A General Model for Kinetics of Heavy Metal Adsorption and

Mar 13, 2013 - Zhenqing Shi*†, Dominic M. Di Toro‡, Herbert E. Allen‡, and Donald L. Sparks§. † School of the Environment, Washington State U...
0 downloads 8 Views 2MB Size
Article pubs.acs.org/est

A General Model for Kinetics of Heavy Metal Adsorption and Desorption on Soils Zhenqing Shi,*,† Dominic M. Di Toro,‡ Herbert E. Allen,‡ and Donald L. Sparks§ †

School of the Environment, Washington State University, Pullman, Washington 99164, United States Center for the Study of Metals in the Environment, Department of Civil and Environmental Engineering, University of Delaware, Newark, Delaware 19716, United States § Environmental Soil Chemistry Research Group, Department of Plant and Soil Sciences, University of Delaware, 152 Townsend Hall, Newark, Delaware 19717, United States ‡

S Supporting Information *

ABSTRACT: In this study, we propose a general kinetics model for heavy metal adsorption and desorption reactions in soils when soil organic matter (SOM) is the dominant adsorbent. The kinetics model, integrated with the equilibrium speciation model WHAM VI, specifically considers metal reactions with SOM and dissolved organic matter (DOM) and accounts for the variations of solution chemistry. Metal reactions with SOM are associated with two groups of sites, one from the monodentate sites and another one from the bidentate and tridentate sites. There are three model parameters, desorption rate coefficients of the two groups of SOM sites for each metal and reactive organic carbon (ROC) for each soil. The applicability of the kinetics model was mainly examined with three elements, Cu, Pb, and Zn, which demonstrate different binding ability with organic matter. The kinetic data were collected with a stirred-flow reactor covering a wide range of experimental conditions, including varying SOM, DOM, Ca, and metal concentrations, reaction pHs, and different flow rates. The kinetics model has been successfully applied to describe heavy metal adsorption and desorption on soils under various reaction conditions.



INTRODUCTION The quantitative understanding of the kinetics of heavy metal adsorption and desorption on soils helps to accurately predict dynamic behavior of heavy metals in the environment. Heavy metal reactions with soils are affected by both solution chemistry (e.g., pH, cation competition, metal concentrations, etc.) and soil composition (e.g., soil organic matter (SOM) content). Developing predictive kinetics models requires consideration of the heterogeneity of soil binding sites and variation of reaction chemistry, and remains a challenge compared with extensive studies on predicting metal partitioning equilibrium in soils using the mechanistic-based speciation models.1−3 In contrast to numerous empirical kinetics models that generally have limited applicability to complex soil systems,4 we have recently developed a mechanistic-based modeling approach for heavy metal adsorption/desorption kinetic reactions in soils.5,6 By integrating the mechanistic-based equilibrium model WHAM,7,8 the kinetics model, in principle, is able to account for the effects of solution chemistry and soil composition on metal adsorption/desorption kinetics in soils. Based an early version of WHAM V,7 the kinetics model was examined using Zn adsorption/desorption kinetic data and was successful in reproducing the experimental results over a wide range of Zn concentrations, SOM contents, and reaction pHs. However, the applicability of this modeling approach to other © 2013 American Chemical Society

metals, especially those demonstrating different reactivity with SOM, has not been tested. Furthermore, for a truly predictive model, the same model input parameters for soil properties should be applicable for different metals, and the model should have the ability to account for additional ligands/adsorbents for metal reactions. For example, the dissolved organic matter (DOM) may significantly affect metal speciation in solutions depending on the affinity of the metal to DOM, which was not considered in our previous study with Zn.6 In this study, we propose a general kinetics model for heavy metal adsorption and desorption on soils under various reaction conditions, including different soils with varying SOM concentrations, different pHs, and various heavy metal, Ca, and DOM concentrations. A more advanced version of WHAM, WHAM VI,8 is used in this study since the performance of WHAM can significantly affect the accuracy of the kinetics model.6,9 We mainly consider Cu, Pb, and Zn as three representative elements, which have different reactivity with SOM. The applicability of the kinetics model to other heavy metals (e.g., Cd and Ni) is also discussed. Received: Revised: Accepted: Published: 3761

November 6, 2012 February 26, 2013 March 13, 2013 March 13, 2013 dx.doi.org/10.1021/es304524p | Environ. Sci. Technol. 2013, 47, 3761−3767

