Role of Heterogeneous Precipitation in Determining the Nature of

Aug 6, 2010 - A detailed kinetic model has been developed to describe the formation of the oxidation products, organically complexed Fe(III) and amorp...
0 downloads 9 Views 607KB Size
Environ. Sci. Technol. 2010, 44, 6667–6673

Role of Heterogeneous Precipitation in Determining the Nature of Products Formed on Oxidation of Fe(II) in Seawater Containing Natural Organic Matter MARK W. BLIGH AND T. DAVID WAITE* School of Civil and Environmental Engineering, The University of New South Wales, Sydney, NSW 2052, Australia

Received April 2, 2010. Revised manuscript received July 9, 2010. Accepted July 26, 2010.

A detailed kinetic model has been developed to describe the formation of the oxidation products, organically complexed Fe(III) and amorphous ferric oxide (AFO), on oxidation of Fe(II) in seawater containing Suwannee River fulvic acid (SRFA). Experimental data were collected using spectrophotometric detection of the Fe(III)-SRFA complex for a range of initial concentrations of Fe(II) and SRFA. Initial sensitivity analysis identified rate constants to which the model was most sensitive including those for heterogeneous precipitation of AFO and Fe(II)-SRFA formation and dissociation which to date have only been determined with a high degree of uncertainty. Using these rate constants as fitting parameters, an accurate fit to the experimental data could be obtained using a kinetic model describing key processes. However, reasonable fits could only be achieved with the inclusion of the heterogeneous precipitation reaction suggesting the importance of this reaction in determining the outcome of oxidation in the presence of organic ligands. The rate constants for Fe(II)-SRFA formation and dissociation were highly correlated and could not be determined uniquely, however their ratio revealed a stability constant of ∼105, 3 orders of magnitude higher than previously reported. The fitted model also suggested that a complex interaction between Fe(II) and SRFA in the initial stages of the oxidation process determines the pathway of Fe(III)-SRFA formation.

mediated reductive dissolution of benthic sediments or as a result of groundwater discharge. The presence of NOM prevents a portion of the incoming inorganic Fe(II) (Fe(II)’) transforming directly to an amorphous ferric oxide (AFO) following oxidation (6), thereby enhancing the levels of dissolved iron and, potentially, its bioavailability. The relative proportions of the products (i.e., Fe(III)-NOM complex and AFO) resulting from the oxidation of Fe(II)’ in seawater in the presence of NOM will depend on the concentrations of Fe(II)’ and NOM present and on the iron binding characteristics of the NOM. While knowledge of the relevant reactions and their rate constants is available, an integrated understanding is lacking. A number of the various competing reactions and transformation pathways, which are illustrated in Figure 1, occur at similar time scales with simultaneous analysis of the complete reaction set required, at least in the first instance, to fully understand the dynamics of reactant loss and product formation. Previous studies in which the oxidation of Fe(II)’ in the presence of NOM has been investigated have generally used a ligand concentration sufficient to prevent the precipitation of AFO or have not considered the outcomes of the oxidation process (7, 8). Rather, the rate of oxidation has been the key interest of such studies. Pullin and Cabaniss (9) measured the products resulting from the oxidation of Fe(II)’ in the presence of SRFA in pH 8 buffer, however these authors were chiefly interested in the effect of the presence or absence of SRFA on the rate of oxidation. Autocatalysis of oxidation by AFO (10, 11) was also investigated by these authors (9), however the high carbonate concentration of seawater and the low [Fe(II)’] that we are interested in render this mechanism insignificant for this study (9, 11, 12). As shown in Figure 1, precipitation of AFO may occur either via a homogeneous pathway where two dissolved inorganic Fe(III) species (Fe(III)’) interact to form AFO or via a heterogeneous pathway where Fe(III)’ reacts with AFO to form more AFO. Given the extremely low concentrations of Fe(III)’ in seawater, the heterogeneous pathway for AFO formation might be expected to play a significant role in the formation of oxidation products however this reaction is

Introduction Iron is an important nutrient required for biological production which has been implicated in both the limitation of productivity (1) and the promotion of excessive growth of some cyanobacterial species (2). Due to its extremely low solubility at pH ∼ 8, the bioavailability of iron is closely linked to the presence of organic molecules capable of binding and maintaining ferric iron (Fe(III)) in the dissolved phase (3, 4). Estuarine systems are supplied with large fluxes of terrigenous natural organic matter (NOM), including fulvic acids, from the surrounding catchment with these organic complexing agents playing an important role in maintaining levels of dissolved iron higher than those of oceanic waters (5, 6). Iron often enters an estuarine water column in the reduced ferrous form (Fe(II)) either as a result of microbially * Corresponding author phone: +61 2 9385 5059; fax: +61 2 9313 834; e-mail: [email protected]. 10.1021/es101046y

