Critical Evaluation of Desorption Phenomena of Heavy Metals from

Rice University, 6100 South Main Street, MS-519,. Houston, Texas 77005-1892. In natural sediments, the majority of heavy metal ions are generally asso...
0 downloads 0 Views 167KB Size
Environ. Sci. Technol. 2003, 37, 5566-5573

Critical Evaluation of Desorption Phenomena of Heavy Metals from Natural Sediments YAN GAO, AMY T. KAN,* AND MASON B. TOMSON Department of Civil and Environmental Engineering, Rice University, 6100 South Main Street, MS-519, Houston, Texas 77005-1892

In natural sediments, the majority of heavy metal ions are generally associated with the solid phase. To become bioavailable, the metal ions must desorb from the solid. Numerous studies of heavy metals in sediments have suggested that sorption and desorption exhibit hysteresis (i.e., the two processes are not reversible), while other studies have suggested that desorption hysteresis does not exist. In this study, sorption/desorption hysteresis of lead (Pb) and cadmium (Cd) was evaluated over the following range of conditions: (i) desorption induced by replacing the supernatant liquid with contaminant-free electrolyte solution; (ii) desorption induced by lowering the solution pH with mineral acid; and (iii) desorption induced by sequestration with EDTA. Given the importance of dissolved organic and inorganic ligands in regulating heavy metal behavior in nature sediments, sorption/desorption experiments were conducted on both untreated and prewashed sediments. Prewashing treatment increases the sorption potential of Cd but not Pb. Desorption hysteresis is observed in both the untreated and the prewashed sediments using the replaced supernatant method, and the desorption hysteresis appears to increase with aging time. Hysteresis is not observed when desorption is initiated by lowering the solution pH. A large fraction of the sorbed heavy metal ions can be easily desorbed by EDTA; between 0.04 and 1.2 mmol/kg Cd and Pb ions are resistant to desorption.

Introduction Sorption/desorption processes have been found to be important for heavy metals and may significantly affect their toxicity and bioavailability in natural environments. Numerous laboratory experiments have demonstrated that sorption of organic compounds onto natural soils is not a fully reversible process (i.e., sorption and desorption often exhibit hysteresis) (1). Sorption/desorption behavior of heavy metals has also been investigated, but mixed results have been reported (2-9). Di Toro et al. (2) presented nickel and cobalt montmorillonite sorption data and demonstrated that reversibility is not complete, and a two-component model was proposed to interpret incompletely reversible behavior. Yin et al. (3) investigated mercury (Hg(II)) desorption from 15 soils and reported that sorption/desorption hysteresis was evident. Similar incomplete reversibility has also been reported by others (4, 5). Verburg and Baveye (6) reviewed the hysteresis previously observed (8-10) in the binary * Corresponding author phone: (713)348-5224; fax: (713)348-5203; e-mail [email protected]. 5566