Environmental Science & Technology

Article

Table 1. Selected Soil Properties particle size composition

a

soil

soil location

pHa

sand %

silt %

clay %

OCb %

fraction of reactive OCc

Boonton Bergen County Loam Boonton Union Loam Codorus Clay Loam Matapeake Silt Loam Montpellier Loamy Sand Nottingham Sandy Loam Rhydtalog Loamy Sand

New Jersey New Jersey Maryland Delaware France United Kingdom United Kingdom

4.9 4.9 6.0 6.4 6.4 4.2 4.7

60 49 27 13 87 64 51

27 35 45 63 4 23 36

13 16 28 24 9 13 13

3.43 7.15 2.53 2.32 0.76 5.20 12.90

0.37 0.30 0.47 0.49 0.60 0.30 0.20

Measured with DI H2O. bMeasured with a Variomax CN analyzer. cObtained from modeling (see details in the manuscript).



MATERIALS AND METHODS Soil Samples. Uncontaminated soils, sampled from the U.S. and European countries, were used to conduct adsorption and desorption experiments. All soil samples were obtained from the 0−20 cm layer, air-dried, and then sieved using a 2 mm screen. These acidic soils, with low clay fractions and low heavy metal concentrations, were selected to cover a wide range of SOM concentrations. Under our experimental conditions, SOM was assumed to be the dominant adsorbent due to the high affinity of metal ions to humic substances.1−3,10 Selected properties of soils for Cu and Zn experiments are presented in Table 1. The properties of soils for the Pb study can be found in the previous study,11 in which one low SOM and another high SOM soil was used. Kinetic Experiments. Two types of experiments were conducted in a stirred-flow reactor. The first type of experiments used uncontaminated soils to study the kinetics of metal adsorption followed by desorption, and the second type of experiments used soils spiked with heavy metals to study the kinetics of metal desorption. (i). Metal Adsorption and Desorption Experiments. Besides the Zn adsorption and desorption kinetic data we collected previously,6 Cu adsorption and desorption kinetic experiments were conducted using the same stirred-flow chamber and experimental setup. Briefly, a 0.3 g soil sample and a Tefloncoated magnetic stir bar were placed into the reaction cell (volume = 6.4 cm3), which was then filled with the background electrolyte (3 mM Ca(NO3)2). A 25 mm diameter filter membrane with a 0.45 μm pore size was used to retain the soil in the reaction cell. The pH (5.5−6.5) was kept at the desired value using 3 mM MES ([2-(N-morpholino) ethane sulfonic acid]) which does not complex metal ions.12 A Cu(NO3)2 stock solution was added to the background electrolyte to prepare different concentrations of Cu solutions (0.85−1.68 mg L−1). To initiate the adsorption experiment, the metal solution was pumped through the chamber. After 3 h of adsorption, desorption was initiated by passing the background electrolyte through the chamber. The desorption continued for another 5 h. For selected experiments, the flow was stopped for 30 min during the adsorption process to determine that the reaction was rate limited rather than being at instantaneous equilibrium.5 The flow rate for all experiments was 0.9 mL min−1. Using this experimental setup, we systematically examined the effects of SOM, Cu concentrations, and pH on Cu adsorption and desorption kinetics. The Cd and Ni adsorption and desorption kinetic experiments were also run for comparison at one selected reaction condition. The kinetic data of Pb adsorption and desorption on soils were published previously,11 and the experiments were

performed with a similar procedure for Cu and Zn experiments. Compared with Cu and Zn experiments, Pb experiments used a slightly larger stirred-flow chamber (8 mL), more soil particles (0.75 g), different background electrolyte (0.05 M NaNO3), and a much higher metal concentration (480 mg L−1). The effects of SOM and flow rates on Pb adsorption and desorption were studied. (ii). Cu and Zn Desorption from Metal Spiked Soils. Two soils, the Matapeake and the Codorus soils, were spiked with Cu and Zn and then air-dried. The detailed metal spiking procedure is presented in the Supporting Information (SI). A 0.3 g spiked soil sample was put in the stirred-flow chamber and preequilibrated with the background electrolyte for 20 min without flow. Then the background electrolyte was continuously pumped through the chamber at a fixed flow rate. Details about the experiments can be found elsewhere.5 Besides the data we published previously,5 we conducted additional experiments and the total experiments matrix include varying DOM concentrations (0 − 15 mg C L−1), Ca concentrations (1 mM to 10 mM), pHs (5.5−6.5), and flow rates (0.5−4 mL min−1). Effluent samples were collected using a fraction collector with fixed time interval. Since the filter membrane with a 0.45 μm pore size was used to retain the soil particles, the solution samples may contain colloidal particles smaller than 0.45 μm. The metal concentrations in solution samples were analyzed by ICP-MS or ICP-OES. Model Descriptions. Two groups of reaction sites are considered in the kinetics model, one fast desorption site, labeled as site 1, and the other slow desorption site, labeled as site 2. The kinetic equations for metal adsorption and desorption on the two sites are dCp1 dt

