Journal of Colloid and Interface Science 222, 184–197 (2000) doi:10.1006/jcis.1999.6621, available online at http://www.idealibrary.com on X-Ray Absorption Spectroscopy of Strontium(II) Coordination I. Static and Thermal Disorder in Crystalline, Hydrated, and Precipitated Solids and in Aqueous Solution Peggy A. O’Day,∗,1 Matthew Newville,† Philip S. Neuhoff,‡ Nita Sahai,∗,2 and Susan A. Carroll§ ∗ Department of Geology, Arizona State University, Tempe, Arizona 85287-1404; †Consortium for Advanced Radiation Sources, University of Chicago, Chicago, Illinois 60637; ‡Department of Geological and Environmental Sciences, Stanford University, Stanford, California 94305-2115; §Earth and Environmental Sciences Directorate, Lawrence Livermore National Laboratory, Livermore, California 94550 Received April 30, 1999; accepted September 30, 1999 Detailed analyses of crystalline, hydrated, and precipitated strontium compounds and an aqueous strontium solution by synchrotron extended X-ray absorption fine structure (EXAFS) were used to quantify local thermal and static disorder and to characterize strontium coordination in a variety of oxygen-ligated bonding environments. Analysis of anharmonic vibrational disorder (i.e., significant contribution from a third cumulant term (C 3 ) in the EXAFS phase-shift function) in compounds with low and high static disorder around strontium showed that first-shell anharmonic contributions were generally not significant above experimental error in the EXAFS fits (R ± 0.02 Å with and without C 3 ). The only case in which a significant apparent decrease in Sr–O distance was observed with increasing temperature, and for which a third cumulant term was significant, was for dilute strontium in aqueous solution. Empirical parameterization of Debye–Waller factor (σ 2 ) for strontium compounds as a function of backscatterer atomic number (Z), interatomic Sr–Z distance, and temperature of spectral data collection showed systematic increases in σ 2 as a function of increasing temperature and Sr–Z bond length. At values of σ 2 greater than ≈0.025 Å2 (for N < 12 and RSr- Z > 3 Å), backscattering was generally not significant above noise levels in spectra of compounds of known crystal structure. Comparison of the EXAFS spectra of freshly precipitated SrCO3 (spectra collected wet) to that of dry, powdered strontianite (SrCO3 (s)) indicated no significant differences in the local atomic structure around strontium. Analysis of partially hydrated strontium in natural Ca-zeolite (heulandite) showed that strontium is substituted only in the calcium (Ca2) site. Backscattering from aluminum and silicon atoms in the zeolite framework were apparent in the EXAFS spectra at low and room temperature at distances from central strontium of <4.2 Å. Comparison of strontium structural coordination determined in this and previous studies suggests that previous EXAFS determinations of hydrated strontium may have underestimated first-shell interatomic distances and coordination numbers because minor contributions to the EXAFS phase-shift and amplitude functions were not accounted for, either theoretically or empirically. °C 2000 Academic Press 1 To whom correspondence should be addressed: E-mail: [email protected]. Present address: Department of Chemistry and Biochemistry, University of Maryland, College Park, MD 29742 USA. 2 0021-9797/00 $35.00 C 2000 by Academic Press Copyright ° All rights of reproduction in any form reserved. Key Words: X-ray absorption spectroscopy; EXAFS; strontium; anharmonic disorder; cumulant expansion; zeolite. INTRODUCTION The mobility of strontium in the environment as the radioactive contaminant 90 Sr is controlled by chemical partitioning between natural waters and minerals in soils and sediments. Because of the large ionic radius of Sr2+ and its chemistry as an alkaline earth cation, strontium typically substitutes for calcium in solids and can retain its waters of hydration during adsorption, ion exchange, and precipitation. Strontium (II) has variable coordination with oxygen and water ligands, reported from 6–12, in solids and aqueous solutions (1, 2). The number of oxygen ligands coordinating strontium is not well established in hydrated compounds such as zeolites or for sorption complexes and may vary from system to system. Knowledge of the degree of hydration, Sr–O interatomic bond distances, and atomic structural differences among hydrated solids and sorbed complexes is important in the formulation of sorption and diffusion models aimed at predicting strontium attenuation in the environment. Thus, accurate molecular characterization is needed in order to distinguish and quantify modes of strontium coordination in solution, within solid phases, and on their surfaces. Recent studies have employed X-ray absorption spectroscopy (XAS) and extended X-ray absorption fine structure (EXAFS) analysis to determine strontium coordination in silicate glasses (3, 4), in natural and synthetic calcite (5), in anorthite and zeoliteA (6), in aqueous solution and oxyhydroxides (2, 7), after sorption to calcite and kaolinite (8), after sorption to hydrous ferric oxide (9), and after sorption to clay minerals (10). Comparison of strontium atomic structural parameters derived from these studies shows significant disagreement. Analysis of large, strongly hydrated cations such as strontium by EXAFS, which probes the local molecular structure around a central absorber atom (usually to ≈6 Å), can be problematic. For these cations, EXAFS backscattering from atoms beyond the first coordination shell is 184 STRONTIUM(II) COORDINATION, I often weak because of static and substitutional disorder, bond vibrational anharmonicity, variable coordination, and long interatomic bond distances. In addition, recent XAS studies have pointed out that multielectron excitations and the presence of hydrogen atoms can have a significant effect on the quantitative analysis of the EXAFS of strontium in solution (7, 11, 12). All of these factors have probably contributed in some degree to the discrepancies among strontium structural parameters derived from EXAFS analysis in previous studies. Moreover, they have influenced the interpretation of strontium coordination in surface-sorbed and precipitated species (see Sahai et al. (13)). In this study, we analyzed EXAFS spectra collected at different temperatures of strontium in aqueous solution and in a series of strontium–oxygen compounds that included ordered crystalline solids, hydrated strontium-bearing natural zeolites, and freshly precipitated strontianite (SrCO3 (s)). We used compounds of well-known crystal structure to constrain fit parameters in the EXAFS scattering function, we fit known compounds to test the accuracy of the estimates, and we used these constrained parameters to fit spectra of strontium in unknown bonding sites (i.e., precipitated samples and zeolites). Based on this analysis, we discuss (i) the effects of static, thermal, and anharmonic vibrational disorder on the quantitative determination of EXAFS interatomic distances for oxygen-ligated strontium; (ii) the empirical derivation of Debye–Waller factors (σ 2 ) for strontium compounds as a function of backscatterer Z , interatomic Sr–Z distance, and temperature, and the ability to identify secondneighbor, low-Z backscatterers in Sr-EXAFS spectra; (iii) the bonding environment of strontium in fresh, homogeneously, or heterogeneously precipitated strontianite (SrCO3 (s)), and as a partially hydrated ion in zeolite channels. These results provide constraints on XAS analysis of strontium, as well as insights into the local molecular coordination of hydrated strontium, which are important for the interpretation of contaminant strontium removal from