Environ. Sci. Technol. 2005, 39, 7915-7920
Kinetic Studies of Synthetic Metaschoepite under Acidic Conditions in Batch and Flow Experiments OLGA RIBA,* COLIN WALKER,† AND K. VALA RAGNARSDOTTIR Department of Earth Sciences, University of Bristol, Queen’s Road, Bristol, BS8 1RJ, United Kingdom
The weathering and corrosion of depleted uranium (DU) forms a complex series of oxidation reactions, ultimately resulting in metaschoepite, UO3‚2H2O. The present work focused on studying the dissolution rate of synthetic UO3‚ 2H2O using batch and flow-through reactors. Under acidic conditions (pH ) 4.4-5.4), atmospheric CO2, room temperature, and 0.1 m ionic strength, the log solubility product, log Ksp ) 5.26 at equilibrium and a pH-dependent rate law R0 ) (0.30 ( 0.15)[H+]0.83(0.1 were established. For consistency, these results were incorporated into the computer program PHREEQC 2.6, and the experimental conditions were simulated. There is generally good agreement between the experimental results and the modeled results. Batch experiments revealed a fast dissolution rate of UO3‚2H2O in the first hour, followed by fluctuations in uranium concentration before equilibrium was attained after 3000 h.
Introduction Depleted uranium (DU), a byproduct of uranium enrichment, has been extensively used in military operations over the past decade because of its armor-piercing capability. DU is a radioactive material with a long half-life (4.5 × 109 years), which makes it problematic because of its chemical toxicity as a heavy metal and its radioactivity, which will become increasingly important with time due to the formation of daughter isotopes. In 2000 the first international assessment on the environmental behavior of DU was carried out by UNEP in Kosovo (1). The study revealed that approximately 10% of the fired DU rounds had hit their designated target, and the remainder was buried in the ground. Seven years after the conflict in Bosnia-Herzegovina, the DU penetrators recovered by UNEP’s team lost approximately 25% of their mass (2). Fast decomposition of the DU penetrators occurs because of the high chemical reactivity of metal uranium under oxidizing conditions, promoting the surface formation of uraninite (UO2). UO2 has been the focus of a large number of thermodynamic and kinetic studies (3-7). Under dry oxidizing conditions UO2 generates several uranium oxides of variable stoichiometry (e.g., U4O9, U3O7, and U3O8) up to * Corresponding author phone: +44 (0)117 331 1176; fax: +44 (0)117 925 5646; e-mail:
[email protected]. Present address: Interface Analysis Centre, University of Bristol, Oldbury House, 121 St. Michael’s Hill, Bristol, BS2 8BS, U.K. † Dental Biophysics, Queen Mary College, Mile End Road, London, E1 4NS, U.K. 10.1021/es050678k CCC: $30.25 Published on Web 09/10/2005
2005 American Chemical Society
UO3, where all of the uranium atoms are hexavalent (8, 9). In the near surface environment UO3 readily hydrates to form schoepite and schoepite-like phases (UO3‚nH2O). These are considered the most important solubility-limiting phases in the dissolution of uranium oxide, found in soils at the Fernald Environment Management Site (10, 11), Oak Ridge National Laboratory (12), and Kosovo (1). Schoepite (UO3‚ 2.25H2O) transforms slowly under ambient conditions to the more stable metaschoepite (UO3‚2H2O) (13). As reviewed by K. V. Ragnarsdottir and L. Charlet (14) and also shown in recent studies of natural systems, carbonate content, redox characteristics, organic matter, biological activity, Fe/Mn oxyhydroxide minerals and silicates forming part of the soil, and soil solution can also be locally significant (15-18); these variables are not considered in this study. The dissolution reaction of UO3‚2H2O under acid conditions can be represented by
UO3‚2H2O + 2H+ ) UO22+ + 3H2O
(1)
The log value of the solubility product, Ksp, of this reaction is given as 4.81, calculated using the suggested free energies of the NEA-TDB (19). This value has been shown to vary from 4.70 to 6.33 (20) as a function of the degree of crystallinity, particle size, electrolyte concentration, pH range, and working atmosphere. Despite the fact that the thermodynamic equilibrium of metaschoepite has been extensively studied, there is a lack of reported data concerning its dissolution kinetics. The aim of this work was to study the dissolution of UO3‚ 2H2O far and close to thermodynamic equilibrium using batch experiments and, using an appropriate flow-through reactor, establish a kinetic rate law to describe the dissolution of metaschoepite under acidic conditions. The derived rate law was checked for consistency by simulating the experimental data using the geochemical computer program PHREEQC 2.6 (21).