= −kd1Cp1 + ka1C ion

(1)

= −kd2Cp2 + ka2C ion

(2)

and dCp2 dt

where ka1,2 (L (g min)−1), kd1,2 (min−1), and Cp1,2 (μg g−1) are adsorption and desorption rate coefficient and soil metal concentration for sites 1 and 2, respectively, and Cion (μg L−1) is the solution concentration of the ionic metal. The equilibrium model WHAM VI is integrated into the kinetics model to account for the effects of SOM concentrations and solution chemistry on reaction kinetics, using the approach we developed previously.6 In WHAM VI, metal ions can bind to either monodentate, bidentate, or tridentate sites of SOM, and also form outer-sphere complexes in the electrical 3762

dx.doi.org/10.1021/es304524p | Environ. Sci. Technol. 2013, 47, 3761−3767

Environmental Science & Technology

Article

dC ion = −kd1K p1mC ion − kd2K p2mC ion + kd1mCp1 dt Q (C ion − C ion,0) + kd2mCp2 − − kd2K p,DOM V

double layer via electrostatic interactions. For the desorption reactions, the bidentate and tridentate complexes involve breaking more than one metal-surface bonds. It is likely that these desorption reactions have slower rates compared with the desorption reactions of the monodentate complexes. Therefore, we assume that the fast desorption reactions are from the monodentate sites and the slow desorption reactions are from the bidentate and tridentate sites. The formation of outersphere complexes is minimal in our experimental conditions, which has little effect on the overall model performance. Therefore, the desorption of outer-sphere complexes is incorporated into the fast desorption reactions to simplify the model calculations. The effect of solution chemistry is incorporated into the adsorption reaction, and the desorption rate coefficients are assumed to be constant irrespective of the reaction conditions.5,6 The adsorption rate coefficients at specific reaction times can be calculated based on the equilibrium partition coefficient, Kpi (L g−1), predicted by WHAM VI for site 1 and 2, respectively kai = kdiK pi(i = 1, 2)

C DOMC ion + kd2CMe − DOM

and dCMe − DOM = kd2K p,DOMC DOMC ion − kd2CMe − DOM dt Q (CMe − DOM − CMe − DOM,0) − V

(6)

where m (g L−1) is the soil particle concentration, Q (L min−1) is the flow rate, and V (L) is the reaction volume of the reactor. Subscript 0 denotes the influent metal concentration. Model Calculations. For model calculations, the desorption rate coefficients, kd1 and kd2, and WHAM predicted partition coefficients, Kp1, Kp2, and Kp,DOM, need to be determined. The major input parameters for WHAM VI calculations include concentrations of particulate humic acid (HA) and fulvic acid (FA), dissolved FA, solution cation concentrations or activities, anion concentrations, and pH. Since SOM consists of a number of components different from humic substances, not all SOM is as active as humic substances for metal binding. A fraction of soil organic carbon (SOC), defined as reactive organic carbon (ROC), is considered responsible for metal binding to soils. Both ROC and dissolved organic carbon (DOC) were converted to organic matter by multiplying a factor of 1.8. The reactive SOM was input as 82% HA and 18% FA, and 65% of DOM was input as dissolved FA in WHAM VI.2 The solution pH was input as the experimental values. Total dissolved cations were input as measured. For anions, NO3−, which is the major anion in all kinetic experiments, was input to balance the charge. The Fe and Al competition effect was accounted for by assuming that Fe(III) and Al activities were controlled by the solubility of their metal hydroxides with solubility constant Ks0, 8.5 for Al and 3 for Fe(III), respectively.2 Additional information about WHAM VI input parameters is presented in the SI. The ROC for each soil and the desorption rate coefficients were obtained from data fitting. We first globally fit all Zn kinetic data from the adsorption and desorption experiments, which cover a wide range of experimental conditions, to obtain ROC for each soil and the desorption rate coefficients for Zn. For Cu and Pb, the same ROC values were used for each soil and the desorption rate coefficients were obtained by fitting the adsorption and desorption kinetic data, respectively. For any soil not used in the Zn experiments, the ROC value was calculated by linear interpolation of ROC values of other soils and the results are presented in the SI. In summary, there are three model fitting parameters, the desorption rate coefficients of the two groups of SOM sites for each metal and the ROC for each soil. Since WHAM VI has been calibrated over a wide range of reaction conditions, there is no need to add empirical fitting parameters to account for the variations in reaction chemistry. To validate the model, the Cu and Zn desorption kinetics was predicted for metal-spiked soils under various conditions and the model predictions were compared with the independently collected data. An implicit finite difference numerical method was used to solve the model equations. The experimental data sets were

