Environ. Sci. Technol. 1998, 32, 2969-2975
Dissolution of Iron Oxides by Phagotrophic Protists: Using a Novel Method To Quantify Reaction Rates KATHERINE A. BARBEAU AND JAMES W. MOFFETT* Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543
In previous work, we have reported the dissolution of iron oxides within the acidic food vacuoles of marine protozoan grazers as evidence of a novel mechanism for the conversion of refractory iron solids to more labile forms in oxic surface waters. This paper expands upon those initial studies and presents a new technique to study the reaction of iron oxides in seawater, based on the synthesis of colloidal ferrihydrite containing an inert tracer. Measuring the accumulation of the tracer in the dissolved phase enables the determination of the rate and extent of iron oxide reaction, even for kinetically slow processes and regardless of the fate of iron in the system. The validity of the method as a means of following the reaction of iron oxides in seawater is shown here in a series of codissolution studies and in several photochemical kinetics experiments. In laboratory studies of the dissolution of colloidal ferrihydrite by protozoan grazers, the inert tracer method enables an improved estimate of the rate of protozoan-mediated iron oxide dissolution, confirming our previous results and providing a useful tool for further studies of phagotrophy as a reaction pathway for refractory iron.
with radiolabeled Fe colloids (10); competitive ligand exchange/cathodic stripping voltammetry with 1-nitroso-2naphthol (11); and Fe-limited diatoms as probes of bioavailable Fe. Results indicated that iron oxide dissolution occurred as a consequence of the ingestion of colloidal iron oxide by protozoan grazers via reaction within the acidic, reducing microenvironment of the phagotrophic digestive vacuole (12). This paper presents the results of further studies of the dissolution of iron oxides by protozoan grazers, using a novel tracer methodology. This method is based on the synthesis of colloidal ferrihydrite uniformly impregnated with the radiotracer 133Ba. In seawater, when the iron oxide structure is altered by chemical reaction, 133Ba is released into the dissolved phase due to a combination of low affinity for the oxide surface and ion exchange with other alkali earth cations in seawater (Mg2+ ,Ca2+). 133Ba remains in the dissolved phase because it is generally nonparticle reactive and not biologically active. The rate of accumulation of dissolved 133Ba in a given system provides a relative measure of the rate at which the iron oxide phase is reacting. Thus, unlike techniques that rely on detection of reactive iron (10, 1315), the accumulation of dissolved 133Ba provides a signal of the extent of reaction undergone by the iron oxide, even though the reaction rate may be low enough that measurable amounts of reactive iron do not accumulate in the system. This paper presents details of the inert tracer method, beginning with the protocol for synthesis of ferrihydrite impregnated with 133Ba. To demonstrate that 133Ba is incorporated throughout the iron oxide matrix and dissolved congruently with Fe, the results of a series of co-dissolution experiments are shown. Several photochemical kinetics experiments with the 133Ba-impregnated ferrihydrite are also presented to demonstrate that the results obtained with this method are consistent with previous studies of iron oxide photochemistry. Finally, the inert tracer method is applied in a series of laboratory culture studies to obtain quantitative estimates of the rate and efficiency of dissolution of colloidal ferrihydrite as mediated by protozoan grazers.
Experimental Methods Introduction Iron, one of the most abundant metals in the earth’s crust, exists in an array of dissolved, particulate, and colloidal forms in aqueous systems (1, 2). In natural surface waters, exchange among these various species is driven by several mechanisms, including photoreduction, ligand complexation, enzymatic reactions, and thermodynamic reactions (3-6). These processes are important to the global geochemical cycle of organic carbon, in some cases as an oxidation pathway for reduced organic compounds and more generally as an influence on the bioavailability of iron, a limiting nutrient, to primary producers (7, 8). Reactions that affect the speciation of iron in surface waters are thus of general interest. Using the iron oxide mineral ferrihydrite as a model phase, we have previously documented the existence of a novel reaction pathway for refractory iron solids in oxygenated waters, mediated by protozoan grazers (9). Four independent techniques were employed to investigate the effect of protozoan grazing on colloidal ferrihydrite in laboratory culture systems: Size fractionation of radiolabeled Fe colloids in systems with EDTA as a dissolved Fe trap; oxine lability * Corresponding
[email protected].