natural waters by sorption or precipitation on mineral surfaces. In our companion study (13), we apply the EXAFS constrained parameters derived here to the spectral analysis of surface sorbed and precipitated strontium. EXPERIMENTAL METHODS EXAFS Materials and Sample Preparation The solid compounds SrO(s) (Alfa Aesar, reagent grade), SrCO3 (s) (natural sample, courtesy of G. E. Brown, Jr.), SrTiO3 (s) (courtesy of G. A. Waychunas), α-SrHPO4 (s) (Aldrich Chemicals, reagent grade), and three strontium-bearing heulandite samples ((Na, Ca)4 Al9 Si27 O72 · 24H2 O, a natural zeolite in which strontium substitutes for calcium), and an aqueous solution of 10−3 M SrCl2 were examined with XAS. The zeolite samples, containing naturally abundant strontium concentrations, were collected as macroscopic crystals from volcanic areas in Eastern Iceland. They were identified as the mineral heulandite with powder X-ray diffraction (XRD) and analyzed chemically by X-ray fluorescence (Stanford University). Solid 185 compounds were hand crushed with an agate mortar and pestle, diluted by physically mixing with B2 O3 , and loaded into Teflon sample holders with Kapton tape windows. Alternatively, some reference samples were prepared by layering several thin films of pure sample powder on adhesive tape to achieve sufficient X-ray absorption (>50–70% of the incoming beam intensity). Precipitated SrCO3 (s) XAS samples were prepared by mixing 0.04 M SrCl2 solutions with 0.1 M Na2 CO3 in the presence or absence of either kaolinite (natural Georgia kaolinite, Clay Minerals Society Source Repository (KGa-1b)) or goethite (synthesized following the method of Schwertmann and Cornell (14); see Sahai et al. (13) for characterization of these solids). Samples were filtered immediately, loaded into Teflon samples holders, and sealed with Kapton tape. X-ray absorption spectra were collected within 2 h of sample loading. For precipitates and solutions collected at cryogenic temperatures, wet samples were quenched by immersion in liquid nitrogen and then placed in a helium cryostat in the beamline hutch. EXAFS Data Collection EXAFS spectra were collected on wiggler beamline IV-3 at the Stanford Synchrotron Radiation Laboratory (SSRL). Beam current varied from about 55–90 mA at an energy of 3 GeV. Two different cuts of a Si(220) monochromator crystal with a d-spacing of 1.92013 Å were used. The incident beam was detuned from 50–70% of the maximum incoming intensity to reject higher-order harmonic reflections. The mid-point of the absorption edge of dry, crystalline SrCO3 (s) reference compound (set to 16,105 eV) was used for energy calibration. Absorption spectra were collected at ambient (≈300 K) and cryogenic temperatures (70 and/or 10–20 K using a helium cryostat). Data collection at cryogenic temperatures amplifies the EXAFS oscillations at higher energy, but careful comparisons of low- and room-temperature spectra were made to evaluate whether sample freezing affected the experimental spectrum. Solid compound spectra were collected in transmission mode using Ar gas-filled ion-chamber detectors (2–4 scans averaged). Spectra for strontium aqueous solution, precipitated samples, and natural heulandite samples were collected in fluorescence mode using a 13-element germanium array detector with, in some cases, an aluminum or arsenic filter to reduce background fluorescence and/or soller slits to increase count rate. For these samples, 3–12 scans were collected to achieve an adequate signal/noise ratio. EXAFS Data Reduction Absorption spectra were analyzed with the program EXAFSPAK (15) which uses a curved-wave formalism and single-scattering approximation. Analysis of first-shell anharmonic vibrational effects for three reference spectra as a function of temperature using cumulant expansion were done with the program FEFFIT (16). Averaged experimental spectra were background-subtracted by fitting a straight line through the preedge region and a cubic spline beyond the absorption edge. 186 O’DAY ET AL. Spectra were normalized using the height of the edge-step near the maximum absorption edge. Normalization was extended into the EXAFS region using a Victoreen polynomial and McMaster coefficients (see Brown et al. (17)). A threshold energy (i.e., energy at which k, the photoelectron wave vector, equals zero) of 16,115 eV was used and spectra were weighted by k 3 to compensate for damping of oscillations at high k. Normalized EXAFS spectra were filtered over a k-range of 2 to 13–15 Å−1 depending on data quality and Fourier-transformed to produce Radial Structure Functions (RSFs) (shown here uncorrected for backscatterer phase shift). Reference phase-shift and amplitude functions used in nonlinear least-squares fitting of experimental spectra were calculated using the ab initio program FEFF (v. 6.0 and 7.02) (18–20). Known crystal structures of strontium(II) compounds determined by XRD studies (from the literature) were used to determine the positions of near-neighbor atoms around strontium (to 6–7 Å away) using the program Atoms (16, 21). Theoretical phase-shift and amplitude functions were calculated with FEFF using atom clusters with strontium as the central absorbing atom. Experimental EXAFS spectra of well-known compounds were fit with the theoretically calculated FEFF functions to estimate values for S02 (amplitude reduction factor) and 1E 0 (difference in threshold energy between theory and experiment) by fixing interatomic distances (R) and backscatterer numbers (N ) and varying S02 and 1E 0 (see O’Day et al. (22)). Reference compound spectra were also used to empirically derive values for σ 2 , the Debye–Waller term (or mean-square displacement of the bond length) which accounts for thermal and static disorder, for each backscattering shell around strontium as a function of backscatterer Z , Sr–Z interatomic distance, and temperature. Empirically determined values for σ 2 were then fixed in fits to determine N in tests on known structures and in samples for which N was not known (e.g., for Al and Si in zeolite samples). Least-squares fits to spectra, typically with R, 1E 0 (set as a single variable for all shells), and either σ 2 or N as adjustable parameters, were performed both on filtered spectra of individual peaks in the RSFs and on full normalized χ(k) spectra with no significant differences in fit results. In order to evaluate whether anharmonicity in Sr–O bonding affects the experimental EXAFS, a third cumulant term was added to the fit variables for the first-shell analysis of three reference spectra, SrO(s), SrCO3 (s), and Sr2+ (aq), as a function of temperature. Spectra were Fourier transformed between k = 2.0 and 11.0–13.0 Å−1 using a Hanning window and the filtered first Sr–O shell was fit between R = 1.1 and 2.7 Å (using FEFFIT and FEFF theoretical phase-shift and amplitude functions). For the cumulant analysis, spectra were fit in R-space rather than k-space. Distance (R), σ 2 , and S02 were allowed to vary in the fit (N fixed). Filtered spectra were fit with and without a third cumulant (C3 ) term, which becomes important when the effective radial distance distribution is non-Gaussian. The EXAFS phase-shift function is expanded by 18 = 2k1C1 − 4/3k 3 1C3 + · · · , where 18 is the difference in phase shift between reference and unknown, k is the photoelectron wave vector, 1C1 is the difference in mean distance R (1R in Å), and 1C3 is the difference in third cumulant (in Å3 ), which is a measure of skew or non-Gaussian form of the radial distribution function (23, 24). RESULTS Third Cumulant Contributions to Strontium EXAFS First-shell Sr–O scattering was fit with and without the third cumulant term (C3 ) in the solid compounds SrO(s) and SrCO3 (s), and in aqueous