(3)

The reaction chemistry and metal nonlinear binding to SOM are accounted for by eq 3, or, specifically, by the WHAM VI predicted equilibrium partition coefficients Kpi at specific reaction conditions. More details are described in the SI. Although not considered in our previous modeling study,6 DOM may affect metal speciation in solutions by complexing metal ions and thus reduce the adsorption rates of the ionic metal. In theory, the effect of DOM can be considered with a similar approach used for metal reactions with SOM. In our experiments with DOM, metal binding to the tridentate and bidentate sites of DOM dominated the overall reaction according to WHAM VI calculations since the total dissolved metal concentrations were low. Thus, to simplify the model calculations, we only use one group of sites, site 2, to account for the metal kinetic reactions with DOM. Since both SOM and DOM are input as humic substances in WHAM VI, we use the desorption rate coefficient for site 2, kd2, as the dissociation rate coefficient of metal reactions with DOM to avoid adding additional fitting parameters. Similarly, the complexation and dissociation rate coefficients of metal reactions with DOM are constrained by the WHAM calculated equilibrium partition coefficient for DOM, Kp,DOM (L (mg C)−1) k f = k bK p,DOM = kd2K p,DOM

(5)

(4)

where kf and kb are complexation and dissociation rate coefficients, respectively. Thus no additional model parameters are added to account for the effect of DOM in our kinetics model. Note that, when the total dissolved metals are high, metal binding to the monodentate sites of DOM can be added into the kinetics model using the same approach. In addition, the interaction between DOM and soil particles is not considered in the current model. To incorporate the effect of DOM into the kinetics model, the total DOM concentration, CDOM (mg C L−1), is used to represent the total reactive ligand concentration of DOM. The mass balance equations for ionic metal concentration, Cion, and metal−DOM complex concentration, CMe‑DOM (μg metal L−1), are 3763

dx.doi.org/10.1021/es304524p | Environ. Sci. Technol. 2013, 47, 3761−3767

Environmental Science & Technology

Article

effluent Zn (Figure 1A) and Cu (Figure 1B) concentrations at higher SOC concentrations during the adsorption process. Much more Cu is adsorbed than Zn under similar experimental conditions based on mass balance calculations of the effluent data (Figure 1), and Cu adsorption also shows stronger pH dependency than Zn adsorption (Figures S1B and S2A, SI). The different adsorption kinetics between Cu and Zn, consistent with the facts that Cu forms stronger complexes with SOM than Zn and the molar exchange ratio of H+/Cu2+ is much higher than that of H+/Zn2+ for metal reactions with natural organic matter,13,14 is explained by the kinetics model through the WHAM VI predicted partition coefficients via the adsorption rate coefficients (eq 3). The Cu concentrations decreased during the stop-flow experiments (Figure 1B), demonstrating that Cu adsorption reaction is rate-limited rather than instantaneous. The model correctly reproduced these observations as well. The kinetics of Cd, Cu, Ni, and Zn adsorption and desorption under similar experimental conditions is presented in Figure 2. While Cu demonstrates significantly different

tabulated in Microsoft Excel 2003 spreadsheets. For each observation time, the square of the difference between the measured and the model calculated dissolved metal concentrations was calculated. The sum of the squares for each data set was summed to obtain the total squared error. The SOLVER program in Excel was used to obtain the model fitting parameters by minimizing the total squared error.