author
S0013-936X(98)00254-5 CCC: $15.00 Published on Web 08/26/1998
phone:
(508)289-3218;
e-mail:
1998 American Chemical Society
Synthesis of 133Ba-Impregnated Colloidal Ferrihydrite. The procedure for synthesizing 133Ba-impregnated ferrihydrite is based on the common method of synthesizing ferrihydrite via neutralization of a ferric salt solution (16, 17). 133Ba (µM concentration) and 59Fe (trace amount) are added to Milli-Q water, and the pH of the water adjusted to ∼5 (with rapid stirring). Fe(NO3)3‚9H2O is then added and dissolved (concentration 2.5 mM Fe). A stoichiometric amount of 0.0375 M NaOH solution (1/5th the volume of the ferric nitrate solution) is then rapidly added to the ferric nitrate salt while stirring to quickly bring the pH to between 7 and 8. This results in the instantaneous precipitation of orange flocs. The solution is allowed to stand in this condition for about 15 min prior to further processing. The initial product of this basic hydrolysis is a 2-line ferrihydrite. This material is, in general, too reactive for practical use in experiments. To decrease the reactivity of the 2-line ferrihydrite, it is heated at 90 °C in a water bath for 5 min (at pH 7-8) and then rapidly cooled to room temperature by submersion in an ice-water bath. The heating step causes the phase to condense somewhat, so that it becomes more ordered but still subcrystalline, more like a 6-line than a 2-line ferrihydrite (10, 18). Before this product can be used in tracer experiments, it is critical to thoroughly rinse away the excess 133Ba not actually VOL. 32, NO. 19, 1998 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2969
incorporated in the oxide matrix. Following the heating/ cooling step, the ferrihydrite suspension is well-mixed and diluted by a factor of 10 with Milli-Q water. The pH is adjusted to 5 by adding dilute HCl. The solution is then briefly ultrasonicated with a 1/8-in. probe in order to disperse the aggregated phase, forming a stable colloidal sol at pH 5. The colloidal sol is loaded into dialysis bags (Spectrum, Spectra/ Por CE sterile membranes, 3500 MWCO, 20 mL sample volume). The colloids are dialyzed in Milli-Q water at pH 5 for 60 h (5 dialysate cycles). Stable BaCl2 (at 10-4 M) is added to the third or fourth batch of dialysate in order to enhance removal of 133Ba from the surface of the ferrihydrite. A fresh batch of colloidal ferrihydrite was synthesized and dialyzed for all experiments. Co-dissolution. Several co-dissolution studies were performed in order to demonstrate uniform incorporation of 133Ba in the oxide and concurrent production of dissolved 133 Ba and dissolved Fe. In one study, co-dissolution of the dialyzed ferrihydrite colloids was perfomed by a cation resin (Bio-Rad 100-200 mesh AG 50W-X8, H+ form) slurry extraction at pH 5. Starting at t0, the colloidal sol was sampled to determine the total concentration of 133Ba and 59Fe prior to extraction. Then cation-exchange resin was added and stirred to create a slurry. Periodically, aliquots of the slurry were removed and filtered through a coarse frit to remove the resin beads, and the aliquot of extracted sol was saved for radioactive counting. In this way, the disappearance of each isotope from the colloidal sol was monitored. Isotopes were counted on a Canberra low-energy germanium detector coupled to an Ortec multichannel analyzer. In another study, a series of three co-dissolution experiments was performed in Sargasso seawater. 133Ba/59Fe colabeled ferrihydrite colloids were added to 0.2 µm-filtered Sargasso seawater at 1 µM concentration, equilibrated 1 h, and then the chemistry of the seawater was adjusted so as to initiate dissolution of the ferrihydrite. In one experiment, seawater pH was reduced to 2.5. In another, seawater pH was reduced to 4, and 100 µM of citrate was added. In the final experiment in this series, seawater pH was adjusted to 2, and a 2 mM concentration of the reducing reagent hydroxylamine hydrochloride was added. Starting at the end of the equilibration period and continuing over a time course (4 h for the pH 2.5 and hydroxylamine experiments, 11.5 h for the citrate experiment), 0.05 µm filtrate and total samples were taken. Filtrate samples were obtained using a syringe filtration device with a 47 mm diameter, 0.05 µm pore-size Nuclepore polycarbonate membrane. Total and filtrate samples of equivalent weight were placed in polypropylene sample cups for determination of 133Ba and 59Fe activity. An aliquot of hydroxylamine hydrochloride solution was added to all samples to dissolve colloidal ferrihydrite in totals and prevent wall loss. Photochemical