solution (10−3 M SrCl2 solution) (Table 1). Strontium oxide (SrO(s)) is a well-ordered solid with a cubic halite structure in which strontium has six first-neighbor oxygen atoms at 2.570 Å and a second shell of strontium atoms at 3.634 Å (25). Including the C3 term resulted in small differences in the fits at all three temperatures relative to the Sr–O distance determined from XRD (Table 1). Spectra for SrO(s) collected at 10 K, 70 K, and room temperature and first-shell fits are shown TABLE 1 Results for First-Shell EXAFS Fits of SrO(s), SrCO3 (s), and Sr2+ (aq) for Data Collected at Low and Room Temperature (RT), with and without a Third Cumulant Term, C3 T (K) 1R (Å)a σ 2 (Å2 ) 10 70 RT 10 70 RT 0.02(1) 0.01(1) −0.01(1) 0.02(1) 0.02(1) 0.04(1) SrO(s)b 0.0078(3) 0.0077(3) 0.0128(3) 0.0077(3) 0.0077(3) 0.0129(4) 10 70 RT 10 70 RT 0.00(1) −0.01(1) −0.03(1) −0.01(2) −0.01(2) −0.01(1) SrCO3 (s)d 0.003(2) 0.004(2) 0.007(1) 0.004(2) 0.004(2) 0.008(2) 10 RT 10 RT 0.03(1) −0.02(1) 0.05(3) 0.03(2) Sr2+ (aq)e 0.008(2) 0.011(2) 0.008(2) 0.011(1) C3 × 103 (Å3 ) c S02 0.95 c c 0.00(1) 0.03(1) 0.5(3) c 1.04 c c −0.2(4) −0.2(3) 0.1(4) c 1.54 c 0.7(7) 1.4(5) Note. Best-fit values are reported with the uncertainty in the statistical fit for the last digit shown in parentheses. a Fitted distance (1R) reported as the difference between the EXAFS fit and crystallographic distance from XRD analysis for fixed NSr-O . b Crystallographic distance for first shell R Sr-O = 2.570 Å; NSr-O = 6 (from Primak et al. (25)). c C forced to be 0.0. 3 d Fit using all five crystallographic distances (R Sr-O = 2.558–2.729 Å) for SrCO3 (see Table 2); 1R is the difference between fit and crystallographic distances for all five Sr–O shells with 1R treated as a single variable; NSr-O = 9. e Fit initially assuming R Sr-O = 2.620 Å and NSr-O = 6 (see text); solution is 10−3 M SrCl2 (aq). STRONTIUM(II) COORDINATION, I 187 FIG. 1. (A) Normalized EXAFS spectra and Fourier transforms of SrO(s) collected at 10 K, 70 K, and room temperature (RT). Solid lines, experimental data; dashed lines, nonlinear least squares fit for the first coordination shell including a third cumulant (C3 ) term (numerical results in Table 1). (B) Spectra of 10−3 M SrCl2 solution collected at 10 K and RT. Solid lines, experimental data; dashed lines, fit for the first coordination shell without C3 term; grey lines, fit including C3 term (Table 1). in Fig. 1A. At lower temperature, EXAFS amplitudes increase significantly, especially at higher photoelectron energy (k). The largest difference between fitted and crystallographic distance (1RSr-O ) was 0.04 Å for the room temperature spectrum fit with C3 ; all other fits gave 1RSr-O of 0.01 or 0.02 Å. The magni- tude of these differences is near to a typical error in EXAFS bond length determination (±0.01–0.02 Å). The expected result of ignoring a significant third cumulant would be a decrease in backscatterer distance with increasing temperature, which is not observed (above error in the bond length fit). In the spectrum 188 O’DAY ET AL. TABLE 2 EXAFS Fit Results for Crystalline Strontium Reference Compounds as a Function of Temperature XRDa Sr–Z EXAFS (10 K) Nb R (Å) R (Å) σ2 EXAFS (70 K) (Å2 ) R (Å) σ2 EXAFS (RT) (Å2 ) R (Å) σ 2 (Å2 ) 1E 0 (eV)c S02d 0.0066 0.0071 2.57 3.96 f 0.0117 0.0125 −7.8 −7.8 −10.6 0.92 Sr–O Sr–Sr 6 12 2.570 3.634 2.58 3.68 0.0066 0.0067 SrO(s)e 2.58 3.67 Sr–O Sr–Ti Sr–Sr Sr–O 12 8 6 24 2.758 3.378 3.900 4.777 2.74 3.38 3.91 4.77 0.0071 0.0034 0.0035 0.0066 SrTiO3 (s)g 2.74 3.39 3.91 4.78 0.0092 0.0035 0.0043 0.0065 2.72 3.39 3.92 4.78 0.0186 0.0080 0.0117 0.0121 −11.0 −10.8 −10.9 1.1 Sr–O 1 2 2 2 2 1 2 1 2 2 2 2 4 2.62 0.0084 SrCO3 (s)h 2.61 0.0087 2.61 0.0136 −6.7 −9.0 −6.9 1.04 3.05 0.0018 3.05 0.0017 3.05 0.0073 4.11 0.0021 4.11 0.0025 4.14 0.0083 4.24 4.90 0.0005 0.0031 4.24 4.90 0.0012 0.0037 4.28 4.92 0.0063 0.0150 2.58 0.0118 2.57 0.0158 −7.2 −8.9 1.0 3.21 3.37 4.09 0.0027 0.0024 0.0069 3.20 3.36 4.06 0.0062 0.0060 0.0232 Sr–C Sr–Sr Sr–Sr Sr–Sr 2.5647 2.5780 2.6401 2.5615 2.7298 3.0263 3.0637 3.4071 3.5152 4.0968 4.1156 4.2592 4.9024 2.5520 2.5608 2.6340 2.6655 2.7250 3.0256 3.0365 3.4011 3.5423 4.0854 4.1171 4.2359 4.8934 SrHPO4 (s)i Sr–O Sr–P Sr–Sr 1 1 1 1 1 1 1 1 1 1 1 1 2.446, 2.516 2.559, 2.573 2.587, 2.575 2.589, 2.581 2.620, 2.591 2.676, 2.608 2.897, 2.634 3.074, 2.693 3.370, 3.211 3.448, 3.597 4.087, 3.933 4.090, 4.090 a Crystallographic distances from X-ray diffraction (XRD) studies; for all compounds, Sr–O scattering was not considered beyond the first coordination shell. of backscatterers from crystal structure; fixed in EXAFS fits. c 1E for fits at 10 K, 70 K, and RT, respectively. 0 d Amplitude reduction factor (S 2 ) was estimated at 10 K and fixed at higher temperatures. 0 e SrO(s): Primak et al. (25). f Increase in bond length from hydration and unit cell expansion. g SrTiO (s): Scheel (28); Hutton and Nelmes (29). 3 h Two X-ray structure determinations were found for SrCO (s): 1st R column, Pannhorst and Löhn (26); 2nd R column, DeVilliers (27). Mean Sr–O distance 3 from XRD = 2.613 Å, median distance = 2.606 Å. i SrHPO (s): Boujada et al. (30). There are 2 positions for Sr in SrHPO (s), with Sr–Z for each site given by the R pairs. Mean Sr–O distance = 2.639 Å, median 4 4 distance = 2.590 Å. Additional Sr–P scattering between 3.60–3.92 Å was not significant in the EXAFS. b Number for SrO(s) at room temperature (Fig. 1A), there is a peak shift to longer distance for second-neighbor Sr–Sr scattering, probably from partial hydration of the compound (see the next section), which did not affect the Sr–O first-shell distance. Strontianite (SrCO3 (s)) has an aragonite structure in which strontium is bonded by nine oxygen atoms at five different Sr–O distances from 2.552 to 2.725 Å determined from XRD (Table 2) (26, 27). In the first-shell cumulant analysis, all five Sr–O crystallographic distances were used for fits with and without C3 , but R was constrained to be a single variable in the least-squares fit (Table 1). At all three temperatures, the differences between fitted and crystallographic distances (1RSr-O ) were from −0.03 STRONTIUM(II) COORDINATION, I to 0.00 Å. These differences were still much less than the variance in Sr–O bond lengths (i.e., the static disorder in the Sr–O shell) determined by XRD (=0.173 Å). In no case was there a statistically significant improvement in the fit with C3 . This analysis points out that for compounds in which static disorder is large, anharmonic effects can be masked by static disorder and EXAFS fit results will differ depending on the number of absorber–backscatterer shells assumed in the fit model. For strontium in aqueous solution, there is disagreement in the literature regarding Sr–O interatomic distances and the number of coordinating water ligands (see Discussion). The spectrum for Sr2+ (aq) was fit with FEFF theoretical phase-shift and amplitude functions calculated for an expanded SrO(s) atomic cluster (RSr-O = 2.62 Å) and initially assumed 6 first-shell oxygen atoms. At low temperature (10 K), 1RSr-O for fits with and without C3 range from −0.02 to 0.05 Å (Table 1). There are no differences in the fitted σ 2 values. Comparison of fits at room temperature, however, show a contraction of the Sr–O distance by 0.05 Å if C3 is ignored (from 1RSr-O = + 0.03 Å with C3 to 1RSr-O = −0.02 Å without C3 ). This contraction with temperature can be seen in Fig. 1B as a shift to lower R of the first-shell peak in the RSF. Including the C3 term in the fit at room temperature corrects the distance contraction and makes it consistent with the low temperature result (RSr-O = 2.65 − 2.67 ± 0.02 Å). Varying the amplitude reduction factor (S02 ) and fixing NSr-O at 6 resulted in fitted S02 values