Dieke Postma

Nitrate Reduction in an Unconfined Sandy Aquifer: Water Chemistry, Reduction Processes, and Geochemical Modeling

Nitrate distribution and reduction processes were investigated in an unconfined sandy aquifer of Quaternary age. Ground water chemistry was studied in a series of eight multilevel samplers along a flow line, deriving water from both arable and forested land. Results show that plumes of nitrate-contaminated groundwater emanate from the agricultural areas and spread through the aquifer. The aquifer can be subdivided into an upper 10- to 15-m thick oxic zone that contains O2 and NO3−, and a lower anoxic zone characterized by Fe2+-rich waters. The redox boundary is very sharp, which suggests that reduction processes of O2 and NO3− occur at rates that are fast compared to the rate of downward water transport. Nitrate-contaminated groundwater contains total contents of dissolved ions that are two to four times higher than in groundwater derived from the forested area. The persistence of the high content of total dissolved ions in the NO3−-free anoxic zone indicates the downward migration of contaminants and that active nitrate reduction is taking place. Nitrate is apparently reduced to N2 because both nitrite and ammonia are absent or found at very low concentrations. Possible electron donors in the reduced zone of the aquifer are organic matter, present as reworked brown coal fragments from the underlying Miocene, and small amounts of pyrite at an average concentration of 3.6 mmol/kg. Electron balances across the redoxcline, based on concentrations of O2, NO3−, SO42− and total inorganic carbon (TIC), indicate that pyrite is by far the dominant electron donor even though organic matter is much more abundant. Groundwater transport and chemical reactions were modeled using the code PHREEQM, which combines a chemical equilibrium model with a one-dimensional mixing cell transport model. Only the vertical component of the water transport was modeled since, in contrast to rates along flow lines, the vertical rates are close to constant as required by the one-dimensional model. Average vertical transport rates of water in the saturated zone were obtained by tritium dating. The modeling process is a two-step procedure. First the sediment column is initialized with natural water containing only oxygen as electron acceptor, and subsequently agricultural waters containing both oxygen and nitrate are fed into the column. The nitrate concentration of agricultural waters entering the saturated zone varies with time, and an input function was therefore constructed by linear mixing of natural waters and agricultural waters. This input function was fed into the column initialized with natural water, and the model run forward in time to the year 1988 where field data are available. Comparison with field data shows that the variation in groundwater chemistry is well described by the model when reduction of oxygen and reduction of nitrate by pyrite oxidation are the only redox reactions occurring. Finally, predictions are made for the distribution of water chemistry in the year 2003. Downward progression of the redoxcline is accelerated by a factor of five due to nitrate pollution of the aquifer, but absolute rates remain small, of the order of a few centimeters per year. The controlling factor for nitrate migration through the aquifer, once it has reached the anoxic zone, is the concentration and distribution of pyrite in the sediments.

Redox zonation: Equilibrium constraints on the Fe(III)/SO4-reduction interface

Abstract The concept of redox zonation during degradation of organic matter, which is usually explained by the overall energy yield of different reactions, has been reevaluated. At least for reduction of Mn(IV), Fe(III), sulfate, and methanogenesis, the sequential occurrence of these processes is much easier explained by a partial equilibrium approach where the fermentive step is overall rate limiting, while the electron accepting processes are considered to be close to equilibrium. Using the partial equilibrium approach, an explanation is sought for the simultaneous occurrence of Fe(III) and sulfate reduction, observed in several field studies. Calculations of conditions for equilibrium between Fe(III) and sulfate reduction indicate that, depending on the stability of the iron oxide, simultaneous reduction of Fe(III) and sulfate is thermodynamically possible under a wide range of sedimentary conditions and sulfate reduction may even occur before Fe(III) reduction. In Fe2+-rich environments, the pH of the porewater has in addition a strong influence on whether Fe(III) or sulfate reduction is favored. In natural sediments, the presence of a wide range of iron oxide stabilities is likely to cause considerable overlap between zones of Fe(III) and sulfate reduction, while a better confined stability range of iron oxides should cause more distinct zones of Fe(III) and sulfate reduction.