 2010 American Chemical Society

Published on Web 08/06/2010

FIGURE 1. Key pathways of iron transformation during the oxidation of Fe(II)’ in the presence of SRFA (L). Solid lines show the forward reactions leading to the oxidation products AFO and Fe(III)L. Dashed lines show the substantially slower back reactions. The percentages of the initial Fe(II)’ that follow each forward pathway are shown for the standard model with initial [Fe(II)’] ) 1 µM and [L] ) 4 mg L-1. The relevant reaction numbers, from Table 2, are shown in parentheses after the pathway name. The thick gray lines show the two pathways for the transformation of Fe(II)’ to Fe(III)L: (1) Associative oxidation (AO) and (2) nonassociative oxidation (NAO). VOL. 44, NO. 17, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6667

rarely included in kinetic models that include AFO precipitation (8, 13). This reaction was included in recent detailed investigations of iron oxyhydroxide precipitation (14, 15) and was also incorporated into a model for iron cycling in coastal surface waters (16) however there was no analysis of the significance (or otherwise) of this reaction. Measurement of the products formed on oxidation of Fe(II)’ is hampered by the difficulty in differentiating between freshly formed AFO, composed of small and highly reactive colloids, and organically complexed Fe(III). Various techniques have been utilized previously, including susceptibility to reduction and ferrozine complexation in a flow injection system (17) and capture by anion exchange resin following acidification (18). Here a spectrophotometric technique is used to determine the concentration of Fe(III)-NOM present throughout the Fe(II) oxidation process. In this study we develop a kinetic model, based on the reactions of Figure 1, to be fitted to measurements of the formation of Fe(III)-NOM undertaken for a range of initial Fe(II)’ and NOM concentrations. We aim to utilize reaction rate constants available from previous studies as a first pass of the model and adjust those rate constants displaying the greatest sensitivity and uncertainty. Of particular interest is the assessment of whether the heterogeneous precipitation of AFO is a significant determinant of the outcome of oxidation of Fe(II)’.

Materials and Methods Reagents. All solutions were prepared using 18Ω Milli-Q water. Clean seawater (pH 8.1 and salinity 36 ppt) was obtained from the Sydney Offshore Reference Station, vacuum filtered using Millipore 0.22 µm membrane filters, and stored in the dark at 4 °C when not in use. For experimental use, seawater was equilibrated with the atmosphere. All pH measurements were made using a calibrated Hanna HI9025 pH meter. All pH adjustments were performed using high purity 30% w/v HCl and 32% w/v NaOH (Fluka puriss p.a plus). The former reagent was diluted to produce a solution of high purity 0.2 M HCl for reagent preparation. A 10 mM Fe(II) in 2 mM HCl stock was prepared by dissolving ferrous ammonium sulfate hexahydrate (Fe(NH4SO4)2.6H2O) (Sigma-Aldrich ACS reagent) in 0.2 M HCl and making up to volume with Milli-Q water. This stock was prepared three-monthly with the pH appropriate to avoid oxidation of Fe(II). A daily working stock of 100 µM Fe(II) was prepared by diluting the Fe(II) stock in 0.4 mM HCl. An 8 mM Fe(III) stock in 8 mM HCl was prepared by dissolving ferric ammonium sulfate dodecahydrate (FeNH4(SO4)2 · 12H2O) (Sigma-Aldrich ACS reagent) in 0.2 M HCl and making up to volume with Milli-Q water. This stock was prepared weekly with pH appropriate to prevent polymerization. Daily working stocks of 1 mM and 100 µM were prepared by diluting the Fe(III) stock in 2 mM HCl. A 2 g L-1 solution of Suwannee River fulvic acid (SRFA) (International Humic Substances Society) was prepared in Milli-Q water. All stocks were stored in the dark at 4 °C when not in use. Spectrophotometric Determination of Fe(III)-SRFA Complex. The concentration of the Fe(III)-SRFA complex was determined by measuring light absorbance at 500 nm, following zeroing with a solution of seawater containing the same concentration of SRFA, using long path length spectrophotometry. Sample was driven with a peristaltic pump through a 1 m liquid capillary waveguide (World Precision Instruments) incorporating a tungsten halogen light source (Ocean Optics LS-1) with a variable neutral density filter and a miniature fiber optic spectrometer (Ocean Optics USB4000). To obtain a stable absorbance reading, the absorbance at 500 nm was baseline-corrected using the absorbance at 700 nm as a reference. During measurement, the sample was 6668

