect of counter ions on adsorption of metal ions also appears to vary with the nature of soil type. For Cd2+, while the presence of sulfate increased adsorption in oxisols, there is little effect on its sorption in xeralfs (Naidu et al., 1994a). In contrast to SO2-4, Cd2+ sorption was enhanced in the presence of PO3-4 in both oxisols and xeralf. These results indicate that the nature of sorption can depend on soil types, as well as composition of the background electrolyte. Barrow (1987) similarly found that PO3-4 pretreatment could increase Zn2+ sorption by soil; he also found that the effect was pH-dependent and concluded that the effect could depend on PO3-4-induced change in both surface charge and pH. Thus generalization of data from work based on one soil to other soil types may not be feasible, given the complex nature of interactions between the counter ions and soil surface, as well as counter ions and metals. Thus regulatory guidelines developed in any single country or locality may not have direct application to areas of different climatic conditions and soil types. For example, regulations developed for the northeastern USA may not be correct for the arid Southwest, since both climate and soil characteristics are drastically different in the two regions.
The two most commonly used background electrolyte anions are Cl- and NO-3. While these anions are commonly present in soil solution, the concentration of the electrolyte solution is often many times greater than that of the anions in soil solution. As already discussed, such high concentrations can modify sorption reactions either through the formation of metal–ligand complexes or modification of soil surface charge density through the formation of covalent bonds.
pH
We also know that pH influences metal retention by soil. For example, Boekhold et al. (1993) reported a doubling of Cd sorption for each 0.5 increase in pH between pH 3.8 and 4.9. In addition, all metals exhibit a sorption edge whereby sorption increases rapidly over a relatively narrow pH range (Fig. 6)
. As indicated, the sorption edge tends to become more distinct and shifts to a higher pH as the amount of metal added increases. An attempt is usually made to keep solution concentration low enough that the rapid increase cannot be readily explained by precipitation of the metal. The rapid increase in sorption is usually assumed to be the result of both variations in the surface charge and shifts in solution ionic species. In the case of Fig. 6, at pH 8.0, both ZnOH+ and Zn(OH)2(aq) will be present in the solution, so the sorption edge could be partially due to a shift in species sorbed from Zn2+ to ZnOH+. As pH increases, there will be increased tendency to shift sorbed species, but there is also an increasing probability that the solid-phase Zn(OH)2 will be forming at the surface. The coinciding of the Zn sorption edge at about pH 8.0 (Fig. 6) with the rapid drop in solution Zn+ concentration (Fig. 5) and concomitant increase in solution Zn(OH)2(aq) would argue for this mechanism. Regardless of the actual mechanism, the presence of such pH relationships makes it difficult to compare soils with inherent pH differences, particularly those also exhibiting widely different sorption capacities. Many sorption studies have simply been conducted at soil equilibration pH. In some studies, the pH has been adjusted, but whether adjustment is by titration at the time of experiment initiation or by prior equilibration with liming material does affect results. Prior equilibration usually provides a more stable pH over the experimental period.
Care must be taken, however, that we do not don "blinders" and consider only the role of pH on metal solubility, since other important factors can also be affected by pH. For example, Suave et al. (1998) have shown a complex interaction between pH, DOC, and Pb in solution. They reported the traditional increase in Pb solubility with decreasing pH below pH 6.5. Above pH 6.5, an increasing amount of DOC was in the soil solution, along with Pb complexed with this material. The result was increased Pb solubility as the pH increased above 6.5 and a concern that at near neutrality and above, the Pb in solution will be more mobile. McBride and Blasiak (1979) demonstrated that Zn and Cu react similarly in soil. They found metal ion activity to continue decreasing, but total ion in solution to increase with pH increase above neutrality, the latter being a result of complexation with DOC at high pH. They also reported that an increasing amount of Zn and Cu passed through a cation-exchange column as pH increased above 
6.0. Even without considering DOC, Harter (1992) argued that despite the decreasing solubility of metal ions with increasing pH, competition with Ca2+ for sorption sites could actually make the ions more mobile in soil solution at high pH. One can therefore only conclude that despite what we do know about the response of trace metals to changing soil-solution pH, there still remains some contradictory indications that need to be resolved.
