Environ. Sci. Technol. 2003, 37, 4642-4649
Molecular-Scale Characterization of Uranium Sorption by Bone Apatite Materials for a Permeable Reactive Barrier Demonstration C . C . F U L L E R , * ,† J . R . B A R G A R , ‡ A N D J. A. DAVIS† U.S. Geological Survey, 345 Middlefield Road, MS 465, Menlo Park, California 94025, and Stanford Synchrotron Radiation Laboratory, Stanford, California 94309
Uranium binding to bone charcoal and bone meal apatite materials was investigated using U LIII-edge EXAFS spectroscopy and synchrotron source XRD measurements of laboratory batch preparations in the absence and presence of dissolved carbonate. Pelletized bone char apatite recovered from a permeable reactive barrier (PRB) at Fry Canyon, UT, was also studied. EXAFS analyses indicate that U(VI) sorption in the absence of dissolved carbonate occurred by surface complexation of U(VI) for sorbed concentrations e 5500 µg U(VI)/g for all materials with the exception of crushed bone char pellets. Either a split or a disordered equatorial oxygen shell was observed, consistent with complexation of uranyl by the apatite surface. A second shell of atoms at a distance of 2.9 Å was required to fit the spectra of samples prepared in the presence of dissolved carbonate (4.8 mM total) and is interpreted as formation of ternary carbonate complexes with sorbed U(VI). A U-P distance at 3.5-3.6 Å was found for most samples under conditions where uranyl phosphate phases did not form, which is consistent with monodentate coordination of uranyl by phosphate groups in the apatite surface. At sorbed concentrations g 5500 µg U(VI)/g in the absence of dissolved carbonate, formation of the uranyl phosphate solid phase, chernikovite, was observed. The presence of dissolved carbonate (4.8 mM total) suppressed the formation of chernikovite, which was not detected even with sorbed U(VI) up to 12 300 µg U(VI)/g in batch samples of bone meal, bone charcoal, and reagent-grade hydroxyapatite. EXAFS spectra of bone char samples recovered from the Fry Canyon PRB were comparable to laboratory samples in the presence of dissolved carbonate where U(VI) sorption occurred by surface complexation. Our findings demonstrate that uranium uptake by bone apatite will probably occur by surface complexation instead of precipitation of uranyl phosphate phases under the groundwater conditions found at many U-contaminated sites.
Introduction Development of effective strategies for remediation of groundwater contamination is needed for protection of * Corresponding author phone: (650)329-4479; fax: (650)329-4545; e-mail:
[email protected]. † U.S. Geological Survey. ‡ Stanford Synchrotron Radiation Laboratory. 4642
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 20, 2003
drinking water supplies. Emerging passive remediation technologies such as permeable reactive barriers (PRBs) require materials that are efficient in removing dissolved contaminants through degradation, precipitation, or sorption of contaminants (1). Because the long-term stability of sequestered contaminants is dependent on their chemical state, investigation of contaminant removal processes under relevant geochemical conditions is critical to the design, operation, and evaluation of remediation devices (2). For example, zerovalent iron, ZVI, has been deployed in PRBs in several uranium-contaminated aquifers based on a reduction and precipitation mechanism for U(VI) removal (1, 3, 4). However, the potential for clogging by mineral precipitation and passivation by formation of corrosion products (3, 4) may compromise the effectiveness of ZVI for U(VI) removal by reductive precipitation. In addition, U(VI) adsorbed to ZVI corrosion products (5), such as ferrihydrite (6), could be re-released by desorption if uncontaminated water enters the PRB. These uncertainties in the U(VI) removal processes by ZVI suggest the need for comparison with alternative PRB materials. Interest in the use of apatite for metal remediation has developed based on the low solubility of metal phosphates that form by interaction with phosphate (P) released from apatite (e.g., refs 7-9). Proposed use of apatite for groundwater uranium remediation is based upon reported low solubility of uranyl phosphates (10, 11). High levels