RESULTS AND DISCUSSION The kinetics model based on WHAM VI reproduced the observations over the range of SOC concentrations, reaction pH, and metal concentrations for both Zn and Cu adsorption and desorption kinetics (Figure 1, Figures S1−S2, SI). The

Figure 2. Kinetics of Cd, Cu, Ni, and Zn adsorption and desorption on the Matapeake soil (pH 6.0; Influent containing similar molar concentrations of heavy metals (μM): [Cd] = 23.7, [Cu] = 22.7, [Ni] = 26.4, [Zn] = 27.5). Influent metal concentrations during the adsorption experiments are annotated in the plot. Soil particle concentration m = 46.9 g L−1. Flow rate Q = 0.9 mL min−1. Solid lines are model calculations.

adsorption and desorption kinetics as discussed previously, the effluent concentrations of Cd, Ni, and Zn approached the influent concentrations after about 2 h adsorption and the effluent concentrations decreased quickly during the desorption process. The adsorption and desorption curves of Cd, Ni, and Zn almost overlap after normalization by the influent metal concentrations (Figure S3, SI) as Cd, Ni, and Zn all form weak complexes with humic substances. Therefore, we tested the model applicability to both Cd and Ni by using the same desorption rate coefficients as Zn. It is promising to see that the kinetics model calculations reasonably match the experimental data with WHAM VI accounting for the difference of metals. Although previous research using different experimental techniques reported that the desorption rate coefficients of Cd, Ni, and Zn differed,15−17 our results suggest similarity of kinetic behavior for Cd, Ni, and Zn under the combined effects of adsorption, desorption reactions and the flow impact, and our kinetics model has the capability to extend to other heavy metals based on WHAM VI.

Figure 1. Kinetics of heavy metal adsorption and desorption on soils with different SOC concentrations. (A) Zn at pH 6.0; and (B) Cu at pH 5.5. Influent Zn or Cu concentrations during the adsorption experiments are annotated in each plot. Soil particle concentration m = 46.9 g L−1. Flow rate Q = 0.9 mL min−1. Solid lines are model calculations.

same ROC values (Table 1), obtained from Zn modeling results, are applicable to modeling Cu adsorption and desorption kinetics. The kinetics of Zn and Cu adsorption and desorption on soils with various SOC concentrations is presented in Figure 1. Overall, except for the stop-flow events, effluent metal concentrations increased with time during the adsorption process due to less adsorption sites available with more metal adsorbed, and during the desorption process the effluent metal concentrations decreased with time as adsorbed metals were released and continuously removed from the reactor (eq 5). Higher SOC concentrations resulted in higher adsorption of both Zn and Cu, as demonstrated by the lower 3764

dx.doi.org/10.1021/es304524p | Environ. Sci. Technol. 2013, 47, 3761−3767

Environmental Science & Technology

Article

reactions with soils are rapid. There have been few studies to assess WHAM VI′s ability to predict the reaction of Pb with SOM or DOM.2,18,19 The success of our kinetics model to account for Pb adsorption/desorption kinetics suggests that WHAM VI has captured key features of Pb reactions with SOM and our kinetics model has general applicability to different heavy metals. The kinetics of metal adsorption and desorption on different soil binding sites were analyzed using model calculations (Figure 4). The total Cu and Zn concentrations in soils (Cp) were relatively low compared with the total Pb concentrations in soils and the bidentate and tridentate sites dominated both Cu and Zn adsorption (Figure 4A and B). However, Pb

The kinetics model also successfully reproduced the kinetics of Pb adsorption and desorption under different flow rates and SOM concentrations (Figure 3). The kinetics of Pb adsorption

Figure 3. Kinetics of Pb adsorption/desorption on soils at pH 5.5: (A) Pb adsorption on soils with different SOM concentrations and flow rates (Influent [Pb] = 480 mg L−1); and (B) Pb adsorption and desorption on the soil with 2.1% SOM at two flow rates. Soil particle concentration m = 93.8 g L−1. Solid lines are model calculations. Note that the results were plotted versus chamber volumes to normalize the effect of flow rates.

on soils is characterized by a quick Pb adsorption in the first a few chamber volumes and the adsorption approaches the adsorption capacity of the soils after about 7−10 chamber volumes (Figure 3A). Recall that Pb kinetic experiments were run at very high Pb concentrations. The fast Pb adsorption is due to the combined effects of high Pb concentrations and the large desorption rate coefficient of the bidentate and tridentate sites (Table 2). The significant drop of Pb concentrations in the solution during the stop-flow experiment confirms that Pb adsorption on soils is rate-limited (Figure 3A). The small effect of flow rates on Pb adsorption and desorption kinetics, as shown by plotting effluent Pb concentrations vs chamber volumes (Figure 3B), is consistent with the fact that Pb Table 2. Desorption Rate Coefficients of Soil Binding Sites

Cu Pb Zn

monodentate sites (kd1) (s−1)

bidentate and tridentate sites (kd2) (s−1)