Metal Loading Rate
Although a wide range in metal concentrations (0.01 to >500 mmol g-1) have been used for sorption studies, relatively few studies typify metal loadings commonly found in the soil environment. High metal loadings often lead to results that are only evident at concentrations much higher than normally encountered in nature. For example, Elrashidi and O'Connor (1982) reported that at low loading rates (0.015 to 1.5 mmol g-1), neither counter ion nor ionic strength affected Zn sorption by nine soils. The effect of loading rate is also illustrated in Fig. 6 for Zn sorption by an alfisol. At a moderate metal loading (45 mmol g-1), the effect of pH on adsorption is decreased and there is only a hint of a sorption edge between about pH 6.3 and 7.0. When loading is decreased further toward the level that just satisfies specific sorption sites, one would expect minimal pH effect on sorption, as observed (Fig. 7)
for Cd adsorption by a smectite-dominated xeralf (Naidu et al., 1994b). They showed that at 0.006 mmol L-1 Cd concentrations, comparable to "real world" situations, pH had no effect on Cd adsorption—essentially all the added Cd was adsorbed. When the concentration of added Cd was increased to 0.03 mmol L-1, however, pH exerted a pronounced effect on sorption although there was still only a rudimentary sorption edge at about pH 4.0. The lack of pH effect on Cd adsorption in high-affinity soils may also explain the lack of pH effect on plant uptake of Cd from certain soils at low loading rates (Oliver et al., 1998). Clearly, at least part of the variability in reported metal-sorption reactions is related to loading rates. Thus more work is needed to assess whether we should continue working at unrealistically high metal loadings or modify our approach to better reflect indigenous levels of metal ions found in soil solution.
View larger version (13K):
[in this window]
[in a new window]
|
Fig. 7. Effect of pH and Cd loading on the proportion of Cd sorbed by a xeralf (after Naidu et al., 1994).
|
|
Organics
Finally, soil chemists have tended toward studying the simplest system possible, assuming it is a model for the next step of complexity; in fact, it is probably no more than a model of the system being studied. We ignored organic matter for a while, and when we began to study sorption by whole soils we ignored probable solution constituents. One of the constituents ignored has been low molecular weight organics; we have often argued that, after all, they would be quickly decomposed by microorganisms. Harter and Naidu (1995) noted that small but significant amounts of low molecular weight organics are now being found in a variety of soils. Furthermore, Lo et al. (1992) found that organic ligands forming complexes in soil solution have a greater influence on sorption than does dissolved organic matter (DOM) from the soil. As demonstrated by Naidu and Harter (1998), these organics can be a very important factor in metal sorption by the soil. Recent studies in our laboratory showed marked changes in the nature of Cd within the rhizosphere soil. Further studies are in progress in the authors' laboratory, investigating the effect of low molecular weight organics on the nature of Cd species in soil solution.
Sakurai and Huang (1995) studied Cd adsorption using montmorillonite and hydroxyaluminium montmorillonite in the presence of varying concentrations of oxalate in the background electrolyte. As expected, these mineral surfaces had extremely high affinity for Cd in the absence of oxalate. They found that the critical concentration of oxalate to inhibit the Cd sorption was between 0.5 and 5 mM and concluded that adsorption was greatly inhibited because of the formation of Cd-oxalate complexes. We have found (Fig. 8)
that organics such as acetate have minimal (oxisol) to no (alfisol) effect on sorption, regardless of concentration. In contrast, strong chelating agents such as oxalates or EDTA can substantially affect sorption, depending on the nature of soils. Oxalate did not affect Cd sorption in the permanently charged alfisol, but dramatically increased Cd sorption in the variably charged oxisol. EDTA very slightly increased Cd sorption by the variably charged oxisol, but in the permanently charged alfisol, no Cd was sorbed until all EDTA bond sites were occupied. Similarly, the isomers maleate and fumerate increased Cd sorption in the variably charged soil and decreased Cd sorption in the permanently charged soil (data not shown). Citrate decreased sorption in both soils, with the decrease greatest in the permanently charged alfisol.