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 17, 2010

maintained at 25 °C in a water jacketed beaker connected to a constant temperature water bath. Amber glass vials were used to contain samples in order to minimize light effects. A calibration curve was established by measuring the absorbance for various [Fe(III)-SRFA] produced by adding Fe(II) to seawater containing 32 mg L-1 of SRFA in duplicate, resulting in an extinction coefficient of 500 ) 1810 M-1cm-1 (Supporting Information (SI) Figure S1). A maximum value for absorbance due to light scattering from particles of AFO was established by adding 1 µM of Fe(II) to seawater without SRFA and was calculated to be equivalent to 0.025 µM Fe(III)-SRFA (SI Figure S2), providing minimal impact on the overall absorbance. Determination of Fe(III)-SRFA Formation Rate Constant. The apparent rate constant for Fe(III)-SRFA formation has been previously determined in seawater using a competitive ligand method with sulfosalicylic acid as the competing ligand (6). This technique enables direct determination of the apparent rate constant kf and therefore provides a relatively high degree of certainty in the resultant value. In order to assess the applicability of this rate constant to the scenario of competition for Fe(III)’ with AFO precipitation reactions, competitive ligand experiments were undertaken where SRFA competes with AFO precipitation for Fe(III)’ following the addition of Fe(III)’ to seawater containing SRFA. These studies were similar to those undertaken by Rose and Waite (15) except that, in the studies described here, the rate constant kp (4.1 × 107 M-1s-1 (15)) for the AFO precipitation reaction (which was the unknown in (12)) will be the known rate constant against which the unknown rate constant for complex formation kf may be determined using the relationship shown in eq 1. (See SI Section 2 for the derivation of eq 1). kp[FeT] [FeT] ) +1 [Fe(III)SRFA] kf[SRFA]

(1)

Two series of experiments were undertaken: (1) [Fe(III)] ) 1 µM and [SRFA] ) 2, 4, 8, 16, or 32 mg L-1 and (2) [SRFA] ) 4 mg L-1 and [Fe(III)] ) 50, 100, 250, 500 nM, 1, or 2 µM. [FeT]/[Fe(III)-SRFA] was plotted against either 1/[SRFA] or [FeT] and kf calculated from the slope of the linear regression. In order to maintain consistency with recent previous studies (6, 19), a weight of 3800 g per mole of binding sites for SRFA is assumed. (Further discussion of the approach used here is provided in SI Section 3). Solutions of seawater containing SRFA at the various concentrations required were prepared 24 h prior to experimentation in order to ensure equilibration. A small degree of pH adjustment was required for higher concentrations. An aliquot of the required volume of the appropriate Fe(III) working stock was pipetted into 10 mL of a rapidly stirred seawater-SRFA mixture at 25 °C. Following 30 s of mixing, sample was drawn into the long path length spectrometer and, after zeroing, the signal was monitored for a few minutes. Experiments were undertaken in duplicate for each combination of [Fe(III)] and [SRFA]. The low concentrations of added iron and the buffering capacity of the seawater ensured that no change of pH occurred during the experiment. Oxidation Experiments. The oxidation of Fe(II)’ in the presence of organics is typically monitored by measuring the decrease in [Fe(II)] as a function of time. The purpose here was not to describe the impact of NOM on the rate of oxidation, as this has been previously investigated (7-9, 20, 21) but, rather, to ascertain the effect of NOM on the composition of the oxidation products at the completion of oxidation. In addition to the difficulty in discerning between Fe(II)-SRFA and Fe(III)-SRFA complexes, it is not possible to follow the very rapid formation of oxidation products in seawater, particularly as SRFA promotes oxidation (8), since at least a

minute is required to establish the full signal in the long path length spectrometer used here. Solutions of seawater containing SRFA were prepared in the manner previously described. The required volume of Fe(II) working stock was pipetted into 20 mL of rapidly stirred seawater-SRFA mixture. After 30 s of mixing, sample was drawn into the long path length spectrometer and, after zeroing, the signal was monitored for 15 min or until a stable reading was achieved. Two series of experiments were undertaken: (1) [Fe(II)] ) 1 µM and [SRFA] ) 1, 2, 4, 8, or 16, mgL-1 and (2) [SRFA] ) 4 mg L-1 and [Fe(II)] ) 50, 100, 200, 500, 750, or 1000 nM. Experiments were undertaken in duplicate for each combination of [Fe(III)] and [SRFA]. Modeling. The software program Presto (22) was used to fit the kinetic model to the experimental data. The number of fitting parameters, the correlations between some parameters, and lack of constraint on some parameters necessitated the use of a box search routine that facilitated the identification of the region of the parameter space containing minima, by calculating the residual for each specified step in the parameter values. A simulated annealing procedure was used to establish starting values for the parameter estimation procedure by searching the parameter space identified in the box search using a stochastic algorithm that jumps within the defined parameter space, calculating the residual at each step and testing whether this new point leads to a lower residual. The size of the steps is decreased depending on the progress made in the residual until the parameter is frozen (annealed). Local minima are able to be tested before the procedure is restarted. Sensitivity analysis of the kinetic model was undertaken using the software program Kintecus (23).

