Uranium isotope fractionation during coprecipitation ...

3 downloads 0 Views 2MB Size Report
Equilibrium isotope fractionation among different aqueous U species is the most likely explanation for these findings. Certain charged U .... Morford and Emerson, 1999). Under oxic ...... Le Heron D. P., Craig J. and Etienne J. L. (2009) Ancient.
Available online at www.sciencedirect.com

ScienceDirect Geochimica et Cosmochimica Acta 188 (2016) 189–207 www.elsevier.com/locate/gca

Uranium isotope fractionation during coprecipitation with aragonite and calcite Xinming Chen a,⇑, Stephen J. Romaniello a, Achim D. Herrmann b, Laura E. Wasylenki c, Ariel D. Anbar a,d b

a School of Earth and Space Exploration, Arizona State University, Tempe, AZ 85287, USA Coastal Studies Institute and Department of Geology and Geophysics, Louisiana State University, Baton Rouge, LA 70803, USA c Department of Geological Sciences, Indiana University, Bloomington, IN 47405, USA d School of Molecular Sciences, Arizona State University, Tempe, AZ 85287, USA

Received 5 December 2015; accepted in revised form 11 May 2016; Available online 21 May 2016

Abstract Natural variations in 238U/235U of marine calcium carbonates might provide a useful way of constraining redox conditions of ancient environments. In order to evaluate the reliability of this proxy, we conducted aragonite and calcite coprecipitation experiments at pH 7.5 and 8.5 to study possible U isotope fractionation during incorporation into these minerals. Small but significant U isotope fractionation was observed in aragonite experiments at pH 8.5, with heavier U isotopes preferentially enriched in the solid phase. 238U/235U of dissolved U in these experiments can be fit by Rayleigh fractionation curves with fractionation factors of 1.00007 + 0.00002/0.00003, 1.00005 ± 0.00001, and 1.00003 ± 0.00001. In contrast, no resolvable U isotope fractionation was observed in an aragonite experiment at pH 7.5 or in calcite experiments at either pH. Equilibrium isotope fractionation among different aqueous U species is the most likely explanation for these findings. Certain charged U species are preferentially incorporated into calcium carbonate relative to the uncharged U species Ca2UO2(CO3)3(aq), which we hypothesize has a lighter equilibrium U isotope composition than most of the charged species. According to this hypothesis, the magnitude of U isotope fractionation should scale with the fraction of dissolved U that is present as Ca2UO2(CO3)3(aq). This expectation is confirmed by equilibrium speciation modeling of our experiments. Theoretical calculation of the U isotope fractionation factors between different U species could further test this hypothesis and our proposed fractionation mechanism. These findings suggest that U isotope variations in ancient carbonates could be controlled by changes in the aqueous speciation of seawater U, particularly changes in seawater pH, PCO2 , Ca2+, or Mg2+ concentrations. In general, these effects are likely to be small (105 yr; Dunk et al., 2002) suggests that, unlike Ce and I, U might provide globally integrated paleoredox information. U concentrations and the U/Th ratios in sedimentary rocks have been widely used as paleoredox proxies (e.g.,