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 8. Effect of low molecular weight organics on Cd sorption by (a) a variably charged oxisol and (b) a permanently charged alfisol, in the presence of NaNO3 at ionic strength of 0.03. Concentration of acetate, oxalate, and citrate was 0.5 mmol L-1, EDTA was 0.01 mmol L-1.
|
|
Typical of all soil chemical reactions, there is a complex relationship between the various solution parameters and sorption of metals by the soil. For example, Yin et al. (1996) showed that Hg(II) sorption decreases with increased pH, but that dissolution of soil organic matter increases with increasing pH. When they removed soil organic matter with hydrogen peroxide Hg(II), sorption by soils was either constant or increased with increasing pH. Similarly, Kalbitz and Wennrich (1998) reported that mobility of Cr, Hg, Cu, and As through soil is positively correlated with DOM. They found, however, that mobility of Cd and Zn was related to pH, not DOM, and that DOM is of minor importance in mobilization of all studied trace metals at soil pH below 4.5.
|
Environmental Factors Affecting Sorption
|
|---|
Pressure and Temperature
Both temperature and pressure affect sorption processes. Applying thermodynamic principles, Aharoni and Ungarish (1977) showed that sorption is increased with increase in both temperature and pressure. Furthermore, they indicated that either an increase in temperature or a decrease in pressure will cause a desorption event, followed by resorption. This latter phenomenon has been successfully used to study very rapid reactions (Zhang and Sparks, 1989, 1990), including cation exchange (Tang and Sparks, 1993). Most agriculturally and environmentally oriented research, however, is conducted under ambient temperature and pressure, whereas comparison of research results requires standardization of both. While temperature and pressure cannot be altered in field work, they can be controlled in the laboratory. Most research of this type does not justify expenditure on elaborate pressure-control mechanisms, since pressure in the laboratory is more or less the same as that in the field. Fluctuations are relatively minor in any given location, and while pressure could be a factor in comparing results from a coastal location with that of a mountain location, other factors will probably have a larger impact on results. Temperature, however, is a different story. Diurnally and annually it can fluctuate over a wide range in both field and laboratory and it can vary significantly between summer and winter. Furthermore, a small change in temperature can significantly influence the sorption process. As noted by Barrow (1986), increasing temperature tends to both change the reaction rate and the equilibrium point. Indeed, in studying the sorption of Ni2+, Zn2+, and Cd2+ by goethite, Brümmer et al. (1988) demonstrated a shift of the sorption edge to a lower pH as temperature increased from 5 to 35°C, the shift being greater than that observed when reaction time was increased from 7 to 21 d. In evaluating Ni2+ sorption by various surfaces, Mehadi (1993) found that increased temperature increased both the reaction rate and the total amount of Ni2+ sorbed by the surface (Fig. 9)
. Furthermore, while at the higher temperature the initial reaction was more rapid, it appeared that some type of secondary reaction caused an extended drift toward equilibrium. Practically, we know that unless a laboratory is climate controlled, temperature can vary significantly between summer and winter. As demonstrated in Fig. 9, a 10-degree temperature difference can substantially alter (i) the sorption kinetics, (ii) solution metal content at equilibrium, and (iii) the time required for equilibrium to be achieved. Such temperature variation may have contributed to variability in experimental results.