Results and Discussion ValidityofSpectrophotometricMeasurementofFe(III)-SRFA. A typical plot of absorbance following the addition of 1 µM Fe(II) to seawater containing 4 mg L-1 SRFA is shown in SI Figure S3. The absorbance of the Fe(III)-SRFA complex (above the background of SRFA absorbance) is due to ligand to metal charge transfer (LMCT) (21) however the lack of well-defined LMCT bands (SI Figure S4) is due to the high variability of the coordination environments of the Fe(III) bound to the complex, ill-defined molecules that comprise SRFA. A small number of studies have previously used absorbance at 500 nm as a measure of the concentration of Fe(III)-NOM complexes (19, 24, 25). Further support for the validity of this method can be deduced from the results of the Fe(III)-SRFA formation experiments where the calculated rate constant kf is very close to that previously determined using a different method (see below). The calibration curve of SI Figure S1, while remaining within the error range, is marginally below the four points of lower [Fe(III)-SRFA]. For [Fe(III)-SRFA] e 0.05 × 10-6 M this represents a potential error in measurement of up to 100%. However this low concentration range is only encountered in one measurement. Determination of kf. Separate values of the apparent rate constant for Fe(III)-SRFA formation kf were calculated from the plots of [FeT]/[Fe(III)-SRFA] against 1/[SRFA] and against [FeT] (Figure 2) using eq 1. Both methods produce values of kf that are very comparable to the previously established value in seawater (6) (Table 1). The experimental series where [SRFA] is constant and [FeT] varies (Figure 2b) resulted in a value for kf with low error that is almost identical to the established value. The greater error of the series in which [FeT] is constant and [SRFA] varies resulted from the curvilinearity of the plot (Figure 2a) which may reflect a deviance from unity in the order relative to [SRFA] of the complex formation reaction. From eq 2, n > 1 (with m ) 1)

FIGURE 2. Plots used to calculate kf from the outcome of the addition of Fe(III) to seawater containing SRFA. (a) Addition of 1 µM Fe(III) into various [SRFA]. (b) Addition of various [Fe(III)] to 4 mg L-1 SRFA. The slope equals (a) kp[FeT]/kf and (b) kp/ kf[SRFA] from which two values for kf were calculated. Bars show 95% confidence intervals for duplicates. would result in the observed convex shape of the plot. In reality there are likely to be a diversity of complex species with both m and n, on average, deviating from unity. mFe(III)′+nSRFA f Fe(III)mSRFAn

(2)

The mixture of different binding sites that comprise SRFA may also contribute to the curvilinearity of the plot. While there may be some dependence of kf on [SRFA], there is no feasible alternative, based on data from this or previous studies, to describing Fe(III)-SRFA as a 1:1 complex. The 95% confidence limits of slope of the plot of Figure 2a result in a range for kf of 5.8 × 106 to 12 × 106 M-1s-1. The possible errors involved should not undermine the essential conclusions drawn from this study. Oxidation Results and Modeling. The concentration of the Fe(III)-SRFA complex measured 15 min after the addition of Fe(II) to seawater containing SRFA increases, as expected, with greater [FeT] and [SRFA] (Figure 3). Only with [SRFA] ) 4 mg L-1 and [FeT] ) 50 and 100 nM was Fe(III) fully complexed with no production of AFO. The oxidation products of the remaining combinations of [FeT] and [SRFA] were comprised of various mixtures of Fe(III)-SRFA complex and AFO. A kinetic model (Table 2) was constructed describing the reactions illustrated in Figure 1. The oxidation reactions are represented as parallel Haber-Weiss mechanisms for inorganic and organically complexed Fe(II) (26) with additional reactions to account for catalyzed and uncatalysed dismutation of superoxide (8). The reactions comprising the model are a subset of the reactions used in an iron redox model previously used to describe Fe speciation in coastal waters (16). The model used here treats SRFA as a single reactant and operates at the level of binding site rather than SRFA molecule (see SI Section 3 for discussion). As a naturally derived compound, SRFA is structurally heterogeneous and binding occurs via different functional groups such as VOL. 44, NO. 17, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6669

TABLE 1. Fe(III)-SRFA Formation Rate Constants kf from This and Other Studies

a

kf ( × 106 M-1s-1)

medium

reference

7.9 ( 0.8 6.1 ( 0.3 6.0 0.9 0.8

seawater seawater seawater 100 mM Hepes pH 8 10 mM Hepes pH 8

this study ([FeT] ) 1 µM) this study ([SRFA] ) 4 mg L-1) 6 19 40

a

Standard error of slope calculation from regression analysis.