Experimental Section Materials. All chemicals used in this study, UO2(NO3)2‚6H2O (s), NaOH (s), NaClO4 (s), HCl, and HNO3, were of analytical grade. All solutions were prepared with a Milli-Q water purifying system, resistivity greater than 18.2 MΩ cm. A batch of metaschoepite was synthesized following an adaptation of published methods (22, 23). A 0.2 M UO2(NO3)2‚ 6H2O solution was continuously stirred and slowly titrated with 1 M NaOH under a nitrogen atmosphere. At pH 4 the product started precipitating, and the solution was further titrated until pH 6. The yellow suspension was aged for 14 days and then rinsed by repeated centrifugation, decantation, and resuspension in deionized water. The product was sonicated in water for 20 min and centrifuged at 14 × 103 rpm for 10 min before it was dried at 50 °C for 24 h. The solid was then ground and sieved to provide a starting material with a grain size of 50-100 µm. The surface area of the synthesized schoepite, 4.51 ( 0.4 m2 g-1, measured on a freeze-dried powder sample was determined by BET-Ar adsorption. Surface area measurements were only taken prior to experimental runs as required by eq 2. Analytical Techniques. The solid samples were characterized before and after experimental runs by X-ray diffraction (XRD), infrared spectroscopy (FT-IR), and environmental scanning electron microscopy (E-SEM) (24, 25). XRD analyses were performed on a Phillips Xpert Proc diffractometer with a Cu KR1 (λ ) 1.5406 Å) radiation source, generator voltage of 40 keV, and tube current of 30 mA. Scans used a step size VOL. 39, NO. 20, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7915
mass spectrometry (ICP-MS) on a Thermo Elemental PQ3 using 10 ppb Bi as an internal standard. The hydrogen-ion concentration was measured with a Metrohm Combined Glass Electrode with “Long Life” reference system (Ag/AgCl) calibrated with DIN/NBS buffer solutions. Batch Reactor. Reactors of HDPE were used for the batch experiments in equilibrium with atmospheric CO2 at room temperature. Approximately 0.5 g of UO3‚2H2O was added to the reactor vessel containing 0.40 L of 0.1 M NaClO4, and solution pH was adjusted to 4.39, 4.55, 4.75, 4.95, or 5.44, giving five datasets from far to near equilibrium conditions. The pH was monitored and manually adjusted with 0.5 m HClO4 whenever the measure drifts more than (0.04 pH units from the initial value (Figure 2 shows the periodicity of pH monitoring together with its value). This process was continued to give a total experimental run time of 4000 h. Aliquots of solution for uranium analysis were taken from the reactor at timed intervals and filtered through 0.22 µm cellulose acetate prior to ICP-MS analyses. FIGURE 1. Scanning electron micrographs of filtered and dried materials before (a), (b) and after (c), (d) 4000 h of reaction of metaschoepite in aqueous solution at pH 5.44. of 0.02° with a scan speed of 2 s per step. Scans were collected from 10° to 50° 2θ with a step size of 0.02° and a scan speed of 2 s per step. FT-IR spectra were acquired in the midregion (400-5500 cm-1) using a Nicolet Nexus FTIR spectrometer. A Globar ceramic beam source, KBr beam splitter, and DTGS detector were used at a resolution of (4 cm-1 with accumulation of 256 scans. Samples were pressed into KBr pellets and purged with dry air prior to collecting each spectrum. The E-SEM used to collect the micrographs of uncoated samples was an ElectroScan instrument with a gaseous secondary electron detector. Solutions were acidified using 10% nitric acid and analyzed for uranium concentrations by inductively coupled plasma
Flow-Through Reactor. To operate in a steady-state mode and avoid measuring the kinetics through fine particles or highly reactive sites, a second set of experiments was performed in a plug flow reactor (26). A thin layer of solid was used in order to minimize diffusion problems and preferential solution pathways and ensure optimum contact between the solid and liquid phases (4, 5, 27). A 0.1 g amount of UO3‚2H2O was sandwiched between two 0.22 µm cellulose nitrate filters 47 mm in diameter and held in a reaction vessel through which the feed solution was pumped continuously by means of a peristaltic pump. The pH of the feed solution was manually adjusted to 4.56, 4.72, 4.95, 5.25, 5.40, 5.61, 5.80, and 5.95 by adding small amounts of 0.1 M perchloric acid or sodium hydroxide solution whenever necessary. Sodium perchlorate (NaClO4) was used as an electrolyte to maintain the ionic strength at 0.1 m for all experiments.