Soil/Solution Ratio
Widely different soil/solution ratios have been used to investigate the adsorption and desorption behavior of trace-metal ions by soils (Table 2). Despite differences in soil/solution ratio, there has often been a tendency to compare data between both laboratories and techniques. Variations in soil/solution ratio may, however, influence the aqueous-phase chemistry of indigenous trace-metal ions, and thereby affect either or both the sorption and desorption process. The sorption reaction primarily depends on three factors: the number of available sites to accept sorbate (sorption capacity), the total amount of sorbate in solution, and the probability of contact between sorption site and the sorbate. The sorbate will also need to displace ions already occupying sorption sites, and different ions will have varying resistance to displacement; this will affect the number of available sites. For a given amount of sorbate, when the soil sorption capacity is high in relation to the amount of sorbate present, most or all sorbate is removed from solution by a small amount of soil; adding more soil to the suspension (i.e., decreasing the soil/solution ratio) cannot increase the amount of sorption observed because the number sorption sites available far exceeds the number of sorbate ions or molecules present. Calculated as sorption per unit weight, sorption will, in fact, usually decrease. If the sorption capacity is low, however, sorbate present is much greater than sorption sites, so removal of sorbate from solution directly depends on the total amount of soil present and will increase as the soil/solution ratio is decreased. Sorption per unit weight will also increase slightly with decrease in the soil/solution ratio because the probability of interaction between sorbate and sorbent increases as the availability of sites increases. This principle is demonstrated in Fig. 10
. When a constant amount of Cd2+ was added to a high sorption capacity xeralf, greatest Cd2+ sorption (per unit weight) occurred from the 1:40 soil solution (Fig. 10a), but the low sorption capacity oxisol sorbed slightly more Cd2+ from the 1:5 soil solution (Fig. 10b). Desorption is similarly affected. For example, Fotovat et al. (1997) demonstrated that changes in soil/solution ratio markedly affect the desorption characteristics of trace metals Cu and Zn. We have also found that changing the soil/solution ratio of As-contaminated soils leads to widely different amounts of As being released to solution. Decreasing the soil/solution ratio leads to increasing cumulative release in As in soil extract (Fig. 11)
. A similar effect of the soil/solution ratio was recorded in desorption of Cr from contaminated soils. The soil/solution ratio affects both the rate and extent of sorption, as demonstrated by Tan and Teo (1987). Evaluating the effect of the soil/solution ratio on equilibrium solid-phase concentration, they noted that for the same initial concentration, the equilibrium solid-phase concentration decreased with the increasing soil/solution ratio.
|
A Comparison of Techniques
|
|---|
To confound the situation, two different experimental techniques, batch equilibrium and soil-column leaching tests, are commonly used to study soil adsorption–desorption characteristics. Sorption studies conducted with batch methods apply to soil suspensions that are assumed to be completely dispersed, exposing all soil particle surfaces for interaction with the contaminants. Batch techniques do not remove desorbed species from the sorbing environment, so allow the products to accumulate in the system. The retained reaction products may interfere with data analysis and interpretation because (i) reactions may not be unidirectional and reverse reactions are often not considered, and (ii) desorbed species may cause secondary reactions, for example, precipitation. On the other hand, retention of the reaction products may better reproduce field conditions because surface reactions normally occur in micropores or within relative static surface films in the macropores. Mixing rates of soil and solution can also greatly affect results (Ogwada and Sparks, 1986). If mixing rates are too slow, reaction interpretation is confounded by diffusion reactions. If the mixing rate is too rapid, new specific surface area and sorption sites can be created due to abrasion as particles collide in suspension. Magnetic stirrers are particularly apt to abrade particles, but we have found that immersion stirring rates up to 580 rpm did not cause an increase in surface area. As the rate of the immersion stirrer approached 1000 rpm, however, an increase in soil surface area was observed.