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 24, 2003

exchange of cations on 2:1 clay minerals and examined five mechanisms used to explain the hysteresis. They indicated that none of them alone could account fully for the cationexchange hysteresis. In contrast, Campbell et al. (10) conducted cadmium sorption/desorption experiments on a montmorillonite-humic acid mixture and showed that most of the bound cadmium eluted from the columns. Evidence has been reported for complete reversibility of cadmium sorption on illite, kaolinite, and an oxisol (7, 11, 12). Padmanabham (8) reported that desorption of copper from goethite showed considerable hysteresis whereas all the sorbed lead was readily desorbable. Likewise, Ainsworth et al. (9) indicated that sorption/desorption hysteresis was obvious for cadmium and cobalt on hydrous ferric oxide but that little hysteresis was observed for lead. Clearly, our understanding of the heavy metal sorption/desorption mechanism is limited, and further investigations of the reversibility of heavy metal sorption is needed. In the past decade, extended X-ray absorption fine structure (EXAFS) has been employed to explore heavy metal sorption mechanisms (13-15). Scheidegger et al. (16), via EXAFS analysis, reported that nickel forms mixed Ni-Al polynuclear hydroxides at the surface of clay minerals and metal oxides. Similarly, Ford and Sparks (13) observed the formation of Zn-Al layered double hydroxide (LDH) on the pyrophyllite surface at pH >7, and Voegelin et al. (15) showed evidence of the formation of Zn-Al LDH at pH 6.5 and low surface coverage. Likewise, Roe et al. (17) and ChisholmBrause et al. (18) reported that Pb formed inner-sphere complexes at the surface of goethite and alumina when surface coverage is low but that the formation of Pb surface polymers was observed when surface coverage is high. In contrast, cadmium appears to be sorbed onto exchangeable sites in soils, and no Cd-Al LDH was observed using EXAFs (15). Clearly, EXAFS has become an important technique in understanding the sorption and desorption mechanisms of heavy metals in soils, but each metal appears to behave differently and must be investigated separately. Complexation of heavy metals with dissolved organic ligands (DOLs) has been found to significantly affect metal uptake on soils and sediments. Complexation reactions generally include metal-ligand ion pairs, soluble metalligand complexes, and chelation. Both simple and complex ligands have been found to be important on heavy metal transportation in soils and sediments (19). One of the classical methods to initiate heavy metal desorption is to replace the supernatant liquid with metal-free electrolyte solution. During this process, colloids and organic and inorganic ligands in the sediments are partially removed, which may interfere with the interpretation of sorption/desorption hysteresis. For instance, Gschwend and Wu (20) reported that, if nonsettling particles are eliminated by washing the sediments, the Kd values for either sorption or desorption reaction are indistinguishable. Likewise, Neal and Sposito (21) found that the sorption isotherms for cadmium with sludge soils was S-shaped but that they became more typical L-shaped after washing the soils. To evaluate the relationship between prewashing treatment and sorption/desorption hysteresis, sorption and desorption experiments in this study will be conducted on both untreated and prewashed sediments. The remobilization of heavy metals from natural sediments is often triggered by (i) acidification, (ii) decrease in the metal concentration in the solution phase, (iii) complexation by soluble chelating agents such as EDTA, and (iv) redox condition changes in sediments. In this study, the first 10.1021/es034392w CCC: $25.00

 2003 American Chemical Society Published on Web 10/28/2003

three processes will be simulated. Desorption experiments will be induced by (i) replacing the supernatant liquid with contaminant-free electrolyte solution, (ii) lowering the solution pH with mineral acid, and (iii) sequestration with EDTA. Therefore, the objective of this study is to investigate the release of heavy metals in response to sediment chemical changes, specifically, the effect of pH, solution metal concentration, chelating agents, and prewashing treatment on heavy metal sorption/desorption hysteresis.