5.4 × 10−2 3.6 × 10−3 2.9 × 10−1

2.5 × 10−4 1.9 × 10−3 2.7 × 10−4

Figure 4. Kinetics of metal adsorption/desorption on the monodentate sites (Cp1), the bidentate and tridentate sites (Cp2), and the total soil sites (Cp) of the Matapeake soil (2.32% SOC) at pH 5.5. Influent metal concentrations during adsorption: (A) [Zn] = 1.91 mg L−1, (B) [Cu] = 1.67 mg L−1, and (C) [Pb] = 481 mg L−1. 3765

dx.doi.org/10.1021/es304524p | Environ. Sci. Technol. 2013, 47, 3761−3767

Environmental Science & Technology

Article

adsorption to both the monodentate sites (Cp1) and the bidentate and tridentate sites (Cp2) were significant since the total Pb concentrations in soils were much higher. Overall, metal reactions with the monodentate sites were faster than with the bidentate and tridentate sites in most conditions, as characterized by both larger adsorption and desorption rate coefficients (Figure 4, Table 2). The bidentate and tridentate sites controlled the long-term metal slow release from soils. Among three metals, the desorption of Zn from the bidentate and tridentate sites was much faster compared with that of Cu and Pb, although the desorption rate coefficient of Zn (kd2) was close to that of Cu and even smaller than that of Pb for the bidentate and tridentate sites (Table 2). This is due to the much weaker binding ability of Zn to SOM, which results in smaller adsorption rate coefficients and faster metal release overall (eqs 1−3). As also shown in Figure 4, the adsorption rate coefficients varied significantly during the kinetic experiments even at the constant pH, and they are highly dependent on the total metal concentrations in each group of SOM sites and the reactivity of the specific metal with SOM. The variations of adsorption rate coefficients are consistent with the nonlinear binding behavior of heavy metals. The wide range of adsorption rate coefficients observed in this study highlights that the nonlinearity of metal binding to SOM should be considered in kinetics models and the assumptions based on a linear adsorption isotherm are usually not appropriate in most reaction conditions. For model validation, we applied the model parameters to predict the kinetics of Cu and Zn release from two metal-spiked soils under the effects of DOM, Ca competition, pH, and flow rates, and compared the model predictions with the data collected independently. The initial distributions of Cu and Zn among different SOM binding sites after Cu and Zn spiking were calculated by WHAM VI (Table S1, SI). Overall, the concentrations of the total dissolved Cu and Zn in the reactor increased to high values during the 20 min pre-equilibration period, and then the effluent Cu and Zn concentrations decreased with the desorption time (Figure 5 and Figures S4, S5, S7−S9, SI). The increase in DOM concentrations significantly enhanced Cu release from both soils (Figure 5A and B) but, surprisingly, had little effect on Zn release (Figure 5C). This is explained by the strong complexation of free Cu ions by DOM, which significantly reduced the free Cu ion concentrations (Figure S4A, SI) and thus inhibited the readsorption of free Cu ions to soils. In contrast, a much weaker binding of free Zn ions to DOM resulted in little difference between the free Zn ion concentrations and the total dissolved Zn concentrations (Figure S4B, SI), due to the slow Zn complexation reactions in the flow-through reactor (eq 6). Recall that the complexation rate coefficients of metal reactions with DOM were directly derived from WHAM VI with the desorption rate coefficients (kd2) of SOM (eq 4). Without adding additional model parameters, the kinetics model accounts very well for the DOM effect for both Cu and Zn, which supports the validity of our modeling approach for DOM reactions. Our kinetics model also reasonably predicted the effect of Ca competition on both Cu and Zn desorption kinetics. The increase in Ca concentrations enhanced both Cu and Zn release from the spiked soils (Figure S5A and S5B, SI), explained by the decrease in equilibrium partition coefficients with Ca concentrations (Figure S6, SI), although the Cu desorption

Figure 5. Kinetics of Cu and Zn desorption from spiked soils at different DOM concentrations ([Ca] = 3 mM, pH 5.5): (A) Cu with the Matapeake soil; (B) Cu with the Codorus soil; and (C) Zn with the Matapeake soil. Soil particle concentration m = 46.9 g L−1. Flow rate Q = 1 mL min−1. Solid lines are model calculations.