of 1.53–1.54 (with and without C3 ) at 10 K, indicating that NSr-O must be higher (≈9–10). To estimate NSr-O for Sr2+ (aq), values of S02 were estimated from the other reference compounds with known NSr-O , including strontium-substituted zeolite where first-shell ligands are both oxygen and water (see below). Fits of the Sr2+ (aq) EXAFS sprectrum with S02 fixed between 0.90 and 0.95 indicated 9–10 ± 1 oxygen atoms around strontium in solution. As shown in Fig. 1B, there is no evidence for backscatterers beyond the first Sr–O shell in the Sr2+ (aq) spectrum. At these dilute solution concentrations (10−3 M), significant backscattering from chlorine in SrCl2 aqueous complexes is not expected. Also, there is no evidence that freezing of the aqueous solution resulted in significant differences in the EXAFS fit results for R and N once anharmonic effects were accounted for. Strontium EXAFS of Crystalline Compounds To investigate a range of coordination environments around strontium, normalized EXAFS spectra (k = 2 to 12–13 Å−1 ) of four crystalline strontium compounds were analyzed for first and higher shell near-neighbor atoms (Table 2). In two of the compounds (SrO(s)), examined above, and strontium titanate (SrTiO3 (s)), the crystal structures are cubic at room temperature and the strontium site is highly symmetric (25, 28, 29). In SrO(s), strontium is 6-coordinated and the Sr–O bond length is 2.570 Å. In SrTiO3 (s), oxygen coordination is 12, giving a considerably longer Sr–O bond distance (2.758 Å). Multiplescattering is significant in these two compounds beyond ≈4 Å 189 because of the high symmetry and is amplified at low temperature. In least-squares fits, only single-scattering was included for the first two atomic shells for SrO(s) and the first four shells for SrTiO3 (s), and fits were made to Fourier-filtered data from 0–4.5 Å (uncorrected for phase shift). Best fits are in agreement with crystallographic data to within R ± 0.02 Å or better for almost all shells at all three temperatures, indicating that effects from anharmonic disorder are minor (Table 2). The most notable disagreement between crystallographic and EXAFS distances is for second-neighbor Sr–Sr distances in SrO(s). At room temperature, the fitted RSr-Sr distance is 3.96 Å, significantly longer than crystallographic RSr-Sr = 3.634 Å. Strontium oxide is hydroscopic, and we suspect that hydration occurred during sample preparation, resulting in expansion of the local structure. For both compounds, N was fixed at crystallographic values in order to estimate values for σ 2 and S02 . Fitted σ 2 values are within ≈20% at 10 and 70 K and systematically increase for all shells at room temperature (Table 2). Fitted values of S02 were 0.92 and 1.1 for SrO(s) and SrTiO3 (s), respectively. Tests made by fixing N and varying S02 showed ≈5% variance as a function of temperature. The range in the value determined for S02 between the two compounds may arise from the difference in coordination number (NSr-O = 6 vs 12), but is similar to typical variance (5–10%) reported for S02 (16, 22). Two compounds were analyzed in which the strontium site is highly disordered, strontianite (SrCO3 (s), discussed above) and strontium hydrogen phosphate (α-SrHPO4 (s)). In SrCO3 (s), the structure is orthorhombic and strontium is 9-coordinated by oxygen. Two different structure determinations have been published (26, 27) which give slight differences in calculated Sr–Z interatomic distances (Table 2). The structure of α-SrHPO4 (s) is triclinic with 8-coordinated strontium in two different crystallographic sites, giving rise to a total of 16 unique Sr–O distances from 2.446 to 3.074 Å (Table 2) (30). Despite the large static disorder, the EXAFS spectra could be fit well for both compounds at low and room temperatures as a single first-coordination shell of oxygen atoms: RSr-O = 2.62 Å for SrCO3 (s) and RSr-O = 2.57– 2.58 Å for α-SrHPO4 (s) (Table 2 and Fig. 2). The fitted EXAFS distances are near either the mean or the median of the interatomic distances calculated from XRD studies (with N fixed; Table 2). The value of σ 2 determined in the one-shell Sr–O fit for SrCO3 (s) is slightly higher than σ 2 determined in the first-shell cumulant analysis, which reflects the averaging of Sr– O distances (compare SrCO3 (s) results in Tables 1 and 2). As shown in the cumulant analysis above, any thermal anharmonic effects are masked by the large amount of static disorder. Backscattering from low-Z atoms (carbon and phosphorus) in these two compounds beyond the first shell is relatively weak in the EXAFS due to their low scattering amplitudes and a range of Sr–Z C,P distances (Table 2). For Sr–C and Sr–P scattering, second-neighbor distances (at 3.0–3.4 Å) are fit in the spectra, but scattering at >3.4 Å is not seen in the EXAFS, probably because of low amplitude and cancellation effects from overlapping scattering functions. Strong backscattering from strontium 190 O’DAY ET AL. FIG. 2. (A) Normalized EXAFS spectra and Fourier transforms of crystalline SrCO3 (s) collected at 10 K, 70 K, and room temperature (RT). (B) Spectra of α-SrHPO4 (s) at 12 K and RT. Solid lines, experimental data; dashed lines, nonlinear least squares fits (Table 2). atoms at >4 Å is found for both compounds but decreases significantly at room temperature (Fig. 2). Notably, backscattering is not significant in the EXAFS spectrum when values of σ 2 increase to greater than ≈0.025 Å with increasing temperature (Table 2). Precipitated Strontianite (SrCO3 ) Absorption spectra were collected at room temperature for strontium precipitated from solutions supersaturated with respect to SrCO3 (s) in the presence and absence of two finegrained minerals, kaolinite and goethite. In all three samples, strontium was found to be coordinated in a local environment identical to that of strontium in crystalline SrCO3 (s) (Fig. 3 and Table 3). Least-squares fits of the EXAFS spectra showed no significant differences in R or σ 2 (N fixed) among crystalline and precipitated SrCO3 (s). Backscattering from metal atoms in kaolinite (aluminum or silicon) or goethite (iron) could not be fit in the spectra of the precipitated samples. Fits in which σ 2 was fixed (values from crystalline SrCO3 (s) at room temperature) and N was allowed to vary for each shell showed no significant change (N ± 1) in the refined value for N . This result indicates that the local strontium coordination in the fresh precipitates is similar to that of the crystalline compound and that cluster size STRONTIUM(II) COORDINATION, I 191 FIG. 3. Normalized EXAFS spectra and Fourier transforms of SrCO3 precipitated from aqueous solution homogeneously (ppc-h) and in the presence of fine-grained kaolinite (ppc-k) or goethite (ppc-g). These spectra are compared with that of crystalline SrCO3 (s) from Fig. 2. All spectra were collected at room temperature. Solid lines, experimental data; dashed lines, nonlinear least squares fits (Table 3). is large enough to reproduce the EXAFS scattering of the crystalline compound (i.e., at least nanometer-sized; see Thompson et al. (31)). The spectrum for strontium precipitated in the presence of goethite (SrCO3 (ppc-g)) is of poor quality, partly due to low signal-to-noise from fluorescence quenching by iron in goethite. For this spectrum, only Sr–O and Sr–C shells were fit. It is likely, however, that scattering from strontium atoms at greater TABLE 3 EXAFS Fit Results for SrCO3 Precipitated Samples Nb σ 2 (Å2 ) Sr–Za R (Å) Sr–O Sr–C Sr–Sr Sr–Sr Sr–Sr SrCO3 precipitate, homogeneous (ppt-h) 2.61 9 0.0163 3.04 3 0.0062 4.14 4 0.0100 4.29 2 0.0058 4.90 4 0.0182 Sr–O Sr–C Sr–Sr Sr–Sr Sr–Sr 2.61 3.03 4.14 4.29 4.90 Sr–O Sr–C SrCO3 precipitate w/goethitec (ppt-g) 2.60 9 0.0120 