of uranium sorption have been reported for both synthetic hydroxyapatite and bone apatite (12-15), although the sorption mechanism has not been evaluated under conditions relevant to PRB applications. Because of their effectiveness in metal stabilization, bone and phosphate rock apatite have been proposed as lower-cost alternatives to synthetic hydroxyapatite in designing remediation strategies (16, 17). In this paper, we present analysis of extended X-ray absorption fine structure (EXAFS) and synchrotron source X-ray diffraction (SR-XRD) measurements to characterize the mechanism of U(VI) sorption by bone meal and bone charcoal apatites. U(VI) sorption by these materials was studied in conjunction with a PRB field demonstration to evaluate insitu sequestration of groundwater uranium at Fry Canyon, UT (18). Samples were prepared in the laboratory by reaction with U(VI) both in the presence and absence of dissolved carbonate. In addition, bone char pellets recovered from the PRB were analyzed. The EXAFS and SR-XRD results reported here build on findings for U(VI) sorbed by reagent-grade hydroxyapatite, HA (19), by investigating U(VI) sorbed on economic commercial alternatives to synthetic HA, such as bone meal and bone charcoal apatite. The previous results are used here as a model to interpret these new data. Fry Canyon PRB Field Demonstration. Groundwater at the Fry Canyon, UT site (pH 7; 4.8 mM total carbonate, [CO3]T) is contaminated with up to 85 µM dissolved U(VI), [U], that has leached from tailings at an abandoned ore upgrader facility. A pelletized bone char apatite PRB was installed within the groundwater uranium plume in 1997 as part of a field demonstration to determine the feasibility and longevity of PRBs for remediation (18). Batch uptake and column tests were conducted to evaluate U(VI) sorption by phosphate rock, bone meal, and bone charcoal for PRB use (20). U(VI) uptake by bone meal and bone charcoal was 1.5-2 orders of magnitude greater than by phosphate rock in both batch and column experiments. The pellitized bone charcoal was chosen for the PRB demonstration because of its sorption capacity and permeability. Reversibility tests showed that 65-80% of U(VI) sorbed in column experiments for all 10.1021/es0343959 CCC: $25.00
2003 American Chemical Society Published on Web 09/11/2003
TABLE 1. Physical Characteristics of Bone Apatite Materials material
ID
bone meal, fertilizer grade
BB1
bone-black charcoal
BK
bone char pellet
CP3
bone char pellet
CP5
source Dale Alley Company St. Joseph, Missouri EM Scientific
surface area (m2/g)a 6.6
grain-sizeb 100% < 63 µm
64
3-15 µm
Cercona of America, Inc. Dayton, Ohio
33
Cercona of America, Inc. Dayton, Ohio
44
7.5% > 4 mm 65.3% > 2-4 mm 23.7% > 1-2 mm 3.5% < 1 mm 13.8% > 4 mm 42.9% > 2-4 mm 39.4% > 1-2 mm 3.8% < 1 mm
composition (weight percent)c
phases identifiedd
∼70% apatite ∼30% organic 88% Ca3(PO4)2 12% C 74-76% Ca3(PO3)2 10-12% C 12-15% Al2O3, P2O5e
HA
74-76% Ca3(PO3)2 10-12% C 12-15% Al2O3, P2O5e
WHT
HA WHT
a Surface area determined by single-point N absorption. b Grain size determined by sieving of dry material. Grain size of bone charcoal provided 2 by manufacturer. c Compositional analysis from manufacturer. d Phases identified by sealed-tube XRD. HA hydroxyapatite (Ca5(PO4)3OH); WHT e whitlockite β-Ca3(PO4)2. Binder from pelletizing process.
materials was released during elution with U-free artificial groundwater, with about half of the uranium release occurring slowly over hundreds of pore volumes (20). Previous Study of Uranium Interaction with Synthetic Hydroxyapatite. Reagent-grade synthetic hydroxyapatite (HA) was highly effective in lowering dissolved uranium concentration in batch uptake experiments with near complete removal to 7000 µg U(VI)/g. At sorbed concentrations lower than this value chernikovite was not detected. Instead, U(VI) sorption was found to be controlled by formation of inner-sphere complexes of the uranyl oxocation with the HA surface. This behavior was found to occur even when the initial solution conditions exceeded the apparent chernikovite solubility product. Ternary complexes of carbonate with the uranyl-HA surface complex were observed in the presence of [CO3]T similar to the Fry Canyon groundwater.
