Molybdenum Speciation in Uranium Mine Tailings Using X-Ray

Dec 13, 2010 - Collection of Samples and Bulk Chemistry. Seven uranium mine tailings core samples (four collected during the 2005. DTMF drilling progr...
0 downloads 0 Views 313KB Size
Environ. Sci. Technol. 2011, 45, 455–460

Molybdenum Speciation in Uranium Mine Tailings Using X-Ray Absorption Spectroscopy J O S E P H E S S I L F I E - D U G H A N , * ,† INGRID J. PICKERING,† M. JIM HENDRY,† GRAHAM N. GEORGE,† A N D T O M K O T Z E R †,‡ Department of Geological Sciences, University of Saskatchewan, 114 Science Place, Saskatoon, SK, Canada, S7N 5E2, and Cameco Corporation, 2121-11th Street West, Saskatoon, SK, Canada, S7M 1J3

Received August 27, 2010. Revised manuscript received November 20, 2010. Accepted November 25, 2010.

Uranium (U) mill tailings in northern Saskatchewan, Canada, contain elevated concentrations of molybdenum (Mo). The potential for long-term (>10 000 years) mobilization of Mo from the tailings management facilities to regional groundwater systems is an environmental concern. To assist in characterizing long-term stability, X-ray absorption spectroscopy was used to define the chemical (redox and molecular) speciation of Mo in tailings samples from the Deilmann Tailings Management Facility (DTMF) at the Key Lake operations of Cameco Corporation. Comparison of Mo K near-edge X-ray absorption spectra of tailings samples and reference compounds of known oxidation states indicates Mo exists mainly as molybdate (+6 oxidation state). Principal component analysis of tailings samples spectra followed by linear combination fitting using spectra of reference compounds indicates that various proportions of NiMoO4 and CaMoO4 complexes, as well as molybdate adsorbed onto ferrihydrite, are the Mo species present in the U mine tailings. Tailings samples with low Fe/Mo (113) molar ratios are dominated by NiMoO4, whereas those with high Fe/Mo (>708) and low Ni/Mo (6 unacceptible. Based on the results of the PCA and target transform, linear combination fitting (LCF) analysis was applied to determine the fraction of each Mo species present in each mine tailings sample. In general least-squares fitting is preferred to alternatives because it affords the opportunity to make small corrections for energy shifts between the individual spectra in the data set, although in our case we did not do this. The underlying principle of LCF analysis is the additive nature of absorption from each species in the sample; this is presently the most common approach for quantitative XAS near-edge analysis (18). PCA and TARGET routines in EXAFSPAK were used for the principal component analysis and target transform, respectively; the DATFIT routine was used for the linear combination fitting analysis.

Results and Discussion Oxidation State of Mo in Mine Tailings. An important and common application of synchrotron analyses on environmental and industrial compounds is the utilization of the shift of the absorption edge position of the collected XAS near-edge spectra to determine the oxidation state of the absorbing atom (14, 18). For similar ligand environments of a given atom, higher oxidation states result in higher edge energy shifts and vice versa, a consequence of the atoms in higher oxidation states having fewer electrons than protons, which results in the energy states of the electrons to be lowered slightly and the absorption edge energy to increase (29). Comparison of the near-edge position of the normalized Mo K-edge absorption spectra for the uranium mine tailings samples and reference compounds of known oxidation states (Figure 1) demonstrates that Mo(VI) is the dominant form

FIGURE 1. X-ray absorption near-edge structure of reference compounds with various molybdenum oxidation states and experimental (solid) and linear combination fits (dashed curves) for the Mo K-edge near-edge spectra of (a) 2005 and (b) 2008 tailings samples. of molybdenum in the mine tailings. The characteristic preedge feature in the near-edge spectra indicates that Mo in the tailings occurs mostly as a molybdate. This peak is due to 1s f 4d transitions that are formally dipole forbidden but gain intensity in a noncentrosymmetric Mo environment due to admixing of the 4d levels with p orbitals (30). Principal Component Analysis (PCA). The result of the PCA, a plot of the eigenvectors against energy (Supporting Information Figure S1), shows the likelihood of the presence of three Mo species in the seven tailings samples. Reconstructions of the data set systematically excluding successively smaller components suggested that a minimum of three components were required. The use of two components gave miss-matches in the intensities of the 1s f 4d region of the spectrum, which is the best resolved feature, and that least affected by background subtraction. Inspection of IND values indicated minima at either two or three components, depending upon the exact energy range of the data used in the analysis (19950-20150 eV suggests three components, and 19900-20200 eV two). The combined reconstruction and IND values thus suggest that the data can be represented by three components. VOL. 45, NO. 2, 2011 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