3.04 3 0.0025 SrCO3 precipitate w/kaolinite (ppt-k) 9 0.0146 3 0.0064 4 0.0092 2 0.0058 4 0.0167 1E 0 (eV) −6.6 Strontium in Natural Zeolites −6.3 −8.4 Z = backscattering atom. factor (S02 ) = 1.04; N fixed in fits. c In precipitate with goethite, background noise obscures scattering beyond first two shells. a b Scale distances is present but obscured by noise (note, for example, shoulders on major EXAFS oscillations at k = 6 and 8.3 Å−1 that are not fit in the spectrum for SrCO3 (ppc-g) in Fig. 3). However, fits including strontium atoms at 4.1 and 4.3 Å were not found to be significantly better than those with only oxygen and carbon shells. This spectrum points out the effects of noise on the interpretation of EXAFS spectra. Fourier transformations of noisy spectra often produce peaks in the RSF that could be interpreted as backscattering from atoms beyond the first coordination shell (e.g., peaks in the RSF of SrCO3 (ppc-g) at ≈5 Å in Fig. 3). Changing spline nodes in background subtraction and using different limits for the k-range of the Fourier transform changed the position of artifact RSF peaks and were useful for distinguishing real from spurious backscatterers (17, 32). Incorporation of strontium into the channels of three natural zeolites was examined by EXAFS to determine the coordination of hydrated strontium and its location in the mineral structure, and to evaluate backscattering from aluminum and silicon atoms in the mineral framework. All three samples were identified by XRD as heulandite, a calcium-rich zeolite in which Sr2+ substitutes naturally for Ca2+ in concentrations on the order of 0–9 wt% SrO (33, 34). Chemical analyses are given in Table 4, which show bulk strontium concentrations of 2000–4500 ppm. The EXAFS spectra of the zeolites at both room and low temperature have asymmetric oscillations which indicate backscattering from second-neighbor metal atoms, either silicon or aluminum, although at room temperature these features are significantly damped (Fig. 4). For quantitative fits to the EXAFS spectra, 192 O’DAY ET AL. TABLE 4 Bulk Chemical Analyses of Natural Zeolites Weight % oxide Zeolite #1 (Z1) Zeolite #2 (Z2) Zeolite #3 (Z3) SiO2 TiO2 Al2 O3 Fe2 O3 MnO MgO CaO Na2 O K2 O P2 O5 LOI(+) LOI(−) LOI TOT 69.385 0.011 19.153 0.058 0.002 0.039 8.559 1.688 0.421 0 3.67 12.43 16.09 69.242 0.011 19.4 0 0.005 0.011 8.736 1.722 0.408 0.001 3.74 12.41 16.14 67.731 0.008 19.019 0.008 0 0.056 7.054 1.471 1.845 0.001 3.67 11.83 15.50 Nb Zr Sr U Rb Pb Ga Zn Cu Cr Ti V Ba Trace elements (ppm) 0.2 0 0 0 2013.4 2244.2 0.9 0.8 5.5 9.4 0.5 0 0.8 1 1.4 1.1 9.9 6.7 0 0 0.01 0 7.3 6.2 1221 987.1 0 338 4478.5 0 9.8 0 1.3 3.7 24.8 1.8 0 8.3 5144.2 Note. All three zeolites were identified as heulandite by X-ray diffraction. LOI is loss on ignition. Formulas based on 18 framework O: Z1: Ca0.89 Na0.32 K0.05 Sr0.03 Ba0.01 Al2.20 Si6.75 O18 • 5.56H2 O Z2: Ca0.91 Na0.28 K0.05 Sr0.04 Ba0.01 Al2.22 Si6.72 O18 • 5.56H2 O Z3: Ca0.75 Na0.28 K0.23 Sr0.07 Ba0.05 Al2.21 Si6.68 O18 • 5.28H2 O theoretical phase-shift and amplitude reference functions were calculated using the crystal structure determination of heulandite (Table 5) (35, 36). In this structure, there are three unique calcium positions in the zeolite channels and each calcium is coordinated by five water molecules and three framework oxygen atoms. Framework oxygen atoms coordinating calcium tend to be associated with aluminum-substituted tetrahedra and all three sites exhibit a high degree of static disorder in Ca–O bond lengths (35). In least-squares fits to the EXAFS data, we considered strontium substitution in each of the three calcium sites and an initial assumption of eight-fold coordination of strontium by oxygen. For all three zeolite samples, fit results were similar regardless of the temperature of data collection (Table 5). First-shell Sr–O distances of 2.62 Å at low temperature and 2.59–2.60 Å at room temperature are similar to values derived above for SrCO3 (s). The consistent decrease in interatomic distance with increasing temperature may indicate a small anharmonic contribution to EXAFS scattering as noted above. A series of fits were made with first-shell N varied and σ 2 fixed on typical values obtained from the strontium reference compounds. In addition, filtered first-shell fits were examined in which both N and σ 2 were varied simultaneously. Final fits were done assuming eight-coordinated strontium (N fixed) based on known coordination of calcium in heulandite. Overall, these results indicate variation in firstshell N of ±1 among the different fit models, with no major differences noted for the three samples. Second-neighbor scattering was fit with two shells of silicon or aluminum atoms (which cannot be distinguished from each other as backscatterers). The closer Sr–Si/Al shell at 3.47–3.49 Å has significant amplitude and is well fit by one scattering shell (Fig. 4). Mismatch between the data and fit for the Sr–Si/Al shell at 4.09–4.15 Å is probably a result of static disorder and weaker backscattering at this longer distance. Scattering from nearest-neighbor strontium or calcium atoms within channels would occur at >4.7 Å and was not apparent in the spectra at low or room temperature. Similar to the crystalline compounds, values of σ 2 systematically increase with increasing temperature of data collection. However, fitted values of σ 2 for Sr–Al/Si scattering (N = 2) are <0.0025 Å2 at room temperature and scattering for both shells is still apparent in the RSF, although significantly damped for the shell at 4.09–4.15 Å (Fig. 4). This result shows that low numbers of aluminum or silicon atoms present at distances between 3.5 and 4 Å from central strontium can contribute significant scattering amplitude to the EXAFS at room temperature, regardless of Sr–O static disorder or partial strontium hydration. Backscattering from one or two aluminum or silicon atoms at >4.2 Å from strontium would probably not be significant at room temperature. FIG. 4. Normalized EXAFS spectra and Fourier transforms of strontium in natural zeolites (identified by XRD as heulandite) collected at 14 K and room temperature (RT). Bulk chemical analyses for Zeolite #1 (Z1), Zeolite #2 (Z2), and Zeolite #3 (Z3) given in Table 4. Solid lines, experimental data; dashed lines, nonlinear least squares fits (Table 5). 193 STRONTIUM(II) COORDINATION, I TABLE 5 EXAFS Fit Results for Strontium-Bearing Zeolite (Heulandite) Samples EXAFS (10 K) Sr–Z Na R (Å) EXAFS (RT) σ2 (Å2 ) 1E 0 (eV) Na R (Å) σ 2 (Å2 ) 1E 0 (eV) 8 2 2 2.60 3.49 4.14 0.0144 0.0139 0.0167 −10.0 8 1.9 2.2 2.60 3.47 4.12 0.0140 0.0114 0.0163 −10.1 8 2.3 2 2.59 3.47 4.15 0.0148 0.0170 0.0202 −10.7 Zeolite 1 (Z1) Sr–O Sr–Al/Si Sr–Al/Si Sr–O Sr–Al/Sib Sr–Al/Sib 8 1.9 2.2 2.62 3.48 4.10 0.0091 0.0059 0.0075 Sr–O Sr–Al/Sib Sr–Al/Sib 8 2.3 2 2.62 3.48 4.09 0.0096 0.0056 0.0070 Na1 Zeolite 2 (Z2) −9.6 Zeolite 3 (Z3) −8.6 XRD: Heulandite Ca3.7 Na1.3 K0.8 Al8.9 Si27.1 O72 • 21.4H2 Oc Ca2 K3 Site N R (Å) N R (Å) N R (Å) Ca–Od 2 1 2 2 2 2.361 2.540 2.648 2.906 3.121 2 1 2 1 2 2.390 2.535 2.563 2.596 2.715 2 2 2 2 2.909 2.916 3.112 3.133 Ca–Si/Al 2 2 2 3.500 3.507 4.281 2 2 3.263 3.867 2 2 2 2 3.678 3.893 3.928 4.195 S02 estimated at 0.92 based on NSr-O = 8; N fixed in least-squares fit unless noted. N varied at 10 K; N fixed on 10 K value at RT. c Interatomic distances from the XRD study of Gunter et al. (36) based on space group C2/m with three unique nonframework cation positions. d Ca–O distances not listed beyond the first coordination shell. a b DISCUSSION Static and Anharmonic Disorder in Strontium Compounds The analysis of the EXAFS spectra of strontium reference compounds given above demonstrates the relative influences of static versus thermal