Materials and Experimental Methods Sorbent. Utica sediment, the bottom sediment collected from Utica Harbor, NY, near a gas manufacturing plant, was used as the sorbent in this study. The organic carbon content of the Utica sediment is 2.75%, and the specific surface area was determined by the single-point N2-BET method to be 3.99 m2/g (Micromeritics, Inc., Norcross, GA). The sediment was air-dried, homogenized with mortar and pestle, and then sieved through cheesecloth to remove vegetative matter and pebbles. The total metal concentration in the sediment was determined by ICP after acid digestion of the sediment with nitric acid reflux (1:1) and hydrogen peroxide oxidation (30%) following the EPA SW-846 Method 3050. The primary metal ions in the acid extract of the sediment were 0.79% Ca, 2.3% Fe, 0.02% Mn, and 0.38% Mg (by weight of sediments). The sediment also contained 0.06 µmol/g Cd, 0.32 µmol/g Pb, 4.9 µmol/g Cu, 2.7 µmol/g Zn, and 2.3 µmol/g Cr as the predominant trace elements. In this research, only the desorption of Cd and Pb was studied. In the following presentation, most data treatment does not consider the background Cd and Pb concentrations in the sediment since the freshly added metal concentration is typically much higher than background concentration. The typical sorbed concentrations were 0.5-5 µmol/g for Cd and 5-30 µmol/g for Pb. Sorbates and Solutions. Cadmium and lead solutions were made from Cd(NO3)2‚4H2O and Pb(NO3)2 powders. All the sorption/desorption experiments were carried out in 0.01 M NaNO3 and 0.01 M NaN3 (as a bacterial inhibitor). Reagentgrade or better chemicals were used to make all solutions. Water used in this research was prepared by passing deionized water (Continental Water Co., Bedford, MA) through a Barnstead Ultrapure Mixed Bed Cartridge (Barnstead Co., Boston, MA) to remove silica and CO2. The water was further purified with an Amberlite XAD-2 resin (Rohm & Hass Co., Philadelphia, PA) column to remove trace organic materials. Sorption and Desorption Experiments. Studies of the sorption/desorption of cadmium and lead on the Utica sediments were carried out in batch experiments. The Cd and Pb solution (40 mL) at various concentration levels (5500 µM) was transferred to 50-mL plastic tubes, and 1 g of dry sediment was added. For sorption edge experiments, total initial metal concentrations of 0.1 and 0.36 mM Cd and Pb were used. The solution pH was adjusted to a desired value by adding trace amounts of either 0.95 N HNO3 or 1.5 N NaOH solution. For some experiments conducted at pH 6-7, 5 mM sodium piperazine-N,N′-bis(2-ethanesulfonate) (PIPES) was added to the solution as a pH buffer. The sediment-water mixtures were then sealed and placed on a slowly rotating rack that provided gentle end-over-end mixing (40 rpm) during the reaction period. At the end of each experiment, reaction tubes were centrifuged at 6000 rpm for 50 min, and the supernatant solutions were withdrawn with a syringe and filtered through 0.2-µm Nalgene syringe filters (SFCA). The pH of each solution was measured immediately after sampling. An Orion-Ross combination glass electrode and an expandable ion analyzer EA 920 (Orion Research) were used for the pH measurements. The glass electrode was calibrated at 25 °C using buffers at pH 4.0, 7.0,

and 10.0. The filtered samples were then acidified with 1% nitric acid and analyzed for solution metal concentrations. Our preliminary kinetics studies have shown that Cd and Pb sorption on Utica sediments occurs rapidly and that more than 95% of the reaction was completed within the first 24 h (data not shown). The duration of each sorption experiment was generally 2-4 d. Several sorption experiments were extended to 30 d to study the effect of aging on the reversibility of desorption. Desorption experiments were initiated in three different ways: (1) by lowering the solution pH with acid, (2) by replacing the supernatant liquid with contaminant-free electrolyte solution in order to lower the solution-phase adsorbate concentration, and (3) by sequestration with a solution containing 0.2 mM sodium ethylenediaminetetraacetate (EDTA) in background electrolyte solution at pH 7-8. Similar kinetics studies indicated that most Cd and Pb (95%) desorb within the first 24 h (data not shown). The duration of each desorption experiment was generally 2-5 d. Desorption by lowering pH (method 1) was accomplished by first preparing several sample vials contaminated with a Cd (0.1 mM) or Pb (0.36 mM) solution, and then allowing these sample vials to equilibrate at pH 7.27 and 6.41 for the Cd and Pb sorption experiments, respectively. After 2-4 d, various amounts of acid were added to these sample vials to lower the solution pH, allowing the solution to re-equilibrate for several days before analyzing for the solution-phase metal concentrations. For desorption method 2, desorption was initiated by replacing ∼90% of the supernatant solution with clean electrolyte solution and allowing desorption to re-equilibrate over a period of days. After the first desorption experiment, the sample was centrifuged, and heavy metal concentration in the supernatant solution was analyzed. The procedure was repeated by replacing the supernatant solution with fresh electrolyte solution to the sample vial, initiating another cycle of desorption. For the EDTA desorption experiments (method 3), first, Cd and Pb solutions at various concentrations were added to sample vials containing 1 g of Utica sediments at pH 7 for the sorption experiments. After 3 and 30 d sorption time, successive desorption was induced by replacing the supernatant solution with 0.2 mM EDTA solution at pH 7-8. Each EDTA desorption step was allowed to equilibrate for 1-5 d, and the metal concentration in the aqueous phase was analyzed. Successive EDTA desorption was repeated 4-5 times by replacing the supernatant solution with heavy metalfree EDTA solution, similar to method 2. A Utica sediment sample, without the addition of Cd or Pb, was also extracted by EDTA, and the extractability of these “native” metals from Utica sediments by EDTA was also compared. Sorption and desorption of cadmium and lead using method 2 were also run on prewashed Utica sediments. For the prewashing treatment, the sediment was dispersed in a clean electrolyte solution for a short time, and then the sediment was separated from the supernatant solution by centrifugation. For the second and subsequent prewashing treatments, the supernatant solution was replaced with clean electrolyte solution, and the above two procedures were repeated. The sediment was washed 1-5 times in the background electrolyte solution before the initiation of sorption and desorption experiments. Both sorption edge and desorption experiments with method 2 were conducted on sediments prewashed 1 and 3 times. Sorption time was 2-3 d, and desorption time was 3-4 d. Extended desorption time of 110 d on two samples was also tested to determine desorption reversibility. Solution metal concentrations were measured on an ICPAES Plasma 400 and an Elan 9000 ICP-MS (Perkin-Elmer Co.). The detection limits of Elan 9000 for Cd and Pb are 0.02 VOL. 37, NO. 24, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5567