TABLE 2. Standard Kinetic Model no.

reaction

k (M-1s-1)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

Fe(II)’ + O2 f Fe(III)’ + O2Fe(II)’ + O2- f Fe(III)’ + H2O2 Fe(II)‘ + H2O2 f Fe(III)’ + OH + OHFe(II)‘ + OH f Fe(III)’ + OHFe(III)’ + O2- f Fe(II) + O2 Fe(III)’ + Fe(III)’ f 2AFO Fe(III) + AFO f 2AFO AFO f Fe(III)’ O2- + O2- f H2O2 CuII + O2- f CuI + O2 CuI + O2- f CuII + H2O Fe(II)’ + L f Fe(II)L Fe(II)L f Fe(II)’ + L Fe(III)’ + L f Fe(III)L Fe(III)L f Fe(III)’ + L Fe(II)L + O2 f Fe(III)L + O2Fe(II)L + O2- f Fe(III)L + H2O2 Fe(II)L + H2O2 f Fe(III)L + OH + OHFe(II)L + OH f Fe(III)L + OH-

13 a 1 × 107 a 3.1 × 104 a 5 × 108 a 1.5 × 108 a 4.1 × 107 b 3.8 × 107 c 1 × 10-5 b,f 1 × 105 a 6.6 × 108 a 2 × 109 a 4.5 × 104 c 4.5 × 10-1 c,f 1.3 × 106 d 3.2 × 10-4 e,f 1 × 102 a 1 × 107 a 5 × 103 a 5 × 108 a

a From ref 8. b From ref 15. c Fitted. d Reduced from established value (6) in order to enable fitting within constraints. e Dissociation rate constant for single ligand class - order of magnitude above and below weak and strong class rate constants (19). f First order reaction (s-1).

FIGURE 3. Comparison of the observed [Fe(III)-SRFA], at 15 min following the addition of Fe(II) to seawater containing SRFA, with predictions of the model fitted to both sets of data. (a) Addition of 1 µM Fe(II) into various [SRFA]. (b). Addition of various [Fe(II)] into 4 mg L-1 SRFA. Also shown in (a) is the model output when reaction 7 is omitted such that AFO precipitation proceeds only via reaction 6. Bars show 95% confidence intervals for duplicates.

phenolic and carboxylic groups (27). Two-ligand models, with weak and strong binding sites, have been used with some success to describe complex dissociation (6, 18, 19). However, the rate of complex formation is thought to be controlled by the rate of water loss of the metal ion of the outer-sphere complex (28), providing support for the use of a single rate constant (18). Since the complex dissociation reactions, especially for Fe(III)-SRFA, are slower and play a lesser role in determining the outcome of oxidation, we have adopted a single ligand model and avoided the additional complexity and assumptions of a multiligand model. Initially, values were assigned to rate constants using the current best available information from the literature (SI Table S1) and this model was subjected to sensitivity analysis in order to determine which rate constants deserved more careful scrutiny with regards to the uncertainty and applicability of the previously established value (see SI). This analysis (SI Figure S5) revealed that the outcome of the model was particularly sensitive to the values assigned to the rate constants k7, k12, and k14. Of these values only k14 has been measured directly both in a previous study (6) and in this study. 6670

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 17, 2010

The previously established values for k6 and k7 were not able to be measured separately in a competitive ligand experiment and were assigned the same value of 4.1 × 107 M-1s-1 (15). However, since precipitation is predominantly homogeneous under these experimental conditions, this value can be used confidently for k6 while only representing an upper limit for k7. The greater coordination of Fe centers in AFO and the reduced opportunities of collision suggests k7 < k6 however there is no information available to support an alternate value of k7 at this stage. The only published value for k12, 2.5 × 104 M-1s-1, was measured indirectly via modeling of the loss of Fe(II)’ in the presence of SRFA in seawater under anoxic conditions (6). This approach was hampered by the need to define weak and strong binding groups within SRFA and the dependency of the formation rate constants on concentrations assigned to these different groups and on the correlations between the rate constants for Fe(II)-SRFA complex formation and dissociation (k12 and k13), thereby reducing the certainty attached to this value for k12. Due to the combination of high uncertainty and high model sensitivity attached to k7 and k12, these rate constants were used as fitting parameters in the model (Table 2). k13 was also included as a fitting parameter since, while the initial sensitivity analysis resulted in a low NSC value for k13, its likely correlation with k12, together with the uncertainty associated with k12, points to the requirement that it be fitted. A preliminary box search, with k7, k12, and k13 as variable parameters, showed that the rate constant k14 for Fe(III)-SRFA formation, previously established at 6 × 106 M-1s-1 (6) and confirmed in this study, was too large to allow an acceptable