Methods Materials. Commercially available bone apatites studied included fertilizer grade bone meal (BB1) and powdered reagent-grade bone-black charcoal (BK1). These materials contained hydroxyapatite as indicated by sealed-tube XRD analyses. Two highly porous, pelletized bone charcoal prototypes (CP3 and CP5) manufactured by Cercona of America, Inc. (Dayton, OH) by calcining fertilizer grade bone meal in the absence of air at 1100 °C in the presence of an aluminum phosphate binder also were studied. The XRD pattern of the bone char pellets indicated the presence of whitlockite, a dehydrated form of apatite that likely formed from the bone meal apatite when calcined. All materials were used as received. The physical characteristics of these materials are presented elsewhere (20) and summarized in Table 1. Bone char pellets were chosen for the PRB demonstration based on a laboratory study of the CP3 pellets. The CP5 pellets were subsequently produced in sufficient quantity for the PRB demonstration project. The CP3 and CP5 differed in bone meal source and firing conditions (20). Sample Preparation. U(VI)-reacted bone meal, bone charcoal, and bone char pellets were prepared in batch suspensions for SR-XRD and EXAFS measurement by the methods used for HA samples (19) summarized in the following paragraphs. Sample conditions (Table 2) were chosen to provide a range of sorbed U(VI) concentrations to evaluate U(VI) sorption mechanisms at the [CO3]T found in Fry Canyon groundwater and to test for precipitation of uranyl
phosphates. Samples were also prepared with no added carbonate and in the absence of CO2. One-liter suspensions of bone apatites were prepared from the dry solid and either in artificial Fry Canyon groundwater (AGW) or in 0.1 M NaNO3 solutions. The suspension pH was adjusted to 7 and preequilibrated for 24 h. A 0.01 M uranyl nitrate solution was then added incrementally to obtain the desired total U(VI) concentration, [UT], and the pH was readjusted to 7. Samples were equilibrated for 48-120 h on a shaker table. No significant change in U(VI) sorption was observed after 24 h in batch tests (20). Samples were prepared in 1-L polyethylene bottles that were open to the atmosphere during uranium additions, pH adjustment, and recovery of solid, with the exception of CO2-free samples (CP3-3, BK1-1). No attempt was made to control pH after the initial pH adjustment following U(VI) addition in order to evaluate the pH response of the system. Following U(VI) sorption, the apatite solid was recovered by settling and centrifugation, and an aliquot of supernatant was retained for U, Ca, and P analyses. The recovered solid was then loaded into SR-XRD and EXAFS sample holders. Batch samples CP3-3 and BK1-2 were prepared in 0.1 M NaNO3 in the absence of dissolved carbonate and CO2 in Teflon-lined, sealed glass reaction vessels with continuous purging with CO2-free air. Solid recovery and mounting for SR-XRD and EXAFS was performed in an argon-filled glovebag. The CP3 material (samples CP3-3, CP3-4) was ground prior to U(VI) sorption to allow packing into sample holders in the absence of CO2. A sample of the CP5 bone char pellets, the formulation used in the PRB, was reacted with uranium in a flowing, packed column with an AGW influent containing 50 µM [UT]. The column apparatus and operation is described in ref 20. After complete (100%) breakthrough of uranium, the CP5 solid was ground for SR-XRD measurement. Total uranium uptake was determined by integrating the column effluent breakthrough curve. EXAFS spectra were collected on two samples of the CP5 bone char pellets that were recovered during coring of the upgradient side of the PRB in May 1998 after 18 months of operation. The coring and subsampling procedure is described elsewhere (20). The sorbed uranium concentration of the PRB apatite was determined using a 0.1 M HNO3 extraction. Sample Measurement and Analysis Procedures. Dissolved U(VI) was measured by kinetic phosphorescence (mdl 0.002 µM, precision (2%). U(VI) removal was determined from the difference between [UT] and [U] after equilibration. Dissolved Ca and P were determined by ICP-OES (mdl 0.2 VOL. 37, NO. 20, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4643