REDOX ZONING, RATES OF SULFATE REDUCTION AND INTERACTIONS WITH FE-REDUCTION AND METHANOGENESIS IN A SHALLOW SANDY AQUIFER, ROMO, DENMARK

Abstract Fe-oxide reduction, sulfate reduction and methanogenesis, have been studied in a shallow aquifer with the main focus on sulfate reduction. Direct measurements of sulfate reduction rates have for the first time been applied in an aquifer system. Rates were much lower than reported in other anoxic environments - three orders of magnitude lower than in marine settings and one order of magnitude lower than in lacustrine environments, and varied substantially mainly due to differences in the reactivity of the organic matter. At the extremely low substrate levels in the aquifer, sulfate reduction rates are not primarily limited by low sulfate concentrations. The produced sulfide forms framboidal pyrite via a FeS precursor, with elemental sulfur as an intermediate. Fe-oxide reduction rates were comparable to sulfate reduction rates, but appeared to depend more on Fe-oxide reactivity than organic matter reactivity. Low sulfate concentrations, combined with low-reactivity Fe(III), in the aquifer sediment, has led to an increased appreciation of the existence of concomitant redox processes. This indicates that competitive exclusion is not always effective, and raises questions as to what H 2 data reflect in such a system. Calculations of the in situ energy yield for Fe- and sulfate reduction via H 2 oxidation are ∼2.5 kcal/mol H 2 , indicating that thermodynamic equilibrium is approached. The calculated available energy yield for methanogenesis was very low indicating that CH 4 production must occur in micro-environments where higher H 2 concentrations prevail. The system may be described as being in a state of partial equilibrium, where the overall rate of organic matter oxidation is controlled by the rate of fermentation of the organic matter, and terminal electron acceptor processes occur at close to equilibrium conditions . This partial equilibrium depends on other processes in the system, in this case an increase in pH due to calcite dissolution appears to induce a shift from predominantly Fe-reducing to predominantly sulfate reducing conditions by changing the energy available to Fe-oxide reduction. Numerical modeling using the partial equilibrium approach was successful in modeling this complex of interacting processes.

Kinetics of reductive bulk dissolution of lepidocrocite, ferrihydrite, and goethite

—The variation in Fe-oxide reactivity was investigated by studying the kinetics of bulk reductive dissolution of a suite of synthetic Fe-oxides in 10 mM ascorbic acid at pH 3. The Fe-oxides comprised three different ferrihydrites, five lepidocrocites, and a poorly crystalline goethite. During one of the reduction experiments, lepidocrocite crystals were subsampled and the change in crystal habit and size distribution was studied by transmission electron microscopy. The rate of complete dissolution was described by the function J/m0 = k′(m/m0)γ where J is the overall rate of dissolution (mol/s), m0 the initial amount of iron oxides, and m/m0 the undissolved mineral fraction. Rate laws were derived for the different iron oxides and showed a variation in initial rates of about two orders of magnitudes; 2-line ferrihydrites being most reactive with k′ = 7.6–6.6 × 10−4 · s−1, whereas the initial rate for 6-line ferrihydrite is an order of magnitude lower 7.4 × 10−5 · s−1 and comparable to the quite homogeneous group of lepidocrocites (3.2–8.1 × 10−5 · s−1) with finally the initial rate of goethite being one order of magnitude lower again (5.4 × 10−6 · s−1). The transmission electron microscopy results for lepidocrocite showed strong etch-pitting of the crystals parallel to the c-axis resulting ultimately in disintegration of the crystals. For the different iron oxides, the initial rate was independent of the specific surface area, emphasizing the importance of the crystal structure for the dissolution rate. However, among the lepidocrocites the initial rate was proportional to the specific surface area. The exponent, γ was found to vary from a value near 1.0 for one of the 2-line ferrihydrites, two of the lepidocrocites and the goethite, to values close to 2.3 for the other 2-line ferrihydrite and the 6-line ferrihydrite. Thus, the largest variation in reduction rate during bulk dissolution is found for ferrihydrite. For the lepidocrocites, the preparations that predominantly consist of single domain crystals yielded γ-values near 1.4–1.6, whereas the multidomainic crystal preparations yielded values of 1.0–1.1. The parameter γ collects the effects of factors, such as the crystal geometry, the particle size distribution and the reactive site density. The relative importance of these factors was evaluated and particularly the particle size distribution appears to be of importance for iron oxides.