FIGURE 2. Uranium concentration (mM), pH value, and modeling data plotted as a function of time in batch experiments: set up at pH 4.39, 4.55, and 4.75 for (a) 60 min, (b) 600 h, and (c) 3000 h and at pH 4.95 and 5.44 for (d) 60 min, (e) 600 h, and (f) 3000 h. Errors of the uranium concentrations are too small to be plotted; standard deviations for all measurements are of the order of 0.01. 7916
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 20, 2005
The flow rate of the feed solution was maintained between 0.30 and 0.56 dm3 h-1 for the experiments run at pH e 5.40 and between 0.17 and 0.33 dm3 h-1 for experiments run at pH g 5.61. This undertaking was to ensure the effluent solution was undersaturated with respect to metaschoepite. The flow was measured by weighing the output solutions during a controlled interval of time. The absence of particles in the samples was assured by additional filtering with 0.22 µm cellulose acetate prior to any analysis. Once the pH of the outlet solution was constant for several hours, therefore indicative of steady-state conditions, a new experiment was started using the same solid material. At the end of each experiment simple mass balance calculations determined the amount of solid remaining in the reactor. Geochemical Modeling. The consistency of the experimental data from the batch and flow-through experiments was checked with the computer program PHREEQC 2.6 using the database released by Lawrence Livermore National Laboratory (revision 1.11) with formation constants for the uranyl dissolved species from the NEA database (19). Constants are extrapolated to zero ionic strength using ion association and Debye-Hu ¨ ckel expressions as part of the chemical equilibrium calculations (21). The rate equation was modified from nonlinear transitionstate theory (28, 29), where terms are included to account for the initial surface area and changing concentration of metaschoepite, given by
(
)
m IAP rate ) Kr A 1m0 Ksp
(2)
where Kr is the constant rate (mol m-2 s-1), A is the total surface area of metaschoepite used in the experiment (m2), m is the current number of moles of metaschoepite (mol), m0 is the initial number of moles of metaschoepite (mol), and IAP is the ionic activity product. The rate is therefore approximately equal to the product of the first two terms of the rate equation, eq 2, when the system is super-undersaturated, very far from equilibrium, and zero at equilibrium.