Experiments. The photochemistry of iron oxides in natural waters is an environmentally significant phenomenon and as such has been the object of much research. Previous studies provide a context within which results using a novel tracer methodology can be interpreted. Particularly applicable in this case is the work of Waite and Morel and Wells and Mayer (19, 20), which involved reactions of synthetic iron oxides in seawater under different conditions of pH and chromophore concentration. The light source used in the experiments described here was a 1000-W Hg/Xe lamp equipped with an infrared filter and a Pyrex ultraviolet filter. The lamp produced a wellcollimated beam of light with a broad (∼solar) spectrum, about 10 times as intense as natural sunlight. Samples were irradiated in quartz cells (30 mL capacity, 10 cm path length) in front of the light source for a maximum of 1 h. Although all samples started at room temperature, some increase in sample temperature was observed during longer irradiation 2970
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 19, 1998
periods. Dissolved 59Fe and 133Ba activity in irradiated samples and nonirradiated blanks were determined by syringe filtration as described above. Samples were also taken for total activity. Several photochemical kinetics experiments were performed in 0.2 µm-filtered, aged Sargasso seawater (SSW) with 100 µM citrate added at pH 8 and at pH 6.5. All irradiations were performed on the same day with the same batch of colloidal ferrihydrite. 133Ba/59Fe co-labeled colloidal ferrihydrite (2 µM Fe) was added to pH 8 SSW with citrate and allowed to equilibrate for 1 h, and then irradiations were performed (sequentially) for 5, 25, and 50 min. Blank (nonirradiated) samples were also taken for each of those time periods. The pH of the SSW + citrate was then adjusted to 6.5 by adding HCl, and three more irradiations were performed (and blank samples taken) for 5, 25, and 50 min. Protozoan Grazing Experiments. Three different grazing experiments are described here. All were carried out in 250mL polycarbonate bottles containing 200 mL of culture (basal medium 0.2 µm-filtered sterilized Vineyard Sound seawater, VSW) and incubated at room temperature without shaking, under normal laboratory lighting. Bacterial prey in all experiments was Halomonas halodurans, a common marine heterotroph. Bacteria were grown up separately in yeast extract or glucose media and rinsed and resuspended 3× by centrifugation before being added to experimental systems. Several heterotrophic protozoan species were used in these experiments, all taken from the culture collection of D. Caron at WHOI. These species included Cafeteria sp. (flagellate, 2-4 µm, clone Cflag), Uronema (scuticociliate, ∼10 µm, clone BBcil), Pteridomonas (flagellate, 3-5 µm, clone NB1), and a hymenostome ciliate (∼8 µm, clone Hcil). Protists were inoculated into model grazing systems at t0, and experiments generally ran for about 50 h. 133Ba/59Fe co-labeled colloidal ferrihydrite was added to all experiments at concentrations ∼1-2 µM Fe. In the first grazing experiment, production of dissolved 133 Ba and 59Fe from co-labeled colloids by grazers was compared in model systems both with and without EDTA added as a trap for dissolved Fe. The following model systems were set up, each in duplicate: VSW with 133Ba/59Fe ferrihydrite; VSW with 133Ba/59Fe ferrihydrite and 10 µM EDTA; VSW with 133Ba/59Fe ferrihydrite, H. halodurans, and Cafeteria; and VSW with 133Ba/59Fe ferrihydrite, H. halodurans, Cafeteria, and 10 µM EDTA. The H. halodurans concentration at t0 was ∼2.5 × 107 cells/mL; the Cafeteria concentration was ∼104 cells/mL. Dissolved 133Ba and 59Fe in this experiment were isolated by a dialysis technique. Sterile cellulose ester dispodialyzers (Spectrum, 3500 MWCO, 5 mL capacity) were filled with VSW and added to each model system. The dispodialyzers were allowed to remain suspended in the model systems for about 24 h, during which time dissolved 133Ba and 59Fe were collected inside the bags by passive diffusion. For collection, the dispodialyzers were removed, and their contents weighed into polypropylene sample cups. For total samples, an equivalent weight of whole culture was removed at the same time and placed in a sample cup. In the second grazing experiment, 133Ba-impregnated iron oxides were used to make a quantitative estimation of the rate and efficiency of colloidal ferrihydrite dissolution by Cafeteria. The following model systems were set up: VSW with 133Ba/59Fe ferrihydrite and heat-killed H. halodurans (duplicate); VSW with 133Ba/59Fe ferrihydrite, heat-killed H. halodurans, and Cafeteria (triplicate); and VSW with heatkilled H. halodurans and Cafeteria (duplicate). To ensure that the colloidal ferrihydrite was uniformly associated with bacterial biomass, several hours prior to the start of the experiment, heat-killed H. halodurans and 133Ba/ 59Fe colloidal ferrihydrite were mixed together at a concen-