THE REACTIVITY OF IRON OXIDES IN SEDIMENTS : A KINETIC APPROACH

A kinetic approach is presented that allows a quantitative description of the reactivity of ferric oxides for both synthetic polydisperse ferrihydrite and assemblages of ferric oxides found in natural sediments. Results from reductive dissolution experiments with ascorbic acid at pH 3, and ferrihydrite or ferric oxide contained in sediment are given. The rate of ferrihydrite dissolution was found to be proportional to the initial amount of ferrihydrite in solution, nearly independent of the ascorbic acid concentration within the range 5 to 15 mM, and strongly dependent on the mineral fraction dissolved. The reactivity of ferrihydrite could be described as J/m0 = 4 · 10−4 (m/m0)1.1, where J is the overall rate of dissolution (mol/s), m0 the initial amount of ferrihydrite, and m/m0 the undissolved mineral fraction. Assemblages of ferric oxides in natural sediments can be described as a reactive continuum with a Gamma distribution of ferric oxide reactivity types. This leads to the same rate law as used to describe ferrihydrite dissolution, although the rate constant and exponential term now only have statistical significance. The reactivity of ferric oxides of an oxidized marine sediment from the Bight of Aarhus is then described by J/m0 = 7.4 · 10−3(m/m0)4.7, and J/m0 = 5.3 · 10−5 (m/m0)2.75 was found for an oxidized aquifer sediment from the island of Romo. Thus, initial reduction rates varied by more than two orders of magnitude, while the change in reactivity during dissolution of 90% of the ferric oxides present may be almost five orders of magnitude. Chemical extraction techniques for reactive iron cover over a wide range in reactivity.

Concentration of Mn and separation from Fe in sediments—I. Kinetics and stoichiometry of the reaction between birnessite and dissolved Fe(II) at 10°C

Abstract Redox reactions between Fe2+ in solution and Mn-oxides are proposed as a mechanism for concentration of Mn in sediments both during weathering and diagenesis in marine sediments, e.g. the formation of Mn-nodules. If such a mechanism is to be effective, then reaction rates between Fe2+ and Mn-oxides should be fast. The kinetics and stoichiometry of the reaction between dissolved Fe2+ and synthetically prepared birnessite (Mn7O13·5H2O) were studied experimentally in the pH range 3–6. Results show a stoichiometry which at pH 4 FeOOH is precipitated and excess Fe2+ consumption compared to the theoretical stoichiometry is observed. The excess Fe2+ consumption is not due to a formation of a quantitative MnOOH layer but rather to adsorption. Reaction kinetics are very fast at pH 4 the reaction is fast initially until 17% of the bimessite has dissolved and changes then to a slower stage. The later stage can be described by the equation: J = km 0 (H + ) −0.45 [Fe 2+ ] γ ( m m 0 ) β where J is the overall rate of Mn2+ release, m0 and m the mass of birnessite at time t = 0 and t > 0, β = 6.76−0.94 pH and γ has values of 0.76 at pH 5 and 0.39 at pH 6. The rate constant k is 7.2·10−7 moles s−1 g−1 (moles/1)−0.31 at pH 5 and 9.6·10−8 moles s−1 g−1 (moles/1)0.06 at pH 6. Diffusion calculations show that the rate is controlled by surface reaction and it is tentatively proposed that the availability of vacancies in octahedral [MnO6]sheets of the birnessite surface could be rate controlling. It is concluded that reactions between Fe(II) and birnessite, and probably other Mn-oxides, are fast enough to be important in natural environments at the earth surface.

Reduction of Mn-oxides by ferrous iron in a flow system: column experiment and reactive transport modeling

Abstract The reduction of Mn-oxide by Fe 2+ was studied in column experiments, using a column filled with natural Mn-oxide coated sand. Analysis of the Mn-oxide indicated the presence of both Mn(III) and Mn(IV) in the Mn-oxide. The initial exchange capacity of the column was determined by displacement of adsorbed Ca 2+ with Mg 2+ . Subsequently a FeCl 2 solution was injected into the column causing the reduction of the Mn-oxide and the precipitation of Fe(OH) 3 . Finally the exchange capacity of the column containing newly formed Fe(OH) 3 was determined by injection of a KBr solution. During injection of the FeCl 2 solution into the column, an ion distribution pattern was observed in the effluent that suggests the formation of separate reaction fronts for Mn(III)-oxide and Mn(IV)-oxide travelling at different velocities through the column. At the proximal reaction front, Fe 2+ reacts with MnO 2 producing Fe(OH) 3 , Mn 2+ and H + . The protons are transported downstream and cause the disproportionation of MnOOH at a separate reaction front. Between the two Mn reaction fronts, the dissolution and precipitation of Fe(OH) 3 and Al(OH) 3 act as proton buffers. Reactive transport modeling, using the code PHREEQC 2.0, was done to quantify and analyze the reaction controls and the coupling between transport and chemical processes. A model containing only mineral equilibria constraints for birnessite, manganite, gibbsite, and ferrihydrite, was able to explain the overall reaction pattern with the sequential appearance of Mn 2+ , Al 3+ , Fe 3+ , and Fe 2+ in the column outlet solution. However, the initial breakthrough of a peak of Ca 2+ and the observed pH buffering indicated that exchange processes were of importance as well. The amount of potential exchangers, such as birnessite and ferrihydrite, did vary in the course of the experiment. A model containing surface complexation coupled to varying concentrations of birnessite and ferrihydrite and a constant charge exchanger in addition to mineral equilibria provided a satisfactory description of the distribution of all solutes in time and space. However, the observed concentration profiles are more gradual than indicated by the equilibrium model. Reaction kinetics for the dissolution of MnO 2 and MnOOH and dissolution of Al(OH) 3 were incorporated in the model, which explained the shape of the breakthrough curves satisfactorily. The results of this study emphasize the importance of understanding the interplay between chemical reactions and transport in addition to interactions between redox, proton buffering, and adsorption processes when dealing with natural sediments. Reactive transport modeling is a powerful tool to analyze and quantify such interactions.