Anderson et al., 1989; Wignall and Twitchett, 1996; Morford and Emerson, 1999). Under oxic conditions, U exists as soluble U(VI) in the form of uranyl carbonate complexes (Langmuir, 1978). In anoxic marine settings, U (VI) is reduced to insoluble U(IV) and subsequently adsorbed or precipitated as UO2, U3O7 or U3O8 in the anoxic sediments (Klinkhammer and Palmer, 1991; McManus et al., 2005). Thus, variations of U concentrations in anoxic facies like black shales can reflect changes in redox conditions (Algeo and Maynard, 2004; Tribovillard et al., 2006; Partin et al., 2013). Since Th is only present as relatively insoluble Th(IV) in seawater, there is no significant shift in authigenic uptake of Th under changing redox conditions. Hence, the variation of U/Th ratio in the sediments is also used as a redox proxy. Uranium isotopes are being actively explored and applied as an additional source of ocean paleoredox information (Stirling et al., 2007; Weyer et al., 2008; MontoyaPino et al., 2010; Brennecka et al., 2011a; Asael et al., 2013; Kendall et al., 2013, 2015; Andersen et al., 2014; Dahl et al., 2014; Goto et al., 2014; Noordmann et al., 2015; Tissot and Dauphas, 2015; Lau et al., 2016). Although natural variations in 238U/235U can be caused by processes such as leaching and adsorption (Stirling et al., 2007; Brennecka et al., 2011b; Heiss et al., 2012), redox variations are the primary drivers of U isotope fractionation on Earth, with heavier U isotopes 238U preferentially enriched in reduced species (Stirling et al., 2007, 2015; Weyer et al., 2008; Basu et al., 2014; Bopp et al., 2010; Stylo et al., 2015). According to these studies, widespread anoxia in the oceans causes more U to be scavenged by anoxic sediments, preferentially sequestering 238U from seawater and leading to a decrease in the 238U/235U ratio of seawater. Because of the long residence time of U, it has been proposed that U isotope variations can be used to infer changes in global redox conditions (Weyer et al., 2008; Tissot and Dauphas, 2015). For example, variations of 238U/235U and U concentration in black shales were used to quantify the spatial extent of marine anoxia during Oceanic Anoxic Event 2 (Montoya-Pino et al., 2010) and indicate oxidative U mobilization at 2.50 Ga (Kendall et al., 2013). Because of the wide spatial and temporal distribution of carbonate sedimentary rocks, variation of 238U/235U in carbonate rocks is being explored as a paleoredox proxy. 238 U/235U in modern primary carbonate precipitates (corals, green algae, ooids, etc) is close to that of seawater (Stirling et al., 2007; Weyer et al., 2008; Romaniello et al., 2013; Tissot and Dauphas, 2015), suggesting that marine carbonates capture the 238U/235U of coexisting seawater. Based on this finding, variations of 238U/235U and U/Th in a carbonate section at Dawen in southern China were interpreted as reflecting an enhancement of oceanic anoxia immediately before the end-Permian mass extinction (Brennecka et al., 2011a; Lau et al., 2016). So far, no resolvable U isotope fractionation has been observed between modern aragonite and calcite samples and seawater (Weyer et al., 2008; Romaniello et al., 2013). However, many factors have varied in the past, including seawater pH and CO2 ion concentration 3 (Berner and Kothavala, 2001; Royer et al., 2004;

X. Chen et al. / Geochimica et Cosmochimica Acta 188 (2016) 189–207

Ridgwell, 2005), the relative abundance of cations (Ca2+, Mg2+, and Sr2+, Hortia et al., 2002; Lowenstein et al., 2003), and the primary polymorph of CaCO3 (i.e., aragonite vs. calcite, Ries, 2004, 2010). This might have caused U isotope fractionation during U incorporation into carbonate rocks in the geological past. In particular, variations in aqueous U speciation might lead to isotopic fractionation of 238U/235U ratios during coprecipitation with carbonates if different aqueous U species are isotopically fractionated from one another and certain U species are preferentially incorporated into the solid phase. It is also possible that 238U/235U could be fractionated during incorporation into different polymorphs of CaCO3 due to coordination changes required to accommodate U in the mineral structures. Previous experimental studies found that the predominant U(VI) species in seawater, UO2(CO3)4 3 , was directly incorporated into aragonite without coordination change (Reeder et al., 2000). In contrast, the coordination of UO2(CO3)4 3 was altered during incorporation into calcite, adopting an equatorial coordination number of 5 (Reeder et al., 2000, 2001). Similarly, studies of natural calcite samples have suggested that uranyl substitutes for Ca with an equatorial coordination number of 4 (Kelly et al., 2003, 2006). This difference in the local bonding environment could lead to isotopic fractionation. For example, coordination changes are known to cause resolvable isotope fractionation during adsorption of metal complexes (i.e., U, Mo, Tl, Zn and Cu) onto synthetic Kbirnessite (Wasylenki et al., 2011; Brennecka et al., 2011b; Little et al., 2014; Bryan et al., 2015). In this study, we conducted U(VI) coprecipitation experiments with aragonite and calcite to investigate the effect of varied U speciation (induced by pH) on U isotope fractionation during U incorporation into these two CaCO3 polymorphs. The goal of these experiments was to test whether calcite and aragonite record seawater 238U/235U under a variety of conditions. Experiments were conducted at two different pHs (7.5 and 8.5) that capture most of the range of pH expected in different marine environments over the Phanerozoic. Although the dissolved U concentrations in our experiments were much higher than that in seawater, we do not expect U speciation and U isotope fractionation to depend on U concentration, as discussed below. Additionally, we find that the growth rates of CaCO3 in our experiments fell within those of natural CaCO3 (Owen et al., 2002; Stoll et al., 2002; Wingrad et al., 2006), suggesting that our experimental work is relevant to natural environments. 2. MATERIALS AND METHODS As detailed below, U(VI) coprecipitation experiments with aragonite and calcite were conducted at pH 7.5 and 8.5. Aqueous and solid samples were collected during each experiment to measure U concentration and 238 U/235U. After completion of each experiment, approximately 1 g of the bulk precipitate was collected for X-ray diffraction (XRD) analysis to verify the CaCO3 polymorph. The U(VI) speciation in the aqueous solution was calculated using the software PHREEQC (Parkhurst and