457

FIGURE 2. X-ray absorption near-edge structure of Mo reference compounds. MoO4_Fe(OH)3 is molybdate adsorbed on ferrihydrite. Target transforms were used to relate the principal components of the sample spectra to those of the candidate physical reference standards (Figure 2). SPOIL values (SI Table S1) clearly indicate that MoS2 is not present, with a value close to 7. As expected, most of the molybdates are in the good-fair region with SPOIL values between 3 and 4.5. The exceptions were CaMoO4 and PbMoO4 which showed a SPOIL values of 4.8 and 5.7, respectively, indicating a poor match. The best matches based on SPOIL alone were NiMoO4 and molybdate adsorbed on ferrihydrite. Simple examination of the variance between target and target transform also indicates that these are the best matches. Insofar as the third component suggested by PCA is concerned, H2MoO4 and MoO3 cannot be present because the pH of the samples is high (∼10). Moreover, FeMoO4, with a poor variance but good SPOIL, is most unlikely to be present because the sample conditions are too oxidizing, and PbMoO4 can also be excluded on the basis of chemical composition (21, 22). Thus, the only remaining candidate for the most minor component is CaMoO4. We conclude that the data are best represented by NiMoO4 and molybdate adsorbed on ferrihydrite, with a possible minor component which resembles CaMoO4. Target transform results for NiMoO4, molybdate adsorbed on ferrihydrite (MoO4-Fe(OH)3), and CaMoO4 is available in Supporting Information Figure S2a-c. Superimposed plots (Supporting Information Figure S2d) of the near-edge spectra of the three likely Mo components of the mine tailings over the energy range 19 900-20 200 eV demonstrate that these three spectra show distinguishable spectral features and thus offer a good basis for a reliable quantitative near-edge analysis. The identification of NiMoO4, MoO4_Fe(OH)3, and CaMoO4 as the Mo species present in the mine tailings is in agreement with the results of diagnostic equilibrium modeling conducted (using PHREEQCI) on representative mine tailing samples (from the DTMF) using porewater chemical analysis data. Saturation indices (SI) (between -0.36 and 1.44) from the results of the equilibrium modeling identified the same mineral phases, among others, as possible solubility controls on Mo in the mine tailings (22, 24). Quantification of Mo Species in Tailings Samples. The LCF analysis was conducted over an energy range of 19 900 to 20 200 eV. Sample spectra from the 2005 (E2-GC70; Figure 458

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 45, NO. 2, 2011

FIGURE 3. Experimental and linear combination fits for the Mo K-edge near-edge spectra of (a) GC70 and (b) GC30 tailings samples together with spectra of the three standards, scaled according to their fractional contributions to the fitted spectra. 3a) and 2008 (E6-GC30; Figure 3b) sampling campaigns are provided as illustrative examples. Both are superimposed with the spectra of the reference standard compounds, scaled according to their fractions in the LCF, which contribute to the fitted curve. The intensity of the single absorption maximum in the E2-GC70 spectrum is primarily due to MoO4_Fe(OH)3 with significant contributions from the spectra of NiMoO4 and to a lesser extent CaMoO4 (Figure 3a). Likewise, the intensity of the single absorption maximum in the E6-GC30 spectrum (Figure 3b) is primarily due to NiMoO4 but again with significant contributions from the spectra of MoO4_Fe(OH)3 and to a lesser extent CaMoO4. Combinations of these three components adequately explain the spectra for all seven tailings samples (Figures 1; Table 2). The resulting near-edge fit fractions (Table 2) correspond to the amount of the respective normalized reference standard spectra required to yield a good match between simulated and experimental near-edge spectra. The residual values (Table 2) can be used as a measure of the goodness of the fit, whereas smaller residuals are indicative of better fit. The total values (Table 2) also reflect goodness of fit; the totals are not constrained in the fits, so better fits are represented by the totals closest to 1.00. Results obtained when the LCF was conducted over a shorter energy range (19 950-20 100 eV) yield similar results. LCF results for the 2005 samples (Table 2) indicate NiMoO4 is the dominant Mo mineral phase in E2-GC83 and E2-GC97

TABLE 2. Type and Amount of Mo Species in Uranium Mine Tailings Obtained from a Combination of PCA, Target Transform, and LCF of near-Edge Spectraa samples

NiMoO4

MoO4_Fe(OH)3

CaMoO4

total

residual (×103)

(2005) E2-GC70b E2-GC73b E2-GC83c E2-GC97c

0.29 ( 0.04 0.26 ( 0.05 0.95 ( 0.03 0.83 ( 0.04

0.65 ( 0.05 0.60 ( 0.03 0.06 ( 0.03 0.09 ( 0.02

0.06 ( 0.03 0.14 ( 0.04 0.00 0.09 ( 0.03

1.00 1.00 1.01 1.01

289 333 303 212

(2008) E6-GC22c 0.91 ( 0.02 E6-GC29c 0.45 ( 0.03 E6-GC30c 0.74 ( 0.03

0.09 ( 0.02 0.16 ( 0.02 0.18 ( 0.01

0.00 1.00 0.39 ( 0.02 1.00 0.08 ( 0.02 1.00

183 193 103

a Entries represent fractional amount ( estimated standard deviation from the fit (e.s.d.) (a) 2005 (E2) samples. (b) 2008 (E6) samples. The residual is defined as ∑(Iobs-Icalc)2/N, where Iobsand Icalc are the observed and calculated normalized intensities, respectively, and the summation is over N data points. MoO4_Fe(OH)3 Molybdate adsorbed on ferrihydrite. b McArthur river ore tailings. c Deilmann ore tailings.

whereas molybdate adsorbed onto ferrihydrite (MoO4_ Fe(OH)3) is the dominant Mo phase in E2-GC70 and E2GC73. LCF results for the 2008 samples (E6-GC22, E6-E6GC29, and E6-GC30) (Table 2) show that although NiMoO4 is the dominant mineral phase in all three samples, there is significant variation in the proportion of NiMoO4 present in each of the samples. A plot of DTMF elevation against the Fe/Mo molar ratio of all the samples collected for location E2 (2005) and E6 (2008) (Figure 4a) illustrates a significant difference between the deeper tailings and the shallow tailings in the DTMF. This is also demonstrated in a plot of percentage molybdate adsorbed on ferrihydrite (determined from the XAS experiment) against Fe/Mo ratio (Figure 4a; secondary y-axis). The tailings below 410 m above sea level (masl) have relatively low Fe/Mo ratios (708). Similarly, plots of both DTMF elevation and percentage NiMoO4 (determined form the XAS experiment) against the Ni/Mo molar ratio (Figure 4b; secondary y-axis) also show a significant difference between the deeper and shallow DTMF tailings. Tailings below 410 masl have relatively high Ni/Mo ratios (>113) compared to overlying tailings (ratios