FIGURE 1. Co-dissolution experiments. (A) Co-dissolution by cation resin extraction. (B) Co-dissolution at pH 2.5 in seawater. (C) Codissolution in seawater at pH 4 with 100 µM citrate added. (D) Co-dissolution in seawater at pH 2 with 2 mM hydroxylamine hydrochloride added. tration of 3 × 108 cells/mL bacteria and 10-5 M Fe as colloidal ferrihydrite in VSW. This mixture was placed on a rotary shaker for several hours. Scavenging of the colloidal iron by bacterial cell surfaces was evidenced by the visible formation of orange flocs. For use in the experiment, mixed aliquots of the flocculated ferrihydrite/bacteria suspension were briefly ultrasonicated (10 s) to disaggregate the flocs and then immediately added to VSW in the experimental bottles at 1/10 dilution. This experiment was run over a period of 54 h. Nonlabeled grazing cultures were sampled for protist and bacteria counts throughout this time. Cell counting was done by standard epifluorescence microscopy methods on acridine orangestained samples. In the radiolabeled cultures, three successive 18-h dialysis samples (5000 MWCO) were obtained over the 54-h time course. The final grazing experiment compared the relative efficiency of several different cultured strains of protozoan grazers at dissolving colloidal iron oxides. Species tested included Cafeteria, Uronema, Pteridomonas, and the hymenostome ciliate (Hcil). Replicate cultures of each grazer were set up with live bacterial prey and 133Ba/59Fe-labeled
ferrihydrite along with two control cultures containing only bacteria and labeled ferrihydrite. Additional replicate cultures of each grazer with bacteria and nonlabeled ferrihydrite were set up to sample for cell counts. Bacteria cell concentrations in all cultures at t0 were 6.7 × 106 ( 4.6 × 105 cells/mL. Protozoan cell concentrations at t0 were as follows: Cafeteria, 1.5 × 104 cells/mL; Uronema, 5.8 × 102 cells/mL; Pteridomonas, 7.3 × 103 cells/mL; Hcil, 2.9 × 103 cells/mL. Starting at t0, model systems were sampled every 10 h for dissolved and total 133Ba and 59Fe and for cell counts. Uronema cultures were run for 40 h; cultures of all other organisms were run for 50 h. For determination of dissolved 133Ba and 59Fe, syringe filtration was employed as described previously.
Results and Discussion Co-dissolution Experiments. Results of the co-dissolution experiments are shown in Figure 1, panels A (resin extraction), B (pH 2.5 dissolution), C (pH 4 plus citrate dissolution), and D (hydroxylamine hydrochloride dissolution). It is apparent that the 133Ba and Fe (as 59Fe) dissolve congruently, following the 1:1 dissolution line fairly closely in all cases. This VOL. 32, NO. 19, 1998 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2971
FIGURE 2. Photochemical kinetics experiments in Sargasso seawater with 100 µM citrate added; at pH 6.5 and pH 8. (Black circles) Nonirradiated blanks. (Gray circles) Irradiated samples. (A) Percent dissolved 133Ba at pH 6.5. (B) Percent dissolved 59Fe at pH 6.5. (C) Percent dissolved 133Ba at pH 8. (D) Percent dissolved 59Fe at pH 8. There are no replicate samples; the errors in some of the 59Fe graphs are counting statistics for filtrates with very low 59Fe activity. demonstrates that the 133Ba is evenly distributed within the oxide framework. These data indicate that production of dissolved 133Ba is a good proxy for the dissolution of iron oxides in seawater. It is important to note that 133Ba is only a trace component in the iron oxide. In the final iron oxide product, the Fe/133Ba ratio is about 105/1. Thus, the 133Ba is not present in the oxide as a solid solution or separate phase but rather as a trace component scattered throughout the oxide matrix. This is shown by the co-dissolution results. As a trace component, the 133Ba does not have any effect on the structural or chemical properties of the iron oxide. Such effects are only known to occur at much higher levels of cation substitution (1). Photochemical Experiments. Figure 2, panels A-D, shows the result of the photochemical kinetics experiments. For panels A and B, the experiments at pH 6.5, the initial behavior of 133Ba and 59Fe is similar. Assuming simple firstorder reaction kinetics, the initial rate constant (0-5 min) of dissolved 133Ba release by photodissolution is 3.48 h-1. The initial rate for 59Fe release is similar at 3.22 h-1. 59Fe in this system is likely stabilized in the dissolved phase by citrate complexation. The rate of 133Ba release by photodissolution 2972