TABLE 2. Composition of U(VI) Reacted Apatite Samples Used for EXAFS and SR-XRD Measurements final dissolved concns sample ID
solid (g/L)
solutiona
total U(VI) U uptake uptake added (µM) (µg U/g solid) pH (%)
BK1-2 BK1-1 BK-0106 BK1-3
4 10 1 2
N, no CO2 AGW AGW N
50 250 85 50
BB1-A BB1-B BB1-C BB1-0106
2 1 2 1
N N AGW AGW
25 50 25 60
CP3-3 CP3-4 CP3-5 CP3-0106
2.5 2.5 2.5 2
N, no CO2 AGW N AGW AGW GW GW
CP5-COL column b PRB R2-3 b PRB R2-2 b HA-0106
1.35 AGW
[U] (µM)
Bone-Black Charcoal Powder 2960 8.5 >99 0.2 5530 7.3 93 18 8800 7.7 43 48 5460 9.3 92 4.1
[P] (mM)
[Ca] EXAFS spectra SR-XRD phases (mM) collectedb identifiedb,c
0.01 0.02 0.02 99 0.03 2670 7.0 56 22 4740 8.4 >99 0.18 1180 7.8 20 40
0.33 0.03 0.06 0.04
0.68 3.51 0.16 3.67
X X b X
WHT, CHN b WHT b
50 6-10d 6-10d
Pelletized Bone Charcoal (CP5) 1250 7.0 b b 690 7.0 b 0.71 190 7.0 b 0.71
b 9.1 9.1
b 6.74 6.74
b X X
WHT b b
0.047
0.46
X
HA
75
Bone meal 2970 6.4 11850 6.4 2540 6.8 11200 7.0
>99 >99 76 78
Reagent-Grade Hydroxyapatite Powder 12300 7.5 92 6.3
a
Background electrolyte: N 0.1 M NaNO3; AGW artificial Fry Canyon groundwater (3.9 mM Ca, 12.4 mM Na, 2.4 mM Mg, 0.14 mM K, 6.8 mM SO4, 6.8 mM Cl, 4.8 mM carbonate); GW groundwater at Fry Canyon (see ref 18). b Data not collected . c HA, hydroxyapatite; WHT, whitlockite; CHN, chernikovite. d Range of dissolved uranium measured in samples from PRB wells adjacent to core samples.
and 30 µM, respectively). P measured by ICP-OES is assumed to represent dissolved orthophosphate species. Fluorescent-yield U-LIII EXAFS spectra of wet U(VI)reacted HA samples were collected at SSRL beam lines 4-1, 4-3, and 11-2 (Si (220) and (111) monochromator crystals), using multielement Ge detector arrays (13 or 30 elements) or a Kr-gas-filled Lytle-type ionization chamber equipped with Sr X-ray filters. Spectra were background subtracted, splined, and fit in k-space using EXAFSPAK (21) or SixPACK (22) and IFEFFIT (23). EXAFS spectra were fitted with linear combinations of model spectra using EXAFSPAK to estimate the contribution of species in samples. This procedure was tested by fitting spectra from physical mixtures of known amounts of autunite (Ca(UO2)2(PO4)2‚ 10H2O, which has an indistinguishable spectrum from chernikovite) and schoepite (UO2(OH)2‚H2O). The resulting calibration curve was found to have a slope of 1.0, and the average 1-sigma uncertainty was 6%, which derives primarily from low-frequency low-amplitude structure originating in the spline fit. EXAFS spectra were also fit using a shell-byshell approach based on theoretical phase and amplitude parameters. Residual spectra were obtained by subtracting Oax and Oeq shells from the spline-subtracted EXAFS. FEFF 7 (24) phase and amplitude functions were generated as described previously (25). In cases where Debye-Waller factors (σ2) were too highly correlated with coordination number (CN) to allow independent fitting of σ2 and CN, the σ2 values were set to average values. This procedure is justified by the similarity of U(VI) local structures in the samples being compared and facilitates comparison of fit-derived CNs. Three transdioxo uranyl multiple scattering (MS) paths (1. UdOaxdUdOax2, 2. UdOaxdOax2, and 3. UdOaxdUdOax) were added to all fits, and their CNs, R, and σ values were calculated from the Oax as described previously (26). Based on fits to aqueous, σ2 for MS paths 1 and 2 were calculated as 3 × σ2Oax. SR-XRD was used to determine the existence and identity of crystalline uranyl phases in U(VI)-reacted bone apatites. SR-XRD intensity data were collected on wet solids using the 4644
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 20, 2003
2-circle diffractometer on SSRL beam line 2-1 using the method and parameters described in ref 19. Wet samples were packed into 0.5-mm deep aluminum holders and covered loosely with 4-mil Kapton film. Scans were collected for both a wide region (2-100° 2θ) and a narrow region (5-20° 2θ). The highest intensity peaks for uranyl phosphate phases occur in this narrow region, which is relatively free of background scattering from the Kapton film.