FIGURE 1. Plots of percent sorption of (a) Cd and (b) Pb on unwashed Utica sediments vs equilibrium solution pH values where the filled circles are sorption edge data and open triangles and squares are desorption edge data. and 0.008 µg/L, respectively, and the relative standard deviation (RSD) of three replicate analyses was generally below 5%. Solid-phase metal concentrations were calculated from mass balance. Several sediment samples were extracted with 1 N HCl for 16 h after sorption/desorption experiments to determine the residual metal concentrations (22). For selected samples, the dissolved organic carbon content of the solution was analyzed. Dissolved organic carbon (DOC) concentrations were measured using a total organic carbon analyzer TOC-5050A (Shimadzu Corporation) following acidification to pH 2 by HCl and stripping of the inorganic carbon with N2 gas. The RSD of five experiments was generally below 2%.

Results and Discussion Sorption Edges and Desorption Initiated by Lowering the Solution pH. In Figure 1, the fractions of Cd and Pb sorption on Utica sediments versus pH are plotted. The sorption followed a trend of Pb > Cd. Greater sorbed fractions occur for both metals at higher pH values. The fractions of Cd and Pb sorbed at pH 5 are ∼38 and 96% and increase to 91 and 100% at pH 7, respectively, indicating a strong pH effect. These sorption edge data are consistent with numerous previous findings (23-25). Christensen (24) reported an increase of 1.77-fold of the Cd distribution coefficient (Kd ) qads/C at 0.002-0.026 µmol/L) for each 0.5-unit increase in pH for 63 Danish agricultural soils. A Freundlich isotherm (that will be discussed below) with parameters fitted from our data would predict a change of 1.69- and 1.50-fold increase in Kd for Cd and Pb, respectively, per 0.5-unit increase in pH at similar concentrations. The desorption edge (i.e., the percent of Cd and Pb left on the solid phase after desorption by lowering the solution

FIGURE 2. Plots of (a) Cd and (b) Pb sorption (symbols) on unwashed Utica sediments at four different pH values. The lines are the Freundlich isotherms calculated using eq 1 with the parameters given in Table 1. pH) are also shown in Figure 1. Duplicate desorption experiments for Cd are plotted, and excellent reproducibility is observed. As shown in Figure 1, the desorption edges of Cd and Pb are almost identical to their respective sorption edges, suggesting that their sorption on Utica sediments is completely reversible when desorption is initiated by lowering solution pH. It is important to note that the complete reversibility of sorption and desorption of Cd and Pb by reducing pH is only restricted to freshly sorbed metals (