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 19, 1998
apparently decreases with time, with dissolved 133Ba reaching a plateau by 50 min. A decreasing photoreduction rate constant with time is consistent with previous studies of iron oxide photolysis at pH 6.5 and higher (19, 21). Decreasing rates of photolysis with time have been attributed to depletion of organic chromophores and/or the inactivation of the iron oxide surface by photolysis products (20, 22). In this work, an additional factor may be the depletion of 133Ba at the iron oxide surface or dissolution of a major fraction of the solid. The large drop in dissolved 59Fe at 50 min could be caused by photolysis loss of citrate due to the extended irradiation period and the intense iron redox cycling in this system, despite the initial excess of citrate relative to Fe. Further work is needed to confirm the validity of this interpretation. In Figure 2C, for the experiment at pH 8, the increase in dissolved 133Ba in the lighted flask relative to the dark control shows clear evidence of photodissolution, although the initial rate constant (0.14 h-1 for 0-5 min) is about 25× less than that observed at pH 6.5. The pH dependence of the photolysis rates derived here is entirely consistent with previous studies (20-22). The lowered dissolution rate of the iron oxide at pH 8 relative to pH 6.5 is most likely a consequence of both
FIGURE 3. First grazing experiment with Cafeteria at t ) 48 h; % dissolved 59Fe (gray bars) vs % dissolved 133Ba (white bars) in model systems with and without 10 µM EDTA. Error bars represent standard deviation of measurements in replicate bottles. the reduced affinity of citrate for the iron oxide surface at higher pH (22) and also the strong pH dependence of the oxidation rate of Fe(II) (23). Dissolved 59Fe in the irradiated sample shows no increase relative to the nonirradiated blank, within counting error (Figure 2D). This indicates that, although photoreduction is altering the iron oxide structure as indicated by the release of 133Ba, Fe is being rapidly recycled at the oxide surface without going into solution. This cycle of Fe reduction, reoxidation, and reprecipitation at the colloid surface probably plays a role in reducing the rate of 133Ba release at pH 8 relative to pH 6.5 by creating a reactive layer on the oxide surface that is depleted in 133Ba and that shields underlying, more 133Ba-rich layers from further reaction. A decrease in the rate of dissolved 133Ba release by photodissolution over time is observed at pH 8 but less so than at pH 6.5. The more prolonged release of dissolved 133Ba at pH 8 may be due to the lower rate of iron redox cycling in this system, concurrent with lower rates of citrate photolysis and surface inactivation than at pH 6.5. Grazing Experiments. Results of the first grazing experiment are shown in Figure 3 at t ) 48 h. It can be seen that Cafeteria produces a significant dissolved 133Ba signal relative to the 0.2 µm filtered control, the magnitude of the signal being the same (within error) both with and without EDTA. In contrast, much more dissolved 59Fe is generated in grazing cultures with EDTA than without. These data demonstrate how 133Ba is being released from the colloids through grazing and accumulating in the dissolved phase, while 59Fe rapidly cycles back onto particle surfaces without building up a measurable dissolved concentration unless EDTA is present to trap Fe in the dissolved phase. Figure 4, panels A and B, shows the results of the second grazing experiment with Cafeteria. Protist and bacteria cell counts in the grazing cultures vs time are shown in Figure 4A. In Figure 4B, dialysis results are presented as the excess percent of dissolved 133Ba in grazing cultures (i.e., values have been corrected for dissolved 133Ba in the nongrazing bacteria controls, which averaged about 7-8% over the entire time course). Because significant wall loss of both 133Ba and 59Fe occurred in the grazing cultures (but not in bacteria controls) in this experiment, dialysis results are normalized to the total 133Ba activity at t0 . Wall loss of 59Fe and 133Ba in grazing cultures appears to be due to adsorption of colloidal ferrihydrite to bottle walls. (We have ascertained that 133Ba
FIGURE 4. Second grazing experiment with Cafeteria. (A) Population dynamics of Cafeteria (9) and bacteria (() in grazing cultures. (B) Dialysis data for excess % 133Ba