Results and Discussion Batch Uptake. The final pH of samples prepared in AGW was within or near the pH range measured in the PRB (Table 2). The pH of three samples prepared in 0.1 M NaNO3 increased significantly during equilibration likely because of a lack of buffering. The percent uranium sorption by bone apatites was greater in the samples prepared in 0.1 M NaNO3 in comparison to samples prepared in AGW with 4.8 mM total carbonate at comparable pH (Table 2). This difference likely results from the formation of aqueous uranyl carbonate complexes that dominate solution speciation in the AGW (>99% Ca2UO2(CO3)3,aq, UO2(CO3)22-, UO2(CO3)34-). Speciation calculations used the stability constants listed in ref 20 and included ternary calcium uranyl carbonato complexes (27). U(VI) speciation in Fry Canyon groundwater and in the PRB was also dominated by uranyl carbonate complexes (>99% CaUO2(CO3)32-,UO2(CO3)22-, UO2(CO3)34-). The speciation for samples prepared in 0.1 M NaNO3 was dominated by phosphate (>88% UO2PO4-, UO2HPO4,aq) or hydroxyl complexes (>99% UO2(OH)2,aq, UO2(OH)3-, (UO2)3(OH)7-) for pH e 7.5 and pH g 8.4, respectively. The ion activity product for chernikovite (log IAPCHN) was calculated for final solution concentrations as an indicator of the saturation state with respect to chernikovite as the solubility-limiting phase (log IAPCHN ) 2log {H+} + -3 2log {UO2+ 2 } + 2log{PO4 }, where {} are free ion activities). The log IAPCHN averaged -56.4 for samples in which chernikovite was detected by EXAFS or SR-XRD, as described below (CP3-3, CP3-4, BB-1-B, Figure 1), and is in good
FIGURE 1. The log ion activity product for chernikovite (log IAPCHN) versus log sorbed U(VI) concentration for bone apatite samples in Table 2. Open squares represent samples prepared in 0.1 M NaNO3 with no added dissolved carbonate; filled diamonds are samples prepared in artificial groundwater (4.8 mM [CO3]T); filled triangles are samples recovered from the Fry Canyon PRB. Dashed line represents the average log IAPCHN previously observed for U(VI)reacted synthetic apatite where chernikovite dominated U(VI) uptake by synthetic HA (19). Circled symbols represent samples where chernikovite was detected by EXAFS or SR-XRD. agreement with the average log IAPCHN of -56 calculated for U(VI)-reacted HA where chernikovite precipitation dominated U(VI) uptake (19). The lower IAPCHN values for the remaining samples suggest U(VI) sorption by processes other than chernikovite formation. The elevated log IAPCHN for the bone meal samples (BB1-A, -B, Figure 1) that exceed the apparent solubility for chernikovite is attributed to residual organic matter in the fertilizer grade bone meal, which may contribute dissolved P species other than orthophosphate. Bone char pellets were recovered from the Fry Canyon PRB after 18 months operation. During this period, greater than 90% of dissolved uranium (6-10 µM) entering the PRB was removed from an estimated 140 pore volumes of groundwater (18). A sample recovered within 5 cm of the upgradient face of the PRB (R2-3) had uranium uptake approximately equal to the maximum uptake observed for the CP5 material in a column experiments with 10 µM [UT] AGW (650 µg U(VI)/g (20)), indicating the upgradient face of PRB was approaching the equilibrium sorption expected for these conditions. A log IAPCHN of -60.1 was calculated for groundwater collected from the monitoring well closest to the cores samples at the time of coring. SR-XRD Results. Crystalline uranium phosphate, hydroxide, and carbonate phases were not detected in the SRXRD patterns of any of the samples with the exception of CP3-3, which contains the (001) reflections for both chernikovite and autunite (Figure 2). Previously, we reported detection limits for these uranyl phosphates of 2500 µg U(VI)/g for chernikovite and 350 µg U(VI)/g for autunite for the 7.7° and 6.8° 2θ peaks, respectively (19). The heights of the chernikovite and autunite (001) peaks are similar suggesting chernikovite is about 10 times more concentrated than autunite (Figure 2). EXAFS Results. The X-ray absorption near edge structure (XANES) spectra of U(VI)-reacted bone meal, bone charcoal, and bone char pellets all contain a XANES peak at 17 188 eV (see Figures 1 and 2 in Supporting Information), which has previously been attributed to uranium in the +6 oxidation state (26). In addition, the intensity maximum for the absorption edge occurs at the position characteristic of U(VI) as opposed to U(IV). Both observations indicate that uranium
FIGURE 2. X-ray diffraction region scan (6-9.5° 2θ) for U(VI)-reacted and unreacted bone charcoal apatites. Sample codes are listed in Table 2. The NR designation refers to solids not reacted with U(VI) but equilibrated in AGW. Reflections for autunite (Ca(UO2)2(PO4)2‚ 10H2O), chernikovite (H2(UO2)2(PO4)2‚10H2O), and hydroxyapatite are shown. Incident X-ray beam energy of 10 keV (λ ) 1.24 Å) was used. Data were collected at 0.02° 2θ intervals, 5 s dwell times for up to 180 scans and averaged to accurately define peak shapes.