Soil-column studies, on the other hand, are performed with either "intact" or "repacked" cores in which samples have a definite matrix and soil structure. Thus in column studies, all particles may not be exposed for interaction with the sorbate, and sorption capacity may be underestimated. This can even occur in thin-film "columns." As demonstrated by Wise (1993) and Sugita and Gillham (1995), spatial heterogeneity of soil chemical properties can also contribute to differences between batch and column studies. Hence, while batch studies study only metal sorption by soil suspensions in a closed system, soil-column studies evaluate both sorption and diffusion of the sorbate in an open system. While column studies may better replicate the sorbate movement through many soil environments, when soil water infiltration is retarded, batch studies may more closely approximate the behavior of the sorbate, and Streck and Richter (1997) have demonstrated that under some conditions, migration of contaminants can be adequately modeled by batch sorption. Miller et al. (1989) combined the features of batch and flow techniques in developing a continuous-flow stirred reaction cell in which the solids are kept in suspension as the solution is passed through the cell. The problem comes in attempting to compare results obtained by the three different techniques, particularly with respect to the soil/solution ratio. In the column and stirred-flow systems, the soil/solution ratio is taken as the ratio within the column or cell, whereas in batch systems, the total solution is considered to be in equilibrium with the soil. Thus unless flow rate approaches zero in the column or flow cell, soil/solution ratios are not comparable to batch systems.
Regardless, if the rate of the reaction is concentration-dependent (Eq. [5]) for any given metal addition, the adsorption reaction rate may be different but the partition coefficient will be constant.
 | (5) |
In column studies, however, the amount of M2+ sorbed is related to both solute concentration (constant) and the solution-flow rate. Bajracharya et al. (1996) compared Cd sorption under batch and flow conditions. Using the batch method (1:50, soil/solution ratio), their Cd partition coefficients (determined using linear Freundlich isotherms) were 60 to 80% higher than those from the column method (5:1, soil/solution ratio and various flow rates). Given that the partition coefficient reflects a system at equilibrium, the conclusions drawn by Bajracharya et al. (1996) regarding the widely different Kd values for column and batch studies is of concern in the absence of sufficient evidence that both systems had achieved equilibrium. Their results contrast with those of other investigators who found similar parameters from batch and column studies. For instance, Theis et al. (1988) found similar total reactive surface-site density for Cd from column and batch studies, while Boekhold and Van der Zee (1992) reported that batch-determined Freundlich sorption coefficients for Cd adequately described the observed column Cd breakthrough curves. Another factor is that columns are usually pretreated or conditioned with the background solution before initiating an experiment. If the soil used in the batch experiment is not similarly treated, differences in results between the two techniques can be expected, as illustrated by Grolimund et al. (1995). These results suggest that considerable caution must be exercised in interpreting such data, since the batch and column studies represent totally different systems and are not directly comparable. While the batch study steady-state condition is equilibrium (Fig. 12)
, steady state in column studies, with a constant solution metal concentration, is adsorption site saturation (Fig. 13)
. In batch studies, sorbate concentration decreases (Fig. 9) and desorbed ion increases (Fig. 11) with increasing equilibration time until equilibrium is attained; in column studies, ions removed from solution by adsorption are constantly being replaced and the desorbed ions are flushed out. In other words, in a batch study, steady state occurs when sorption-reaction energy is equal to desorption-reaction energy—one definition of equilibrium—but in the flow-through system the reverse reaction is eliminated so the sorption reaction must go to completion before steady state is achieved. As illustrated by Burgisser et al. (1993), if the column is viewed as a stack of sorption cells, removal of sorbate from solution can be viewed as a step function; as the sites in any one cell approach saturation, fewer M2+ ions are removed from the solution and move into the next cell. Finally "breakthrough" occurs and effluent sorbate concentration slowly increases until the effluent concentration equals the input concentration. This implies that there are more M2+ ions available for sorption in the flow-through or column studies and that greater exposure of M2+ ions increases sorption of M2+ by soil particles, consistent with the LeChatellier's principle of equilibrium. It is apparent from the above discussions that great care must be taken when comparing batch and flow-through or column systems.