Results and Discussion Batch Experiments. The scanning electron micrographs of metaschoepite before reaction (Figure 1a) show plate-like crystals that are approximately 3 µm across, which aggregate to form a heterogeneous, almost botryoidal-like texture (Figure 1b). After 4000 h of reaction at pH 5.44, the plate-like crystals are still identifiable but the aggregates are more homogeneous (Figure 1c and 1d). The results of the batch experiments obtained using different hydrogen-ion concentrations are depicted in Figure 2 together with the results obtained from the geochemical modeling using PHREEQC 2.6. The experimental data shows common trends among the five experiments that can be summarized as follows. (a) There is a fast linear increase of uranium concentration in solution for the first 20 min of the experiment. (b) Uranium concentrations follow an exponential increase that reach a steady state (i.e., no change in uranium concentration) after 1 h for experiments at pH 4.55 and 4.75 and 10 h for experiments at pH 4.39, 4.95, and 5.44. (c) After 100 h a series of ‘steps’ was marked out by changes in uranium concentrations, different in both magnitude and direction, either an increase or decrease in uranium concentration, depending on the hydrogen-ion concentration of the respective experiments. The positions of the changes with respect to time are remarkably consistent for all experiments. Decreases in uranium concentrations were only observed for experiments run at pH 4.39, 4.95, and 5.44 and only after 250-450 h. All experiments showed a rapid increase
TABLE 1. Experimentally Determined Values of logKsp of Synthetic Metaschoepite experimental conditions ref
electrolyte
pH
present work 36 20 35 37 38 39 40
0.1 M NaClO4 0.1 M NaNO3 none 0.5 M NaClO4 0.1 M NaClO4 0.1 M NaClO4 0.1 M NaClO4 0.1 M NaClO4
4.4-5.4 6 5.5-6.1 4.7, 6.3 3.8-7 4.5-5.5 4.07-5.92 2.8-4.6
atmosphere ambient ambient ambient argon ambient nitrogen ambient ambient
log Ksp (I ) 0) 5.13-5.33 5.02-5.19 5.38-5.46 4.70 5.14 5.13 5.53 5.02
in uranium concentration from 1200 to 1500 h and again from 2500 to 3000 h. (d) After 3000 h a steady state was maintained where the uranium concentration remained constant and the hydrogenion concentration was equal to the initial pH value. This was assumed to indicate thermodynamic equilibrium. XRD traces and FTIR spectra were taken for all experiments and confirmed no structural alteration of the metaschoepite and that no secondary phases, by incorporation of Na+ into the interlayers of the metaschoepite structure, were precipitated (20, 30-33). Therefore, the observed changes in solution chemistry are solely due to the dissolution of metaschoepite. This was also confirmed by Dı´az Arocas and Grambow (34), who conducted similar experiments between pH 4 and 6 in 0.5 m NaClO4 solutions. The fluctuations in uranium concentration with time can be tentatively explained by the result of systems approaching thermodynamic equilibrium and going through a steady state and subsequent dynamic behavior of dissolution and reprecipitation processes, which are also affected by the undertaken pH adjustments. These processes persist until the pH remains constant after 3000 h. Similar fluctuations were also observed by Dı´az Arocas (35) in her study of the influence of precipitation kinetics on the solubility and degree of crystallinity of U(VI) phases. Using the data from these five experiments it was possible to derive log Ksp values. The log Ksp only varies from 5.13 to 5.33 with an average of 5.26 and standard deviation of 0.08. This value is in good agreement with other log Ksp values reported in the literature under similar experimental conditions, shown in Table 1. The dissolution rate, R0, of metaschoepite was calculated from the slope of the initial linear segment (far from equilibrium) of the plot of the uranium concentration versus time (from Figure 2) and normalized with the surface area. Values of pR0 are plotted versus their respective pH in Figure 3. The slope of the best-fit line was used to determine the reaction order of 1.39 ( 0.01. Reaction orders less than or equal to 1 have been previously determined for other dissolving solids since most of the dissolution reactions are critically dependent on the coordinative interactions taking place on the surface (41, 42). However, this does not apply to metaschoepite because the dissolution reaction is very rapid. Instead, using a batch reactor, the rate is expected to be controlled by the diffusion of U(VI) species away from the mineral surface into the bulk of the solution. The rate equation, eq 2, was used to model the variation of uranium concentration as a function of time for each experimental run, and this was plotted alongside the experimental data in Figure 2. It was evaluated using the dissolution rates, R0 (shown in Figure 3), as Kr and the experimentally derived Ksp (shown in Table 1). The measured U concentration and model prediction are congruent in the early stages of dissolution (