191

Appelo, 2004) to explore the relationship between pH, U speciation, and isotope fractionation. 2.1. Materials All labware was washed using reagent-grade nitric acid (20%), hydrochloric acid (20%), and 18.2 MX-cm deionized water, following standard trace metal cleanlab protocols (Howard and Statham, 1993). Nitric acid and hydrochloric acid used to dissolve samples and purify U for isotope analyses were Omega trace metal grade (Fisher Scientific, Lot #4114020 (HCl) and #1114010 (HNO3)). Chemical reagents (NaCl, anhydrous MgCl2, Na2CO3, NaHCO3, CaCl22H2O, 30% H2O2) were ACS grade (Fisher Scientific). ICP-MS standard (PU1KN, Ricca Chemical Company LLC, Lot #: 4101230) was used as the uranyl nitrate solution in the U coprecipitation experiments. 2.2. Experimental setup U coprecipitation experiments were performed using the constant addition method (Zhong and Mucci, 1996; Reeder et al., 2000) with a 1 L Erlenmeyer flask as the reactor (Fig. 1). Prior to each experiment, a background NaCl electrolyte solution with U(VI) (50 lM) was added into the reactor. CaCl2 and Na2CO3 or NaHCO3 (concentrations were given below) were delivered at the same flow rate into the reactor via a dual-channel syringe pump (Model: NE1600, New Era Pump Systems, Inc.) to precipitate CaCO3. The solution in the reactor was well mixed with a Teflon stir bar and bubbled with air to maintain constant PCO2 in the aqueous solution. Experiments were run at ambient lab temperature which is tightly controlled at 24 ± 0.5 °C. However, we found that it was necessary to isolate the reactor from the magnet stir-plate using a thermally insulating pad in order to keep the heat of the stirrer motor from warming the reactor by several degrees centigrade (3 ° C). The pH meter was connected to a computer to record the pH in the aqueous solution every 5 min. 2.3. U coprecipitation with aragonite and calcite procedures Four aragonite experiments (A1, A2, A3, and A*1) and three calcite experiments (C1, C2, and C3) were conducted. Aragonite experiment A*1 was a replicate of experiment A1. Experiments A1, A3, A*1, C1, and C3 were conducted at pH 8.5 ± 0.1, while experiments A2 and C2 were performed at pH 7.5 ± 0.1. The pH was set by adjusting the delivery rates of chemical reagents into the reactor. The experimental conditions are summarized in Table 1. All aragonite experiments were conducted using 0.5 M CaCl2 and 1.0 M NaHCO3. In order to maintain a stable pH over several days while precipitating several grams of CaCO3, it is critical that Ca and carbonate alkalinity are delivered in exact stoichiometric proportions. In order to achieve this, large batches of reagents were prepared gravimetrically and then iteratively refined in a series of preliminary experiments to ensure that a stable pH could be maintained at the desired flow rate. Between preliminary experiments, small adjustments to the reagent concentrations

192

X. Chen et al. / Geochimica et Cosmochimica Acta 188 (2016) 189–207

Fig. 1. Illustration of experimental setup for the constant addition method.