fit (sum of squares residual ∼1 s-1 will not be accounted for, resulting in a lower fitted rate constant than might be calculated using a method that is sensitive to fast dissociation reactions. The modeling of NOM complex dissociation is fraught with problems, especially in the presence of high concentrations of divalent cations (33), due to the possession by SRFA of different ligand classes. The inclusion of heterogeneous precipitation (reaction 7) was necessary in order to achieve good fits of the model to the data using realistic rate constants that fall within the constraints based on existing knowledge. The important role played by this reaction in the fitted model is illustrated in Figure 3a where the model with k7 ) 0 predicts the formation of, on average, twice the concentration of Fe(III)-SRFA actually observed. The difficulties encountered in accounting for experimental observations with a model incorporating only the homogeneous reaction are revealed on fitting such a model to the data. Maintaining k14 ) 1.3 × 106 M-1s-1 and the constraint on k12, and fitting with no constraint on k13, results in a best fit with k13 ) 6.9 M-1 s-1 and a high sum of squares residual (0.82 compared to 0.10 for original fit). This unrealistically high value of k13 is necessary in order to minimize the production of Fe(III)-SRFA via oxidation of Fe(II)-SRFA since the homogeneous precipitation of AFO (reaction 6 in Table 2) competes poorly with the complexation of Fe(III)’ by SRFA (reaction 14). Implications for Natural Systems. The success in fitting the model to the range of experimental conditions in the studies described here provides validation of the modeling approach that may be used in the simulation of natural systems. When Fe(II) concentrations are in the micromolar range, reasonably high levels of NOM are required to maintain a significant proportion of the iron in the dissolved phase following Fe(II) oxidation (Figure 4). Dissolved organic carbon (DOC, ∼50% of NOM) of coastal waters is typically 30 mg L-1 (37). Due to its autocatalytic nature, the heterogeneous precipitation reaction has the potential to compete strongly with complex formation for Fe(III)’ and we speculate that this process plays an important role during rapid oxidation where AFO is, at least initially, present as nanosize particles (38). VOL. 44, NO. 17, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6671

AFO from a reactive form to a nonreactive form was included in the model, with a transformation rate constant previously found to be applicable under these conditions, half of the AFO was found to be nonreactive after 40 min (39). In addition, the rate of ligand-promoted dissolution of AFO formed from the addition of Fe(III) to seawater halved after 2 h (15). As a result, while this model is applicable to the oxidation of a pulse of Fe(II), oxidation of a continuous flux of Fe(II) over time may be less well predicted since aging of the resultant AFO will reduce the importance of the heterogeneous precipitation pathway. The effectiveness of a model incorporating an aging reaction in prediction of the outcome of such a scenario is yet to be established.

Acknowledgments Funding provided by the Australian Research Council and the Moreton Bay Waterways and Catchment Partnership through ARC-Linkage project LP0561150 is gratefully acknowledged.

Supporting Information Available Additional details available. This material is available free of charge via the Internet at http://pubs.acs.org. FIGURE 5. Interaction of initial [Fe(II)] and [SRFA] in determining the relative contributions of various pathways in the formation of oxidation products (Fe(III)-SRFA and AFO) following the oxidation of Fe(II) in seawater containing SRFA. (a) The fraction of Fe(III)-SRFA that has been form via associative oxidation (AO) (see Figure 1) as function of [Fe(II)] for various [SRFA]. (b) Fraction of initial [Fe(II)] that has been complexed (reaction 12) as function of [Fe(II)] for various [SRFA]. This fraction is effectively a net value determined by the balance between dissociation and oxidation of Fe(II)-SRFA. According to our model, competition between complex formation and heterogeneous precipitation is avoided when Fe(II) is complexed prior to oxidation. The best-fit model developed here has been used to examine the pathways of formation of oxidation products for a realistic range of [Fe(II)] and [SRFA] by labeling Fe(III)-SRFA and AFO within the model according to the pathway of formation. Figure 1 illustrates the two pathways of formation for Fe(III)-SRFA: (1) Associative oxidation (AO), where Fe(II)-SRFA is oxidized, and (2) nonassociative oxidation (NAO), where Fe(II) is oxidized prior to complex formation. A complex dependency of the proportion that each pathway contributes to the formation of Fe(III)-SRFA on the initial concentrations of Fe(II) and SRFA present is revealed in Figure 5a and may be summarized and explained as follows: 1. At low [SRFA], increasing [Fe(II)] leads to an increasing contribution from AO due to the increasing impact of the autocatalytic nature of heterogeneous precipitation. 2. At high [SRFA], increasing [Fe(II)] leads to a decreasing contribution from AO due to decreasing complexation of Fe(II) (Figure 5b) and the ineffectiveness of low [AFO] to compete against Fe(III) complexation. 3. At low [Fe(II)], increasing [SRFA] leads to an increasing contribution from AO due to increasing complexation of Fe(II) (Figure 5b). 4. At high [Fe(II)], increasing [SRFA] leads a decreasing contribution from AO due to the greater increase in Fe(III) complexation (reaction 14) compared to the increase in Fe(II) complexation (reaction 12) illustrated by the greater sensitivity of the model to reaction 14 compared to reaction 12 (Figure 4). The role of AFO reactions over longer time periods in a high ionic strength medium such as seawater is likely to be significantly smaller than predicted here since diffusionlimited aggregation and/or transformation to more crystalline forms is likely to substantially reduce the accessibility of AFO sites for reaction. When a reaction for the transformation of 6672