vibrational disorder on the quantitative determination of local bonding. Figure 5 summarizes the fit results for reference compounds (Table 2) and zeolites (Table 5) by plotting fitted interatomic distances and Debye– Waller factors (σ 2 ) as a function of backscatterer Z and temperature of data collection. Anharmonic vibrational disorder between 10 and 300 K has a negligible effect on bond distance determination for nonhydrated strontium compounds. Although systematic decreases in first-shell Sr–O bond length with increasing temperature were noted in most compounds (Fig. 5), the amount of bond length reduction was generally less than the error in the analysis (i.e., <0.02 Å), indicating that the anharmonic vibrational effects were small. As shown above, the addition of a higher order cumulant term (C3 ) in the EXAFS fitting did not significantly improve the fit results. For compounds with large first-shell static disorder (e.g., SrCO3 (s), α-SrHPO4 (s)), average bond lengths determined by EXAFS analysis were near the mean or the median of first-shell distances determined by XRD and varied within error with increasing temperature. For Sr–Z shells beyond the first shell, both increases and decreases in interatomic distances were observed with increasing temperature (Fig. 5), indicating no systematic thermal vibrational effects on distance determination. The differences in fitted bond lengths may simply reflect errors in the EXAFS and XRD analyses. Phase changes with temperature appear to have no effect on strontium coordination in this set of compounds. For example, SrTiO3 (s) undergoes a ferroelastic phase transition (so-called “soft mode”) from a cubic to a tetragonal space group at Tc = 105 K. Previous work has shown that this does not affect unit cell volume or significantly change atom positions around strontium, consistent with our fit results (29, 37, 38). Because N and σ 2 are positively correlated to a high degree in the EXAFS amplitude function and do not vary independently, it is desirable to constrain one or the other for a given shell of atoms during fitting. Previous studies have investigated the transferability of σ 2 among structurally similar compounds (22, 39). As shown in Fig. 5 for strontium compounds, there is a 194 O’DAY ET AL. Through cross-fits of reference compounds, we estimate errors in N at ±1 for strontium second-neighbor atoms and ±2 for light atoms (C, P, Si, Al) up to ≈4 Å from central strontium when σ 2 is fixed on a value from a similar RSr-Z backscatterer. Analysis of the reference compounds also showed that at values of σ 2 greater than ≈0.025 Å2 for N < 12 and RSr-Z > 3 Å, backscattering was generally not significant above noise levels in the spectra. Values of σ 2 derived from EXAFS fits in unknown systems that fall significantly above this value are suspect and probably result from simultaneously varying N and σ 2 during fitting. Comparison of the EXAFS spectra of precipitated and crystalline SrCO3 (s) at room temperature shows little difference among the spectra even though the precipitated samples were run as wet pastes (Fig. 3). There was no evidence of backscattering from atoms unique to kaolinite or goethite in the heterogeneously precipitated samples. Fitted σ 2 values (N fixed) for precipitated SrCO3 (s) spectra are within 20%, and fitted interatomic distances are within ±0.02 Å, of those determined for crystalline SrCO3 (s) at room temperature for all three spectra. This comparison indicates that new SrCO3 precipitates can form rapidly from supersaturated solutions with a local molecular structure around strontium identical to that of crystalline SrCO3 (s). Bonding Environment of Aqueous Strontium FIG. 5. Debye–Waller factors (σ 2 ) from EXAFS fits of crystalline compounds (Table 2) and zeolites (Table 5) shown as a function of interatomic absorber–backscatterer (Sr–Z) distance (no C3 terms were included in these fits). Open symbols, data collected at low temperature (10–15 K and 70 K); closed symbols, data collected at room temperature. In the middle panel, backscattering atoms are C, P, Si, or Ti. little difference in σ 2 values between 10 and 70 K, and then a large increase between 70 K and ambient temperature. There is a general tendency for larger increases in σ 2 with longer Sr– Z interatomic distance among the same set of Sr–Z scatterers, with some exceptions. Large increases in σ 2 with increasing temperature were found for Sr–Sr scattering in α-SrHPO4 (s) at R = 4.09 Å and in SrCO3 (s) at R = 4.91 Å. In general, our empirically determined σ 2 values fall within narrow ranges for a given Sr–Z pair at either low or ambient temperature. Therefore, these values for σ 2 are useful for estimating N in unknown systems in which strontium is bonded in a similar environment. In the analysis reported here, the range of σ 2 for first-shell NSr-O from 6 to 12 is between 0.0066 and 0.0118 Å2 at low temperature and between 0.0117 and 0.0186 Å2 at room temperature. Varying σ 2 within this range corresponded to differences in fitted values of N (for fixed σ 2 ) of up to about ±2. Alternatively, treating both N and σ 2 as unknowns in first-shell Sr–O fits of the reference compounds resulted in variation in N of about ±1 from known values (once S02 had been estimated and fixed). Beyond the first shell, the estimated error in N depended on the number of atoms present and backscatterer Z . The coordination and bonding of strontium in aqueous solution (Sr2+ (aq)) has been determined previously by EXAFS, XRD, and X-ray scattering, with disagreement in the results (summarized in Table 6). Clarification of the coordination environment is important because recent EXAFS studies (see Sahai et al. (13)) suggest that when strontium sorbs from solution to mineral surfaces, its hydration sphere is retained and its coordination is similar to that of Sr2+ (aq). In a recent EXAFS TABLE 6 Determinations of Sr–O Interatomic Distance for Sr2+ (aq) at Room Temperature Solution Sr(ClO4 )2 Sr(NO3 )2 Sr(NO3 )2 SrCl2 SrCl2 SrCl2 Sr(CF3 SO3 )2 SrCl2 SrCl2 [Sr2+ ]aq (M) R (Å) N σ 2 (Å2 ) Method Ref. 0.1 0.2 0.05 0.1 0.1, 3.0 0.001 2.61 2.62 2.62 2.61 2.64 2.65 8a 7.3 8.93 8.3 10.3 9–10 0.0116 n.r. 0.012 0.0115 0.021 0.011 EXAFS EXAFS EXAFS EXAFS EXAFS EXAFS 1 2 3 4 5 6 0.82 1.5 2.0 2.64 2.64 2.64 8.1 8a 8a n.a. n.a. n.a. LAXS XRD XRD 1 7 7 Note. References: 1. Persson et al. (2); 2. Pfund et al. (46); 3. Axe et al. (9); 4. Parkman et al. (8); 5. D’Angelo et al. (7); 6. This study; 7. Caminiti et al. (47). EXAFS, Extended X-ray absorption fine structure; LAXS, Large angle X-ray scattering; XRD, X-ray diffraction. n.a., not applicable; n.r. not reported. a Fixed parameter. STRONTIUM(II) COORDINATION, I analysis of Sr2+ (aq) (0.1 and 3 M SrCl2 ), D’Angelo et al. (7) quantified contributions to EXAFS amplitudes from two factors overlooked in previous analyses, contributions from hydrogen scattering and multielectron excitations (shake-up effects). These factors account for only a few percent of the total EXAFS amplitude and are masked in crystalline compounds by backscattering from atoms beyond the first shell, but they have been noted in the EXAFS of ions in solution (7, 11, 12). In these studies, these contributions were accounted for by employing a background subtraction model for multielectron absorbance and theoretical molecular dynamics calculations for the fit model. Using this analysis, D’Angelo et al. (7) determined a coordination of ≈10 water ligands for hydrated Sr2+ (aq). In addition, their study noted that fits excluding these effects reduced the fitted Sr–O interatomic distance from 2.64 to 2.62 Å even when anharmonicity was included in the fit (their skewness parameter, β; see D’Angelo et al. (11)). As shown in Table 6 and