were made by slightly diluting the excess reagent in order to achieve stoichiometric precisions much better than 0.1%. These solutions were delivered into the reactor simultaneously via two Teflon tubes (0.51 mm diameter) using a dual-channel syringe pump equipped with 60 mL syringes. NaCl (0.5 M) was used as the background electrolyte. In aragonite experiments, a Mg2+/Ca2+ratio >2 was maintained, which inhibits calcite growth by destabilizing the crystal lattice and increasing its solubility (ChoudensSa´nchez and Gonza´lez, 2009; Morse et al., 1997). To maintain a high and constant Mg2+/Ca2+ ratio, MgCl2 was added to the NaCl and CaCl2 solutions. The Mg2+ concentration was 50 mM in experiments A1, A2, and A*1 and 100 mM in experiment A3. Prior to coprecipitation, uranyl nitrate was added into the NaCl solution to achieve an initial U concentration of 50 lM. The high U concentration (about 3600 times that of seawater, 14 nM) ensures that there is enough dissolved U for isotope analysis of the aqueous solution even when >95% of U has been incorporated into the precipitate. Aqueous samples (2 mL) were collected every 12 h and immediately filtered through a 0.45 lm filter. Subsequently, the filtered samples were diluted with concentrated nitric acid to a final concentration of 3 M HNO3. At the termination of each experiment, a portion of the bulk CaCO3 precipitate was dissolved with concentrated HNO3 for U concentration and isotope ratio measurements. The experimental conditions and procedures for calcite experiments C1 and C2 were the same as for aragonite experiments A1 and A2, except that no MgCl2 was used. The experimental procedure for calcite experiment C3 was slightly different from the other calcite experiments. It was modified to precipitate a greater proportion of dissolved U in order to achieve a more precise measurement of the isotope fractionation. To achieve this goal, higher concentrations of CaCl2 (2.0 M) and Na2CO3 (2.0 M) were used. To keep ionic strength constant, 2.0 M NaCl was used as the background electrolyte. One gram of calcite seed

crystal synthesized with reagent grade CaCl2 and NaHCO3 was added into the NaCl solution and mixed with a stir bar for 18 h. Subsequently, uranyl nitrate was introduced into the NaCl solution to attain a U concentration of 50 lM. During the calcite coprecipitation process, aliquots of aqueous solution (2 ml) were collected every 12 h. To compare the U isotope ratios of aqueous solution and instantaneous solid, each day a clean glass microscope slide was submerged into the solution to collect daily precipitate and retrieved after 24 h. When experiment C3 ended, bulk solid was also taken for XRD and isotopic analysis. The preparation procedures for U concentration and its isotopic ratio analysis in aqueous and solid samples were the same as in all other experiments. 2.4. X-ray diffraction (XRD) analysis Approximately 1 g of the bulk precipitate from each experiment was collected, ground using an agate mortar and pestle, and loaded into a sample holder. The crystal structures of synthetic carbonates were characterized by powder XRD (Bruker-AXS D8 Advance, ASU) using Cu Ka radiation (40 kV and 30 mA) and a Ni filter with a scanning speed of 0.005° 2h s1. The time constant was 2 s. The crystal structures of the synthetic calcium carbonates were confirmed by comparing the spectra with those of calcite standard (R040070) and aragonite standard (R040078) from the International Center for Diffraction Data. 2.5. Specific surface area measurement In order to estimate the crystal growth rate, the specific surface area of calcite in experiments C1 and C2 was measured using the Brunauer–Emmett–Teller (BET) method (Brunauer et al., 1938; Gregg and Sing, 1982). About 1 g of the calcite precipitate was dried at 150 °C for 24 h to remove moisture, and the specific surface area was measured at 77 K using N2 as adsorption gas.

X. Chen et al. / Geochimica et Cosmochimica Acta 188 (2016) 189–207

Note: Mg2+ and Ca2+ were concentrations of Mg and Ca after pH was stable. Flow rate in the table is the flow rate of delivering CaCl2 and NaHCO3/Na2CO3 solutions into the reactor.