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 17, 2010

Literature Cited (1) Boyd, P. W.; Law, C. S. The Southern Ocean Iron RElease Experiment (SOIREE) - introduction and summary. Deep-Sea Res. Part II 2001, 48 (11-12), 2425–2438. (2) Ahern, K. S.; Ahern, C. R.; Udy, J. W. In situ field experiment shows Lyngbya majuscula (cyanobacterium) growth stimulated by added iron, phosphorus and nitrogen. Harmful Algae 2008, 7 (4), 389–404. (3) Powell, R. T.; Wilson-Finelli, A. Importance of organic Fe complexing ligands in the Mississippi River plume. Estuarine, Coastal Shelf Sci. 2003, 58 (4), 757–763. (4) Wilhelm, S. W.; Trick, C. G. Iron-limited growth of cyanobacteria: Multiple siderophore production is a common response. Limnol. Oceanogr. 1994, 39 (8), 1979–1984. (5) Morel, F. M. M.; Hering, J. G., Principles and Applications of Aquatic Chemistry; Wiley: New York, 1993. (6) Rose, A. L.; Waite, T. D. Kinetics of iron complexation by dissolved natural organic matter in coastal waters. Mar. Chem. 2003, 84 (1-2), 85–103. (7) Santana-Casiano, J. M.; Gonzalez-Davila, M.; Millero, F. J. The oxidation of Fe(II) in NaCl-HCO3- and seawater solutions in the presence of phthalate and salicylate ions: a kinetic model. Mar. Chem. 2004, 85 (1-2), 27–40. (8) Rose, A. L.; Waite, T. D. Kinetic model for Fe(II) oxidation in seawater in the absence and presence of natural organic matter. Environ. Sci. Technol. 2002, 36 (3), 433–444. (9) Pullin, M. J.; Cabaniss, S. E. The effects of pH, ionic strength, and iron-fulvic acid interactions on the kinetics of nonphotochemical iron transformations. I. Iron(II) oxidation and iron(III) colloid formation. Geochim. Cosmochim. Acta 2003, 67 (21), 4067–4077. (10) Sung, W.; Morgan, J. J. Kinetics and product of ferrous iron oxygenation in aqueous systems. Environ. Sci. Technol. 1980, 14 (5), 561–568. (11) Tamura, H.; Goto, K.; Nagayama, M. Effect of ferric hydroxide on oxygenation of ferrous-ions in neutral solutions. Corros. Sci. 1976, 16 (4), 197–207. (12) King, D. W.; Farlow, R. Role of carbonate speciation on the oxidation of Fe(II) by H2O2. Mar. Chem. 2000, 70 (1-3), 201– 209. (13) Weber, L.; Volker, C.; Schartau, M.; Wolf-Gladrow, D. A. Modeling the speciation and biogeochemistry of iron at the Bermuda Atlantic Time-series Study site. Global Biogeochem. Cycls 2005, 19 (1), . (14) Pham, A. N.; Rose, A. L.; Feitz, A. J.; Waite, T. D. Kinetics of Fe(III) precipitation in aqueous solutions at pH 6.0-9.5 and 25°C. Geochim. Cosmochim. Acta 2006, 70 (3), 640–650. (15) Rose, A. L.; Waite, T. D. Kinetics of hydrolysis and precipitation of ferric iron in seawater. Environ. Sci. Technol. 2003, 37 (17), 3897–3903. (16) Rose, A. L.; Waite, T. D. Predicting iron speciation in coastal waters from the kinetics of sunlight-mediated iron redox cycling. Aquat. Sci. 2003, 65 (4), 375–383.