FIGURE 3. U-LIII EXAFS spectra (solid lines) for U(VI)-reacted bone meal and bone charcoal apatites. Dashed lines represent fits of spectra to two-component linear combination of U(VI) surface complex and chernikovite (Table 3). Uranyl surface complex on HA and chernikovite spectra are from ref (19). Sample conditions corresponding to abbreviations are shown in Table 2. is present in the samples as U(VI). Information regarding the uranium local coordination environment can be gained by VOL. 37, NO. 20, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4645
FIGURE 4. (a). EXAFS spectra (solid lines) for U(VI)-reacted apatites for samples with no detectable uranyl phosphate phase and fits (dashed lines) reported in Table 4. (b) Fourier transformed spectra (solid lines) and fits (dashed lines) illustrating radial distribution of atoms surrounding U(VI). Radial distance, R + δR, is not corrected for phase shift. Sample conditions corresponding to abbreviations are shown in Table 2.
TABLE 3. Two Component, Least-Squares Fits to EXAFS Spectra for U(VI) Reacted Apatite Samples Containing Chernikovitea fitted components sample BB1-B CP3-3 CP3-4 a
0.1 M NaNO3 0.1 M NaNO3 AGW
µg U(VI)/g
pH
% chernikovite
% surface complex
chernikovite µg U(VI)/g
11850 4760 2670
6.4 7.5 7.0
73 84 35
27 16 65
8650 4000 940
Standard deviation (1σ) for all samples is estimated at 6%.
qualitative inspection of the EXAFS spectra from the samples (Figures 3 and 4). The EXAFS spectra for samples BB1-B, CP3-3, and CP3-4 are qualitatively similar to chernikovite formed from U(VI) reaction with HA (Figure 3). In contrast, the spectra of all the other samples most resemble the spectra for the U(VI) surface complex on synthetic HA (Figure 4). These observations suggest that, in total, the U(VI)-reacted samples contained two generally distinct uranyl coordination environments, which correlate with total uranium concentration. At sorbed U(VI) concentrations < 5800 µg U(VI)/g, uranium appears to be dominantly adsorbed as a surface species on HA, whereas precipitation of chernikovite occurred at higher sorbed U(VI) in the absence of CO2 (19). To quantify the extent of chernikovite precipitation in the present system, spectra were fit with linear combinations of spectra from chernikovite, and from the dominant U(VI) surface complex observed previously on synthetic HA (19), which is used here as a model. Fits to three samples, BB1-B, CP3-3, and CP3-4 (Figure 3), required between 35 and 84% chernikovite (Table 3). The calculated 4000 µg U(VI)/g component of chernikovite in sample CP3-4 is consistent with the presence of chernikovite in the SR-XRD pattern for this sample and the 2500 µg U(VI)/g detection limit for this phase (Figure 2). Linear combination fits to the remaining sample spectra did not require a significant component of chernikovite. Mechanisms of U(VI) Sorption. Both the SR-XRD and EXAFS indicated that all samples other than BB1-B, CP3-3, and CP3-4 were free of chernikovite and other crystalline 4646
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 20, 2003