pointed out by D’Angelo et al. (7), most other EXAFS studies of Sr2+ (aq) at room temperature reported shorter Sr–O distances (2.61–2.62 Å) and lower values for NSr-O (7–9) than those of D’Angelo et al. (7). Determinations of Sr2+ (aq) coordination by X-ray diffraction or scattering methods (Table 6) eliminated the distance contraction, but were less sensitive than EXAFS to quantification of N because of overlap among all pairs of scattering atoms. Hence, N is often assumed and not determined in XRD solution studies. D’Angelo et al. (7) also noted that there is considerable uncertainty in their determination of N and R for Sr–H scattering. Most EXAFS analyses ignore contributions from hydrogen atoms because they are usually very small compared to the total EXAFS scattering amplitude. In our EXAFS fits of Sr2+ (aq), we did not explicitly include hydrogen scattering or multielectron absorption. Rather, we used a semi-empirical analysis in which these small contributions to EXAFS scattering were accounted for by calculating theoretical Sr–O phase-shift and amplitude functions with FEFF and then estimating S02 (by fixing N on known values) in fits to the compounds discussed above. We obtained a hydration number of 9–10 ±1 from EXAFS fits of Sr2+ (aq) once S02 was estimated. We also found that including the third cumulant term (C3 ) in the fit of the Sr2+ (aq) spectrum at room temperature sufficiently accounted for anharmonic vibrational disorder in the distance determination. Including C3 increased the fitted Sr–O bond length from 2.60 to 2.65 Å for Sr2+ (aq) at room temperature. At low temperature, fits with and without C3 were within error (2.67 vs 2.65 Å, respectively) and in agreement with room temperature data fit with C3 . This suggests that thermal vibrational disorder is sufficiently damped at low temperature to not require fitting with C3 . These results for Sr2+ (aq) (R = 2.65 ± 0.02 Å and N = 10 ± 1) are in agreement with those of D’Angelo et al. (7). We did not find evidence for backscattering from atoms beyond the first coordination shell in the frozen aqueous sample, although it is not clear whether a crystalline ice structure forms upon sample quenching that might affect strontium coordination. 195 Strontium in Heulandite The coordination of strontium in natural zeolites provides a good test case for EXAFS analysis and is a good structural analog for strontium sorbed to surfaces because, if strontium is occupying a channel cation site, it will have both oxygen and water ligands in its coordination shell. In addition, the coordination site for strontium in natural heulandite and isostructural zeolites has not been investigated previously with EXAFS spectroscopy. Refinements of the crystal structure of heulandite (and the isostructural zeolite clinoptilolite) have identified two to four distinct channel cation sites (35, 36, 40, 41). The heulandite structure refined by Gunter et al. (36) has a composition nearly identical to those of this study (Table 4). Gunter et al. (36) identified three channel cation sites labeled Na1, Ca2, and K3 (Table 5). Site Na1 lies within a ten-member tetrahedral ring channel (A channel) and is coordinated by nine oxygen atoms. Sites Ca2 and K3 both lie in different eight-member ring channels (B and C channels, respectively) and are each coordinated by eight oxygen atoms. These three sites are distinguished by different average cation–oxygen distances and by the coordination number and distances to second nearest-neighbor silicon and/or aluminum atoms. Armbruster (40) assigned strontium in clinoptilolite to the K3 site. The coordination of strontium in the three heulandite samples summarized in Table 5 suggests that it occupies the Ca2 site of Gunter et al. (36) instead of K3. The local coordination environment of strontium in the Ca2 site is shown in Fig. 6. Although the coordination number of oxygen atoms is identical or similar among Na1, Ca2, and K3, these sites exhibit FIG. 6. Crystallographic projection along [001] in heulandite showing the local coordination of Sr in the Ca2 site of Gunter et al. (36). 196 O’DAY ET AL. very different average cation–oxygen distances (2.73, 2.56, and 3.02 Å, respectively). The Sr–O distance derived from EXAFS (2.60 Å) is closest to the average Ca–O distance in Ca2 (2.56 Å) and is too small to be consistent with Na1 or K3. In fact, the average cation–oxygen distance in K3 is considerably longer than any average Sr–O distance observed in strontium–oxygen compounds (e.g., Tables 1 and 2), including the Sr–zeolite brewsterite (42). The second nearest-neighbor Si/A1 coordination number determined from EXAFS (N = 1.9–2.3 ± 2) is also near that of Ca2 (N = 4) and is considerably smaller than that of Na1 and K3 (N = 6 and 8, respectively). The two sets of Sr–A1/Si distances derived from EXAFS fits are each 0.21– 0.29 Å greater than the two Ca2–A1/Si distances in heulandite (Table 5). This difference in RSr-Al/Si derived from EXAFS is consistent with the difference in cation–oxygen bond lengths reported for Sr2+ and Ca2+ in solution (= 0.20–0.24 Å) (1, 2) and consistent with the expected increase in RSr-Al/Si if strontium occupied the Ca2 site. Backscattering from atoms at distances characteristic of the other cation sites is not evident in the EXAFS spectra. The fitted strontium EXAFS distances cannot be derived by simply lengthening or distorting bonds at the other two cation sites and are not a result of overlapping backscattering and amplitude cancellation from the Sr–Al/Si shells at these positions. The preference of strontium for the Ca2 site over either the K3 or Na1 site is reflected in the ion exchange behavior of clinoptilolite. Strontium incorporation into heulandite and clinoptilolite through ion exchange has received considerable attention as a means of attenuation of 90 Sr released from nuclear facilities and waste sites (43, 44). The cation selectivity sequence for clinoptilolite is Cs > K > Sr ≈ Ba > Ca > Na À Li (43). The ease with which Sr2+ exchanges for Na+ and Ca2+ reflects its affinity for Ca2, which commonly has high occupancies of these ions (40). In contrast, the very different coordination shell of the K3 site favors large ions like K+ and Cs+ over Sr2+ , leading to limited Sr2+ exchange for ions on this site. Nearly complete exchange of Sr2+ for Na+ in clinoptilolite (and presumably heulandite) was realized experimentally (e.g., Pabalan and Bertetti (45)), requiring a fraction of strontium to occupy the Na1 site as the total occupancy of Ca2 can be no more than 0.5 (41). However, the lower strontium concentrations in Table 4 are typical of natural samples and are similar to the concentrations likely during ion exchange of clinoptilolite and heulandite with groundwater containing radioactive waste. SUMMARY Quantitative determination by EXAFS of the coordination environment around large, strongly hydrated cations such as strontium requires careful parameterization of variables in the phase-shift and amplitude functions of the EXAFS equation. In this study, we used a combination of ab initio theoretical calculations and empirical parameterization of crystalline, precipitated, and hydrated strontium reference compounds to investigate Sr–O bond anharmonicity and to estimate values for variables that contribute to EXAFS amplitudes (N , σ 2 , and S02 ). For all solid compounds examined, first-shell bond anharmonicity was not significant above the error in the EXAFS analysis and in many compounds was masked by the static disorder around strontium. The only case in which a significant decrease in Sr–O distance was observed with increasing temperature, and for which a third cumulant term added a small correction to the distance determination, was for dilute