(17) Pullin, M. J.; Cabaniss, S. E. Colorimetric flow-injection analysis of dissolved iron in high DOC waters. Water Res. 2001, 35 (2), 363–372. (18) Fujii, M.; Ito, H.; Rose, A. L.; Waite, T. D.; Omura, T. Transformation dynamics and reactivity of dissolved and colloidal iron in coastal waters. Mar. Chem. 2008, 110 (3-4), 165–175. (19) Jones, A. M.; Pham, A. N.; Collins, R. N.; Waite, T. D. Dissociation kinetics of Fe(III)- and Al(III)-natural organic matter complexes at pH 6.0 and 8.0 and 25 degrees C. Geochim. Cosmochim. Acta 2009, 73 (10), 2875–2887. (20) Rose, A. L.; Waite, T. D. Effect of dissolved natural organic matter on the kinetics of ferrous iron oxygenation in seawater. Environ. Sci. Technol. 2003, 37 (21), 4877–4886. (21) Santana-Casiano, J. M.; Gonzalez-Davila, M.; Rodriguez, M. J.; Millero, F. J. The effect of organic compounds in the oxidation kinetics of Fe(II). Mar. Chem. 2000, 70 (1-3), 211–222. (22) Wulkow, M., PrestosSimulation of Kinetic Models; CiT GmbH: Rastede, Germany, 2004. (23) Ianni, J. C. Kintecus, V 3.9, 2006. (24) Blaser, P.; Fluhler, H.; Polomski, J. Metal-binding properties of leaf litter extracts. 1. Discontinuous spectrophotometric titration with iron and copper. Soil Sci. Soc. Am. J. 1980, 44 (4), 709–716. (25) Pandeya, S. B.; Singh, A. K. Discontinuous spectrocolorimetric titration method for determining stability constants of fulvic acid-iron complexes. Aust. J. Soil Res. 1997, 35 (6), 1279–1290. (26) Emmenegger, L.; King, D. W.; Sigg, L.; Sulzberger, B. Oxidation kinetics of Fe(II) in a eutrophic Swiss lake. Environ. Sci. Technol. 1998, 32 (19), 2990–2996. (27) Ritchie, J. D.; Perdue, E. M. Proton-binding study of standard and reference fulvic acids, humic acids, and natural organic matter. Geochim. Cosmochim. Acta 2003, 67 (1), 85–96. (28) Margerum, D. W.; Cayley, G. R.; Weatherburn, D. C.; Pagenkopf, G. K., Kinetics and mechanisms of complex formation and ligand exchange. In Coordination Chemistry, Vol. 2; Martell, A. E., Ed.; American Chemical Society: Washington D.C., 1978. (29) Fujii, M.; Rose, A. L.; Waite, T. D.; Omura, T. Effect of divalent cations on the kinetics of Fe(III) complexation by organic ligands in natural waters. Geochim. Cosmochim. Acta 2008, 72 (5), 1335– 1349.

(30) King, D. W. Role of carbonate speciation on the oxidation rate of Fe(II) in aquatic systems. Environ. Sci. Technol. 1998, 32 (19), 2997–3003. (31) Bao, X. M.; Yu, T. R. Stability constants of Fe2+ chelates with soluble ligands from incubated soils. Biol. Fertil. Soils 1987, 5 (1), 88–92. (32) Schnitzer, M.; Skinner, S. I. M. Organo-metallic interactions in soilss0.5. Stability constants of Cu+2-, Fe+2-, and Zn+2-fulvic acid complexes. Soil Sci. 1966, 102 (6), 361–365. (33) Wu, J. F.; Luther, G. W. Complexation of Fe(III) by natural organic-ligands in the northwest atlantic-ocean by a competitive ligand equilibration method and a kinetic approach. Mar. Chem. 1995, 50 (1-4), 159–177. (34) Chen, Z. Q.; Hu, C. M.; Comny, R. N.; Muller-Karger, F.; Swarzenski, P. Colored dissolved organic matter in Tampa Bay, Florida. Mar. Chem. 2007, 104 (1-2), 98–109. (35) Ford, P.; Tillman, P.; Robson, B.; Webster, I. T. Organic carbon deliveries and their flow related dynamics in the Fitzroy estuary. Mar. Pollut. Bull. 2005, 51 (1-4), 119–127. (36) Lubben, A.; Dellwig, O.; Koch, S.; Beck, M.; Badewien, T.; Fischer, S.; Reuter, R. Distributions and characteristics of dissolved organic matter in temperate coastal waters (Southern North Sea). Ocean Dyn. 2009, 59 (2), 263–275. (37) Albert, S.; O’Neil, J. M.; Udy, J. W.; Ahern, K. S.; O’Sullivan, C. M.; Dennison, W. C. Blooms of the cyanobacterium Lyngbya majuscula in coastal Queensland, Australia: disparate sites, common factors. Mar. Pollut. Bull. 2005, 51 (1-4), 428–437. (38) Bligh, M. W. Formation, Fate and Transformation of Products of Iron Oxidation in Coastal Waters, Ph.D Thesis; The University of New South Wales: Sydney, 2010. (39) Bligh, M. W.; Waite, T. D. Formation, Aggregation and Reactivity of Amorphous Ferric Oxyhydroxides on Dissociation of Fe(III)Organic Complexes in Dilute Aqueous Suspensions. Geochim. Cosmochim. Acta 2010, (in press). (40) Pham, A. N., Kinetics and Mechanism of Various Iron Transformations in Natural Waters at Circumneutral pH, PhD Thesis; The University of New South Wales: Sydney, 2007.

ES101046Y

VOL. 44, NO. 17, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6673