View larger version (11K):
[in this window]
[in a new window]
|
Fig. 12. Relationship between the solution metal ion concentration (C) and the amount of metal ion sorbed (Msurface) during a batch study.
|
|
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 13. Relationship between influx (Co) and effluent (Cf) solution metal concentration and the metal ion retained (Msurface) during a column study.
|
|
|
Limitations of the Existing Sorption Techniques
|
|---|
The attempt to understand natural phenomena from laboratory studies has always been cause for frustration and will probably remain so in the foreseeable future. The problem lies in attempting to evaluate processes in a complex media while needing to simplify the system sufficiently to make sense of research results. Whether one is attempting to obtain sorption isotherms, evaluate kinetics of sorption, or measure the transport of solutes, the processes occurring in soils can only be approximated. As a sorption medium, a soil is a heterogeneous complex of organic matter; iron-, aluminum-, and manganese-oxides; clay minerals; and miscellaneous other minerals. Over the years, attempts have been made to model soil-sorption processes using individual components, but the overall conclusion is that the components model nothing more than the component used. The problem is that organic matter, oxides, and mineral surfaces do not exist in pure form, but organic matter and oxides coat surfaces and oxides are closely associated with organic matter. As such, their individual properties are masked or altered by the association. Zachara and co-workers (Cowan et al., 1992; Zachara et al., 1992; Zachara et al., 1993), for example, were unable to model Cd sorption by soil from mineral sorption information. Despite similar mineralogy, when ion exchange was suppressed by increasing ionic strength, Zachara et al. (1993) found that smectitic soil clays selectively sorbed more Cd than did a pure smectite clay. The soil clays exhibited more oxide-like behavior, and the authors also felt that organic matter and Fe oxides associated with surfaces acted as co-complexants of Cd. Cowan et al. (1992) were able to quantitatively model sorption by specimen minerals and oxides, whereas they were only able to qualitatively model Cd sorption by soil isolates due to imprecise estimation of sorption-site concentrations in the natural isolates.
In addition, soils contain a complex of micropores and macropores that have differing impact on solutes entering the soil. When a solute enters the soil, a combination of things happen: the solute may (i) be sorbed onto the nearest surface, (ii) diffuse into the micropores and be sorbed by surfaces there, (iii) diffuse to and react with intracrystalline sites, (iv) be precipitated with some component already in the soil, or (v) move through the soil with the leachate water. The attempt to understand the processes occurring in even one of these alternatives is a complex undertaking. Sorption studies have been designed to at least partially isolate the different effects, but this can never be done perfectly. Precipitation is the one retention mechanism that can be reasonably well controlled by keeping the concentration of individual components in solution at levels nonconducive to precipitation reactions. Even under such experimental conditions, however, one cannot be absolutely certain that no surface-catalyzed precipitation occurs. Static systems allowing components to achieve equilibrium may approximate conditions in micropores, and flow-through systems may simulate movement through the macropores, but the relative contributions of the sorption reaction and diffusion are difficult to isolate. This is a particular problem when sorption kinetics is being addressed, although Harter and Lehmann (1983) suggested that a log-normal plot of kinetics data such as that in Fig. 9 might allow separation of exchange and diffusion reactions. Finally, a major problem of soil-sorption studies is the near impossibility of measuring sorbate in the sorbed phase. In most cases, sorption can only be imperfectly obtained by difference in solution concentrations at times t = 0 and t > 0.
Now, to these complexities we must add the problem of how soils should be studied. In a very few instances it may be possible to study reactions in situ, but these situations tend to address only a fraction of the total solute–soil interaction problem. To study most soil-chemical processes, it has been necessary to transport the soil to a laboratory where experimental conditions can be carefully controlled. Over the years, a protocol has been developed for removing soils to the laboratory and stablizing them sufficiently for storage until needed. The most frequently used protocol in soil chemistry laboratories calls for air-drying the soil samples as quickly as possible at ambient temperature, grinding them to pass through a 2-mm sieve, and storing the samples until used. The objective is to minimize microbially mediated changes in the soil-chemical equilibrium, but the process of drying also has deleterious effects on the soil. For example, both moist and dry colloid particles possess Bronsted acidity. On really dry aluminosilicate surfaces, exchange H+ dehydrates and migrates to the sites of permanent negative charge within the mineral. This totally changes the interfacial chemistry when the particles are remoistened. Added to this, irreversible drying can occur in certain soils, such as those containing allophanic, other poorly ordered aluminosilicates (e.g., imogolite), and hydrous mica components. This irreversible drying will probably alter the dispersive nature of soils and thereby alter the accessibility of sorption sites to metal ions through changes in macropores and micropores within the minerals. In addition, aggregates are destroyed, the distribution of soil macropores and micropores is completely disrupted, chemical species are largely altered to their stable state, and organic bonds are broken. This becomes the material that will be used to study sorption processes in the soil. The limitations of using the resulting "soils" can be readily identified, but alternative protocols are not necessarily any better; each had its own limitations.