uranium phases. Furthermore, the EXAFS spectra are distinctly different than spectra of U(VI) phosphates, carbonates, and oxide solids (19), indicating that these solids do not account for U(VI) sorption in these samples. U(VI) substitution into Ca sites has been observed for fluoroapatite synthesized by annealing UO2, Ca3(PO4)3, and CaF2 at 1380 °C (28). However, solid solution of U(VI) for Ca in apatite was previously ruled out (19) for room temperature, aqueous systems similar to those described herein because no change in the apatite d spacing was observed over a large range of initial [U] and because of the ubiquitous presence of the transdioxo uranyl group and the absence of the uranate geometry. EXAFS spectra from the present samples were subsequently fit using a shell-by-shell approach and theoretical phase and amplitude parameters (Table 4). EXAFS spectra and corresponding Fourier Transforms (FTs) are plotted in Figure 4 for whole spectra and Figure 5 for residuals obtained by subtracting Oax and Oeq spectra contributions to improve sensitivity for second shell analyses. FT distances reported here differ from the true interatomic distances (R) by a phase shift (δR). Discussions of FT distances herein are given in units of R + δR (Å), whereas discussions of structures are given in R (Å). EXAFS fits to sample spectra (Figure 4, Table 4) show that U(VI) has two Oax at ca. 1.78 Å and Oeq at 2.29-2.40 Å, the geometry indicative of the uranyl cation. The Oeq shell is split into two components in R2-2 CP5 and all the BB1 sample spectra. Two Oeq shells could not be justified in the other spectra and were fit instead with single shells (Table 4). The
TABLE 4. Shell-by-Shell EXAFS Fit Results
sample BK1-2 BK1-1 BK1 0106 BB1-A
BB1-C
BB1 0106
CP3 0106 U-HA 0106 PRB R2-2 PRB R2-3
µg U(VI)/g pH solutiona
Oax
Oeq
2960 8.5 N, No CO2 5530 7.7 AGW 8800 7.7 AGW 2970
CNb
2d
6d
Rc (Å) σ2 (Å2) CNb Rc (Å) σ2 (Å2) CNb Rc (Å) σ2 (Å2) CNb
1.785 (2) 0.0028 (1) 2d 1.816 (1) 0.0020 (1) 2d 1.794 (4) 0.0035 (2) 2d
6.4
Rc (Å)
1.796 (1)
N 2540
σ2 (Å2) CNb
0.0031 (1) 2d
6.8
Rc (Å)
1.806 (2)
AGW 11200
σ2 (Å2) CNb
0.0060 (1) 2d
7.0
Rc (Å)
1.782 (6)
AGW
σ2 (Å2)
0.0046 (3)
1180 7.8 AGW 12300 7.5 AGW 190 7 GW 690
CNb Rc (Å) σ2 (Å2) CNb Rc (Å) σ2 (Å2) CNb Rc (Å) σ2 (Å2) CNb
2d 1.787 (2) 0.0026 (1) 2d 1.764 (3) 0.0033 (1) 2d 1.765 (4) 0.0032 (2) 2d
7
Rc (Å)
1.773 (2)
GW
σ2 (Å2)
0.0058 (1)
2.352 (5) 0.0149 (4) 6.4 (3) 2.415 (2) 0.0120 (5) 6.0 (9) 2.420 (9) 0.015 (2) 2.73 (8) 1.9 (1)e 2.308 (2) 2.492 (4)e 0.005d 2.3 (1) 2.5 (1)e 2.307 (4) 2.458 (4)e 0.0053 (4) 3.7 (9) 2.0 (6)e 2.32 (1) 2.52 (3)e 0.008 (2) 0.008 (2)e 6.5 (4) 2.403 (4) 0.0148 (9) 6.7 (8) 2.293 (9) 0.018 (2) 5.3 (6) 2.314 (9) 0.013 (1) 4.0 (2) 0.7 (3)e 2.336 (5) 2.53 (3)e 0.01d
P/C f
Ca
1.8 (4) P 3.034 (7) 0.013 (2) 1.59 (7) C 2.909 (3) 0.0041d 2.1 (2) C 2.881 (4) 0.0053 (9) 0.7 (2) P
0.3 (1) 3.88 (4) 0.005d
0.4 (1) 3.79 (1) 0.005d 0.2 (2)
2.3 (3) 3.597 (4) 0.013 (1) 0.8 (2) 3.59 (2) 0.01d 1 (1)
3.01 (2)
3.77 (4)
3.52 (3)
0.004 (1) 1.4 (1) C
0.005d
0.01 (1) 0.9 (1)
0.4 (1)
2.917 (3)
3.570 (4)
3.906 (8)
0.0045 (8) 1.7 (3) C
0.007 (1) 0.4 (1)
0.009 (2)
2.87 (1)
3.52 (2)
0.004 (1)
0.005d