kinetics was complicated by the soluble organic matter release from soils in the pre-equilibration period.5 Similar to the results discussed previously, the kinetics model reasonably predicted the effects of pH and the flow rates on both Cu and Zn release from spiked soils (Figures S7−S9, SI). Model Evaluation and Applicability. The modelcalculated kinetics of metal adsorption and desorption is controlled by the desorption rate coefficients and the WHAM predicted partition coefficients at specific reaction conditions. The ability of WHAM VI to predict metal partitioning between soils and solutions has been extensively studied.2,20−22 One important finding of our kinetics model is that the fraction of ROC for each soil is applicable for all heavy metals studied. The association of the monodentate sites and the bidentate and tridentate sites with the kinetically fast and slow sites, respectively, has shown success for a number of heavy metals. In our kinetics model, the desorption rate coefficients are constant for each group of sites, irrespective SOM concentrations and solution chemistry. Previous kinetic studies using the technique of DGT (diffusive gradients in thin films) reported that Zn desorption rate coefficients varied from 7.7 × 10−6 to 1.1 × 10−3 s−1 for soils spiked with heavy metals.16,17,23 The Zn desorption rate coefficient of the bidentate and tridentate sites (2.7 × 10−4 s−1) in our study is within this range. The Cu desorption rate coefficients obtained from this study (Table 2) are larger than those obtained from the DGT 3766

dx.doi.org/10.1021/es304524p | Environ. Sci. Technol. 2013, 47, 3761−3767

Environmental Science & Technology

Article

experiments (1.5 × 10−6 s−1).16 The wide range of desorption rate coefficients reported in the literature suggests those values are affected by the specific experimental conditions and the approaches used to obtain the rate coefficients, and thus may not represent the true desorption rate constants. It is promising to see that the model is applicable to a number of heavy metals that have different reactivity with SOM. The model shows predictive ability. The modeling approach provides a general framework to develop predictive models for the kinetic reactions of heavy metals with soils in natural environments if SOM is the major adsorbent. Although the impact of mineral phases is not discussed in this study, the existing kinetics model can be easily expanded to include metal reactions with mineral phases when needed, such as sorption to clay minerals and precipitation, as shown in our previous study.9 Currently, all model parameters were obtained based on the data collected in a well-controlled stirred-flow reactor. In the field conditions, metal reactions with soils are complicated by other kinetic processes such as diffusion in soil pore water, transport by water flow, and uptake by organisms. For the field contaminated soils with long aging time, the metal speciation could differ significantly from those in laboratory short-term adsorption experiments,24 and the metal release from the field contaminated soils may be slower than that from freshly spiked samples.23 The kinetics model would need to be modified under these conditions. Therefore, it is desired to examine this model in a more realistic soil system in the field, and the applicability of laboratory-derived model parameters to field situations must be established.



