Structural environments of incompatible elements in silicate glass/melt systems: II. U IV, U V, and U VI
The structural environments of trace to minor levels (≈2000 ppm to ≈3.0 wt%) of U in several silicate glasses were examined as a function of oxygen fugacity, melt composition, and melt polymerization using X-ray (XANES and EXAFS) and optical absorption spectroscopies. Glass compositions were diopside (CaMgSi 2O 6: DI), anorthite (CaAlSi 2O 8: AN), albite (NaAlSi 3O 8: AB), sodium trisilicate (Na 2Si 3O 7: TS), a peralkaline composition (Na 3.3AlSi 7O 17: PR, approximately halfway between AB and TS), and a calc-alkaline rhyolite composition (RH). A second set of silicate glasses of the same base compositions containing ≈2000 ppm to ≈3.0 wt% U and ≈0.6 to 2.5 wt% F or Cl was also synthesized. In the glasses synthesized under oxidizing conditions (in air), U VI occurs as uranyl groups with two axial oxygens at ≈ 1.77-1.85 ± 0.02 Å and four to five equatorial oxygens at ≈2.21-2.25 ± 0.03 Å. In glasses synthesized under more reducing conditions ( fO2 ≈ 10 -3-10 -7 atm), U V occurs in moderately distorted 6-coordinated polyhedra [ d(U V-O) ≈ 2.19-2.24 ± 0.03 Å], which may co-exist with smaller numbers of U VI species and/or U VI species. Under the most reducing conditions used ( fO2 ≈ 10 -8-10 -12 atm), U IV occurs in less distorted octahedra [ d(U IV-O) ≈ 2.26-2.29 ± 0.02 Å]. No clear evidence for U-F or U-Cl bonds was found for any of the halogen-containing glasses, suggesting that U-halogen "complexes" are not present. In addition, no U-U (second-neighbor) interactions were detected, indicating that no significant clustering of U atoms is present in any of the glasses studied. Bond strength-bond length calculations and constraints placed on local bonding by Pauling's second rule suggest that U IV and U V in 6-coordinated sites in silicate melts will preferentially bond to nonbridging oxygens (NBO's) rather than bridging oxygens (BO's). The unusually low 6-fold coordination of U IV and U V in relatively depolymerized silicate melts (e.g., peralkaline and halogen-rich melts) results in a high U-O bond strength in the melt that is not observed in crystalline U-bearing minerals. This difference in bond strength is partially responsible for the small crystal-melt partition coefficients of U IV. In addition, the common silicate minerals comprising igneous rocks lack appropriate crystallographic sites which can stably accommodate this large and highly charged cation. These factors help explain the normally incompatible character of U IV during magmatic differentiation. In contrast, the low solubility of U IV and U V in more polymerized silicate melts, such as those produced during the late stages of magmatic differentiation, can be explained by a shortage of NBO's. Increasing amounts of 8-fold coordinated U should favor the incorporation of both U IV and U V in accessory minerals like zircon, thorite, titanite, apatite, uranium oxides, etc., thus its more compatible behavior in the latest stages of magmatic differentiation.