strontium in aqueous solution. This suggests that including higher-order cumulant terms in the EXAFS analysis (C3 and/or C4 terms) is generally not needed and increases the number of adjustable parameters in EXAFS fitting. Including more variables in the EXAFS equation is particularly dangerous in the analysis of noisy spectra without strong backscatterers beyond the first coordination shell where weak or spurious features can be erroneously fit. As shown here, the identification of weak backscattering atoms such as carbon, phosphorus, aluminum, and silicon beyond the first Sr–O shell is difficult due to their low scattering power and requires high-quality data to unambiguously determine their presence and structural position relative to strontium. Low-temperature data collection enhanced scattering from more distant atoms and no structural differences were found among spectra collected at low and room temperature. It is important to note, however, that even weak scatterers were clearly seen in the EXAFS spectra at room temperature at distances of <4 Å and low temperature served only to amplify these features. Empirical parameterization of Debye–Waller factors using reference compounds showed that limited ranges of σ 2 for a given temperature and strontium–backscatterer pair can be derived. This provided a useful constraint for determining N in unknown spectra and served to improve the practical error in fitted N . The results described here indicate that EXAFS determinations of strongly hydrated compounds and surface complexes may underestimate first-shell interatomic distances and coordination numbers if minor phase-shift and amplitude contributions to the EXAFS spectra are not accounted for, either theoretically or empirically. This observation is significant for the interpretation of strontium as a hydrated surface complex, exchanged ion, or precipitate in the quantitative formulation of surface complexation models that seek to combine molecular structural results with macrocscopic uptake behavior. ACKNOWLEDGMENTS This work benefited from helpful discussions with Bruce Ravel, John Rehr, and Paul McMillan. Thanks to Glenn Waychunas and Gordon Brown, Jr., for supplying reference compounds. The manuscript was improved by the helpful comments of Hillary Thompson. This work was supported by the Department of Energy (DOE) Environmental Management Science Program (Grant 55249) and by subcontract (B335246) from Lawrence Livermore National Laboratory to ASU. Work done (partially) at SSRL which is operated by the DOE, Office of Basic Energy Sciences. The SSRL Biotechnology Program is supported by the National Institutes of Health, National Center for Research Resources, Biomedical Technology Program, and by the DOE, Office of Biological and Environmental Research. STRONTIUM(II) COORDINATION, I REFERENCES 1. Maginini, M., Licheri, G., Paschina, G., Piccaluga, G., and Pinna, G., “X-Ray Diffraction of Ions in Aqueous Solutions: Hydration and Complex Formation.” CRC Press, Boca Raton, FL 1988. 2. Persson, I., Sandstrom, M., Yokoyama, H., and Chaudry, M., Z. Naturforsch. A 50, 21 (1995). 3. Kohn, S. C., Charnock, J. M., Henderson, C. M. B., and Greaves, G. N., Contrib. Mineral. Petrol. 105, 359 (1990). 4. Creux, S., Bouchet-Fabre, B., and Gaskell, P. H., J. Non-Cryst. Solids 192– 193, 360 (1995). 5. Pingitore, N. E., Jr., Lytle, F. W., Davies, B. M., Eastman, M. P., Eller, P. G., and Larson, E. M., Geochim. Cosmochim. Acta 56, 1531 (1992). 6. Parise, J. B., Corbin, D. R., and Subramanian, M. A., Mater. Res. Bull. 24, 303 (1989). 7. D’Angelo, P., Nolting, H.-F., and Pavel, N. V., Phys. Rev. A 53, 798 (1996). 8. Parkman, R. H., Charnock, J. M., Livens, F. R., and Vaughan, D. J., Geochim. Cosmochim. Acta. 62, 1481 (1998). 9. Axe, L., Bunker, G. B., Anderson, P. R., and Tyson, T. A., J. Colloid. Interface Sci. 199, 44 (1998). 10. Chen, C.-C., Papelis, C., and Hayes, K. F., in “Adsorption of Metals by Geomedia” (E. A. Jenne, Ed.), p. 333. Academic Press, New York, 1998. 11. D’Angelo, P., Di Nola, A., Filipponi, A., Pavel, N. V., and Roccatano, D., J. Chem. Phys. 100, 985 (1994). 12. Roccatano, D., Berendsen, H. J. C., and D’Angelo, P., J. Chem. Phys. 108, 9487 (1998). 13. Sahai, N., Carroll, S. A., Roberts, S., and O’Day, P. A., J. Colloid. Interface Sci., 223, 198 (2000). 14. Schwertmann, U., and Cornell, R. M., “Iron Oxides in the Laboratory: Preparation and Characterization.” VCH, New York, 1991. 15. George, G. N., and Pickering, I. J., “EXAFSPAK: A Suite of Computer Programs for Analysis of X-Ray Absorption Spectra.” Stanford Synchrotron Radiation Laboratory, 1993. 16. Stern, E. A., Newville, M., Ravel, B., Yacoby, Y., and Haskel, D., Physica B 209, 117 (1995). 17. Brown G. E., Jr., Calas, G., Waychunas, G. A., and Petiau, J., in “Spectroscopic Methods in Mineralogy and Geology” (F. C. Hawthorne, Ed.), Vol. 18, p. 431. Mineralogical Society of America, Washington, DC, 1988. 18. Rehr, J. J., Albers, R. C., and Zabinsky, S. I., Phys. Rev. Lett. 69, 3397 (1992). 19. Rehr, J. J., Jpn. J. Appl. Phys., Part 1 32-2, 8 (1993). 20. Zabinsky, S. I., Rehr, J. J., Ankudinov, A., Albers. R. C., and Eller, M. J., Phys. Rev. B 52, 2995 (1995). 21. Ravel, B., Atoms Homepage: http://leonardo.phys.washington.edu/∼ravel/ atoms/, 1999. 197 22. O’Day, P. A., Rehr, J. J., Zabinsky, S. I., and Brown, G. E., Jr., J. Am. Chem. Soc. 116, 2938 (1994). 23. Bunker, G., Nucl. Instrum. Methods 207, 437 (1983). 24. Crozier, E. D., Rehr, J. J., and Ingalls, R., in “X-ray Absorption: Principles, Applications, Techniques of EXAFS, SEXAFS and XANES” (D. C. Koningsberger and R. Prins, Eds.), p. 373. Wiley, New York, 1988. 25. Primak, W., Kaufman, H., and Ward, R., J. Am. Chem. Soc. 70, 2043 (1948). 26. Pannhorst, W., and Löhn, J., Z. Krist. 131, 455 (1970). 27. DeVilliers, J. P. R., Am. Mineral. 56, 758 (1971). 28. Scheel, H. J., Acta Crystallogr. A 37, 916 (1981). 29. Hutton, J., and Nelmes, R. J., J. Phys. C Solid State Phys. 14, 1713 (1981). 30. Boudjada, A., Masse, R., and Guitel, J. C., Acta Crystallogr. B 34, 2692 (1978). 31. Thompson, H. A., Parks, G. A., and Brown, G. E., Jr., Geochim. Cosmochim. Acta. 63, 1767 (1999). 32. Lytle, F. W., in “Applications of Synchrotron Radiation” (H. Winik, Ed.), Gordon & Breach Science, New York, 1989. 33. Lucchetti, G., Massa, B., and Penco, A. M., Neues Jb. Miner. Mh. 1982, 541 (1982). 34. Gottardi, G., and Galli, E., “Natural Zeolites.” Springer-Verlag, Berlin, 1985. 35. Merkle, A. B., and Slaughter, M., Am. Mineral. 53, 1120 (1968). 36. Gunter, M. E., Armbruster, T., Kohler, T., and Knowles, C. R., Am. Mineral. 79, 675 (1994). 37. Fossheim, K., and Berre, B., Phys. Rev. B 5, 3292 (1972). 38. Fischer, M., Bonello, B., Itie, J. P., Polian, A., Dartyge, E., Fontaine, A., and Tolentino, H., Phys. Rev. B 42, 8494 (1990). 39. O’Day, P. A., Brown, G. E., Jr., and Parks, G. A., J. Colloid Interface Sci. 165, 269 (1994). 40. Armbruster, T., Am. Miner. 78, 260 (1993). 41. Armbruster, T., and Gunter, M. E., Am. Mineral. 76, 1872 (1991). 42. Artioli, G., Smith, J. V., and Kvick, A., Acta Crystallogr. C 41, 492 (1985). 43. Ames, L. L., Jr., Am. Mineral. 45, 689 (1960). 44. Bish, D. L., Vaniman, D. T., Rundberg, R. S., Wolfsberg, K., Daniels, W. R., and Broxton, D. E., in “Radioactive Waste Management” Vol. 3, p. 415. International Atomic Energy Authority, Vienna, 1984. 45. Pabalan, R. T., and Bertetti, F. P., in “Scientific Basis for Nuclear Waste Management XVII” (A. Barkatt and R. A. Van Konynenburg, Eds.), p. 731. Materials Research Society, Pittsburgh, 1994. 46. Pfund, D. M., Darab, J. G., and Fulton, J. L., J. Phys. Chem. 98, 13102 (1994). 47. Caminiti, R., Musinu, A., Paschina, G., and Pinna, G., J. Appl. Crystallogr. 15, 482 (1982).
© Copyright 2026 Paperzz