(6) Shi, Z.; Di Toro, D. M.; Allen, H. E.; Sparks, D. L. A WHAMbased kinetics model for Zn adsorption and desorption to soils. Environ. Sci. Technol. 2008, 42, 5630−5636. (7) Tipping, E. WHAM − A chemical equilibrium model and computer code for waters, sediment, and soils incorporating a discrete site/electrostatic model of ion-binding by humic substances. Comput. Geosci. 1994, 20, 973−1023. (8) Tipping, E. Humic ion-binding model VI: An improved description of the interactions of protons and metal ions with humic substances. Aquat. Geochem. 1998, 4, 3−48. (9) Shi, Z.; Peltier, E.; Sparks, D. L. Kinetics of Ni sorption in soils: Roles of soil organic matter and Ni precipitation. Environ. Sci. Technol. 2012, 46, 2212−2219. (10) Weng, L. P.; Wolthoorn, A.; Lexmond, T. M.; Temminghoff, E. J. M.; van Riemsdijk, W. H. Understanding the effects of soil characteristics on phytotoxicity and bioavailability of nickel using speciation models. Environ. Sci. Technol. 2004, 38, 156−162. (11) Strawn, D. G.; Sparks, D. L. Effects of soil organic matter on the kinetics and mechanisms of Pb(II) sorption and desorption in soil. Soil Sci. Soc. Am. J. 2000, 64, 144−156. (12) Kandegedara, A.; Rorabacher, D. B. Noncomplexing tertiary amines as ″better″ buffers covering the range of pH 3−11. Temperature dependence of their acid dissociation constants. Anal. Chem. 1999, 71, 3140−3144. (13) Benedetti, M. F.; Milne, C. J.; Kinniburgh, D. G.; van Riemsdijk, W. H.; Koopal, L. K. Metal-ion binding to humic substances Application of the nonideal competitive adsorption model. Environ. Sci. Technol. 1995, 29, 446−457. (14) Koopal, L. K.; Saito, T.; Pinheiro, J. P.; van Riemsdijk, W. H. Ion binding to natural organic matter: General considerations and the NICA-Donnan model. Colloid Surf. A 2005, 265, 40−54. (15) Sekaly, A. L. R.; Murimboh, J.; Hassan, N. M.; Mandal, R.; Ben Younes, M. E.; Chakrabarti, C. L.; Back, M. H.; Gregoire, D. C. Kinetic speciation of Co(II), Ni(II), Cu(II), and Zn(II) in model solutions and freshwaters: Lability and the d electron configuration. Environ. Sci. Technol. 2003, 37 (1), 68−74. (16) Ernstberger, H.; Davison, W.; Zhang, H.; Tye, A.; Young, S. Measurement and dynamic modeling of trace metal mobilization in soils using DGT and DIFS. Environ. Sci. Technol. 2002, 36, 349−354. (17) Ernstberger, H.; Zhang, H.; Tye, A.; Young, S.; Davison, W. Desorption kinetics of Cd, Zn, and Ni measured in soils by DGT. Environ. Sci. Technol. 2005, 39, 1591−1597. (18) Warnken, K. W.; Lawlor, A. J.; Lofts, S.; Tipping, E.; Davison, W.; Zhang, H. In situ speciation measurements of trace metals in headwater streams. Environ. Sci. Technol. 2009, 43, 7230−7236. (19) Gustafsson, J. P.; Tiberg, C.; Edkymish, A.; Kleja, D. B. Modelling lead(II) sorption to ferrihydrite and soil organic matter. Environ. Chem. 2011, 8, 485−492. (20) Almas, A. R.; Lofts, S.; Mulder, J.; Tipping, E. Solubility of major cations and Cu, Zn and Cd in soil extracts of some contaminated agricultural soils near a zinc smelter in Norway: Modelling with a multisurface extension of WHAM. Eur. J. Soil Sci. 2007, 58, 1074− 1086. (21) Shi, Z.; Allen, H. E.; Di Toro, D. M.; Lee, S.-Z.; Flores Meza, D. M.; Lofts, S. Predicting cadmium adsorption on soils using WHAM VI. Chemosphere 2007, 69, 605−612. (22) Tipping, E. Modelling the interactions of Hg(II) and methylmercury with humic substances using WHAM/Model VI. Appl. Geochem. 2007, 22, 1624−1635. (23) Zhang, H.; Lombi, E.; Smolders, E.; McGrath, S. Kinetics of Zn release in soils and prediction of Zn concentration in plants using diffusive gradients in thin films. Environ. Sci. Technol. 2004, 38, 3608− 3613. (24) Roberts, D. R.; Scheinost, A. C.; Sparks, D. L. Zinc speciation in a smelter-contaminated soil profile using bulk and microspectroscopic techniques. Environ. Sci. Technol. 2002, 36, 1742−1750.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information includes the details on WHAM VI input parameters and additional modeling results. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*Phone: 509-335-7857; e-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS Funding for this project was provided by the U.S. Environmental Protection Agency through the Center for the Study of Metals in the Environment at the University of Delaware.



REFERENCES

(1) Gustafsson, J. P.; Pechova, P.; Berggren, D. Modeling metal binding to soils: The role of natural organic matter. Environ. Sci. Technol. 2003, 37, 2767−2774. (2) Tipping, E.; Rieuwerts, J.; Pan, G.; Ashmore, M. R.; Lofts, S.; Hill, M. T. R.; Farago, M. E.; Thornton, I. The solid-solution partitioning of heavy metals (Cu, Zn, Cd, Pb) in upland soils of England and Wales. Environ. Pollut. 2003, 125, 213−225. (3) Weng, L. P.; Temminghoff, E. J. M.; van Riemsdijk, W. H. Contribution of individual sorbents to the control of heavy metal activity in sandy soil. Environ. Sci. Technol. 2001, 35, 4436−4443. (4) Sparks, D. L. Kinetics of Soil Chemical Processes; Academic Press: San Diego, CA, 1989. (5) Shi, Z.; Di Toro, D. M.; Allen, H. E.; Ponizovsky, A. A. Modeling kinetics of Cu and Zn release from soils. Environ. Sci. Technol. 2005, 39, 4562−4568. 3767

dx.doi.org/10.1021/es304524p | Environ. Sci. Technol. 2013, 47, 3761−3767