A consistent model for surface complexation on birnessite (−MnO2) and its application to a column experiment

Abstract Available surface complexation models for birnessite required the inclusion of bidentate bonds or the adsorption of cation-hydroxy complexes to account for experimentally observed H+/Mm+ exchange. These models contain inconsistencies and therefore the surface complexation on birnessite was re-examined. Structural data on birnessite indicate that sorption sites are located on three oxygens around a vacancy in the octahedral layer. The three oxygens together carry a charge of −2, i.e., constitute a doubly charged sorption site. Therefore a new surface complexation model was formulated using a doubly charged, diprotic, sorption site where divalent cations adsorbing via inner-sphere complexes bind to the three oxygens. Using the diprotic site concept we have remodeled the experimental data for sorption on birnessite by Murray (1975) using the surface complexation model of Dzombak and Morel (1990) . Intrinsic constants for the surface complexation model were obtained with the non-linear optimization program PEST in combination with a modified version of PHREEQC (Parkhurst, 1995) . The optimized model was subsequently tested against independent data sets for synthetic birnessite by Balistrieri and Murray (1982) and Wang et al. (1996) . It was found to describe the experimental data well. Finally the model was tested against the results of column experiments where cations adsorbed on natural MnO2 coated sand. In this case as well, the diprotic surface complexation model gave an excellent description of the experimental results.

Kinetics of nitrate reduction by detrital Fe(II)-silicates

Abstract The ability of Fe(II)-bearing minerals to reduce nitrate was investigated experimentally in order to asses their potential for nitrate removal in aquifers. Experiments were carried out with a fluidized bed reactor, using arfvedsonite as an example for amphiboles and augite for pyroxenes. Results show that both Fe(II)-bearing silicates are able to reduce nitrate at low rates in the pH range 2 to 7. For arfvedsonite a maximum reduction rate was found around pH 4, while at higher values a pH independent rate of 4 · 10 −17 N mol/cm 2 · sec (25°C) is found. Nitrate reduction rates for augite are on the same order of magnitude. The mechanism appears to be complex; apparently, it is not a direct reaction between nitrate and the dissolving mineral surface, but rather nitrate seems to react with secondary products of silicate dissolution. The most plausible explanation is that freshly precipitated FeOOH catalyzes nitrate reduction by Fe 2+ , as has been reported from other studies. A rough estimate for sandy aquifers indicates that Fe(II)-bearing silicates should be able to reduce nitrate at a rate on the order of magnitude of 4 · 10 −5 NO 3 mol/1 · a, and they can be of importance in aquifers with long groundwater residence times or low nitrate loads.

Formation and solid solution behavior of Ca-rhodochrosites in marine muds of the Baltic deeps

Authigenic Ca-rhodochrosites are found in organic-rich sediments in the deep anoxic basins of the Baltic Sea. Rhodochrosite formation is apparently the result of organic matter degradation. The rhodochrosite contains 10 to 40 mol% CaCO3 and 2 to 5 mol% MgCO3, as determined by microprobe. The composition of the carbonates at a given depth varies by about 5 mol% MnCO3. SEM revealed the rhodochrosite to be present as globular aggregates consisting of microcrystallites. Dissolution features such as cavities within the aggregates are frequently observed and indicate that extensive recrystallization takes place. Pore waters are greatly supersaturated with respect to both rhodochrosite and calcite. Rhodochrosite is only found in sediments where pore waters show the highest degree of supersaturation. Due to ion exchange, Ca2+ diffuses from the underlying freshwater clays. This results in increasing aCa2+aMn2+ ratios in the pore waters from the sediment surface downward. In response to this, the maximum CaCO3MnCO3 ratio also increases with depth. The composition of rhodochrosite which forms under these conditions may be determined by both the porewater aCa2+aMn2+ and precipitation kinetics. However, if their composition is controlled solely by the aCa2+aMn2+ in solution, their behavior can be described by a regular solid solution model.