With trepidation, the lab soil will next be used to study soil-chemical properties and reactions. How does one simply evaluate soil-sorption capacity? In brief, a solution containing the solute must be put in contact with the soil, and the extent of reaction measured. How much solution and how much soil? As discussed above, sorption studies have been conducted with anywhere from a 1:1 to a 1:100 or wider soil/solution ratios, the amount of solution in part depending on the design needs of the experiment. Yet an in situ soil having even a 1:1 solid/solution ratio would be considered wet and would probably be in a reduced state. Thus, even if parameters such as solute concentrations, pH, and ionic strength are kept within the range expected in soil, just the required experimental conditions are a potential source of divergent results from those in the field. This is further exacerbated by the use of metal loading rates that exceed the concentrations commonly found in soil solutions. Indeed, in the soil pH range commonly found in soils, Naidu et al. (1994b) showed contrasting effects of pH on sorption of certain metals by soils with similar mineralogy. Sorption edges are commonly reported in response to changes in pH; in soils with very high sorption capacity a sorption edge has only been observed at high metal loadings. In many studies at low loadings, similar to that recorded in soil solution, no effect of pH has been observed. Does this also explain the contrasting trends observed in soil–plant transfer of metals with varying pH? What about sorption kinetics? In general, two methods are used for studying the rate of reactions: (i) a flow-through method, which may be compared with what is happening in the soil as solute moves through the macropores, and (ii) a batch method, which possibly simulates static conditions that exist in the micropores. Each has advantages and disadvantages of use. A primary problem of both is assessing the extent to which observations can be attributed to the reaction and how much is the result of diffusion processes within the experimental system. Despite the increasing sophistication and complexity of experimental techniques, the question of whether diffusion has really been minimized will probably continue to be raised as a potential source of calculation error.
|
The Future of Sorption Research
|
|---|
So where are we? The current status of metal-sorption research is not unlike the status of cation exchange research before 1940, when Bower and Truog published their article showing that CEC depends on the ion used for its measurement. As more diverse parameters such as the effect of low molecular weight organics or other ligands are incorporated into sorption studies, results can be compared between laboratories only if there is some basis for comparison. Ironically, conferences are held worldwide to discuss the behavior of solutes in soils but limited effort has been directed toward the standardization of techniques that may enable one to make interlaboratory comparisons. We suggest that at least a minimum sorption-study standardization be adopted. We suggest that in all sorption studies, (i) at least one experiment using a 10 mmol L-1 NaNO3 background electrolyte be included; (ii) that all at least include a treatment conducted at 25°C ± 3°C, and (iii) that so far as possible, at least one treatment in each study be at pH between 5.5 and 6. The method of pH adjustment needs to be negotiated among soil chemists, but is probably best accomplished by adding a liming material and stabilizing by incubation with alternate wetting and drying. The choice of background electrolyte can also be negotiated; NaNO3 is probably the best choice, being a salt with environmental implications but minimum probability of forming complexes at the ionic strengths used for these studies. This amount of standardization should allow at least some comparison among laboratories and provide a better framework for evaluating the impact of chosen variables on metal sorption by differing soils and minerals.
|
NOTES
|
|---|
TOP
ABSTRACT
INTRODUCTION
C) per mole of added P (