1.6 (1) C g 2.844 (7) 0.0041d 0.8 (3) C g 2.82 (1) 0.004 (3) 0.5 (1) C g 2.94 (2) 0.0041d 2.1 (3) C
1.2 (1) 3.483 (9) 0.01d 1.1 (1) 3.51 (1) 0.01d 2.1 (2) 3.51 (2) 0.01d 2.5 (8)
2.945 (4)
3.615 (9)
0.005 (1)
0.017 (4)
P
U 1 (1) 3.87 (3) 0.016 (1) 0.9 (1) 3.928 (3) 0.0069 (8) 0.54(7) 3.880 (7) 0.005d
0.6 (2) 3.840 (9) 0.003 (6)
a
Background electrolyte N: 0.1 M NaNO3; AGW artificial groundwater (4.8 mM total carbonate); GW PRB groundwater. b Coordination number ((30%). c Interatomic distance ((0.03 Å). d Parameter value held constant. e Two Oeq shells were fit. f C and P denote parameters for C or P shell, respectively. g P and C provide similar fit quality. Estimated standard deviations (ESDs) are given in parentheses. The 99.5% confidence interval limits are obtained by multiplying the ESDs by 2.81.
σ2 values for these single Oeq shells, ca. 0.015 Å2, were about two times the σ2 values for split Oeq shells, ca. 0.008 Å2, observed previously (19). The large σ2 values in these latter samples thus suggest splitting of the Oeq shell. Distortion and splitting of the Oeq shell is common for uranyl adsorbed on oxide surfaces (25, 29) due to the asymmetry of the coordination environment and is interpreted here as evidence for surface complexation of U(VI). As evident in the residual EXAFS (Figure 5), all spectra contain strong FT components at (i) 2.4 Å and (ii) 3 Å (R + δR). Several of the samples also show a FT component at (iii) 3.7 Å. These shells are interpreted (explanation follows) as arising from the following: (i) C at 2.8-2.9 Å (in the presence of 4.8 mM [CO3]T AGW) or P at ∼3.02 Å (air headspace or CO2 excluded); (ii) P at 3.5-3.6 Å and/or Ca at ∼3.8 Å; and (iii) U at 3.9 Å. Previous investigation of uranyl reacted with HA in the absence of dissolved carbonate has shown that the 2.4 Å FT frequency is well described by a single shell of P at about 3.02 Å (19). The same conclusion was obtained for two samples prepared in either the absence of CO2 or with no added [CO3]T (BK1-2 and BB1-A). The 3.01-3.03 Å U-P distance is consistent with bidentate coordination of the uranyl equatorial plane to a phosphate group located in the apatite surface or to phosphate groups present as ternary ligands (i.e., coordinated to U(VI) trans to the apatite surface).
The proposed conformation of these structures is, therefore, a four-member U-O-P-O ring. In the presence of 4.8 mM [CO3]T (AGW), the EXAFS corresponding to the 2.4 Å FT peak were better fit with one to two C at 2.8 to 2.9 Å than with a P shell (on average χ2 decreased 3-fold for fits with C compared to P) for samples BK1-1, BK1-0106, BB1-C, BB1-0106, and PRB R2-3 (Figure 5, Table 4). This U-C distance is consistent with bidentate coordination of uranyl to carbonate, which is a frequently observed topology in uranyl carbonates (30). The improved fits with carbonate suggests that it displaces phosphate as a coordinating ligand around uranyl, which in turn implies that the bidentate phosphate groups in question are present as ternary ligands. The residual spectra for samples CP30106, PRB R2-2, and HA-0106 were equally well fit with bidentate carbonate with low C CNs (