2145 2525 2075 2053 336 360 250 94.9 98.6 98.2 93.7 14.5 17.5 90.5 8.5 ± 0.1 7.5 ± 0.1 8.5 ± 0.1 8.5 ± 0.1 8.5 ± 0.1 7.5 ± 0.1 8.5 ± 0.1 A1 A2 A3 A*1 C1 C2 C3

0.50 0.50 0.50 0.50 0.50 0.50 2.00

CaCl2 + NaHCO3 + MgCl2 CaCl2 + NaHCO3 + MgCl2 CaCl2 + NaHCO3 + MgCl2 CaCl2 + NaHCO3 + MgCl2 CaCl2 + NaHCO3 CaCl2 + NaHCO3 CaCl2 + Na2CO3

4 40 4 4 4 40 10

160 16 160 160 160 16 168

50 50 100 50 0 0 0

5 5 4 4.8 2 2 3.5

1.94 1.71 1.96 1.95 1.91 1.64 19.44

U in carbonate (lg U/g CaCO3) pH

Electrolyte NaCl (M)

Reagents

Flow rate (ll/min)

Time (h)

Mg2+ (mM)

Ca2+ (mM)

CaCO3 (g)

U drawdown (%)

2.6. U concentration analysis

Experiment

Table 1 Summary and results of CaCO3 coprecipitation experiments.

193

Aqueous samples collected during coprecipitation experiments were diluted to 0.32 M nitric acid for U concentration analysis using quadrupole inductively-coupled plasma mass spectrometry (ICP-MS, Thermo X-series) at Arizona State University (ASU). Solid samples (typically 0.1 g) from each experiment were dissolved with 1 ml of 3 M HNO3 and diluted similarly. The estimated precision for the U concentration measurements by ICP-MS was ±2% based on repeated analysis of check standard solutions during each run. 2.7. U isotope analysis The 233U–236U double-spike method was used to measure U isotope ratios, using multiple collector ICP-MS (MC-ICP-MS) (Weyer et al., 2008). The double-spike method is the most effective way to correct for instrumental mass bias and potential isotope fractionation during U purification (Dodson, 1963; Rudge et al., 2009). About 500 ng of U from each sample was well mixed with sufficient double-spike U solution (IRMM-3636) to achieve a spike:sample molar ratio of 0.0363 (Verbruggen et al., 2008; Weyer et al., 2008). The samples were spiked prior to chromatographic separation to correct for any U isotope fractionation that might occur during U purification. U purification was carried out following the protocol of Weyer et al. (2008). Polypropylene columns (BioRad #7311550) were first rinsed with 10 ml 18.2 MX-cm deionized water. A volume of 0.8 ml UTEVA resin (Eichrom Technologies, LLC) was loaded on a column. The resin was then washed with 4  2.5 ml 0.05 M HCl to remove impurities. The resin was then converted to the nitric form by loading 3  0.8 ml 3 M HNO3. The double-spiked U sample (dissolved in 3 M HNO3) was loaded on the column and rinsed with 5  2 ml 3 M HNO3 to remove all matrix ions except U and Th. Then 10 M HCl (3  0.8 ml) was added to the column to convert the UTEVA resin to chloride form. Th was removed from the resin using a mixture of 5 M HCl and 0.05 M oxalic acid (3  0.8 ml). The oxalic acid left on the resin was rinsed with 3  0.8 ml 5 M HCl. The U adsorbed on the resin was eluted with 7 ml (1 + 1 + 1 + 2 + 2) 0.05 M HCl. The U cuts were dried down and heated with concentrated HNO3 + 30% H2O2 (1 ml + 0.3 ml) to remove any organic residue eluted from the UTEVA resin. The last step was repeated until all organic residue was removed. The U isotope ratios were measured at ASU using MCICP-MS (Thermo Scientific Neptune). Purified samples were dissolved in 0.32 M HNO3 with a U concentration of 50 ppb and introduced into the instrument via an Apex-Q desolvation introduction system. Ion beams of 233 U, 235U, 236U, and 238U were collected with Faraday cups connected to 1012 X, 1012 X, 1012 X, and 1011 X resistors respectively. The typical voltage for 238U from a 50 ppb U solution was 30 V. U isotopic compositions are reported in d notation relative to the standard CRM145 using the following equation:

194

X. Chen et al. / Geochimica et Cosmochimica Acta 188 (2016) 189–207

" 238

d



ð238 U=235 UÞsample

ð238 U=235 UÞCRM-145

#  1  1000

where (238U/235U)sample and (238U/235U)CRM-145 are the U isotope ratio of a sample and the standard, respectively. After chemistry, each sample was measured at least three times. The U isotope precision is presented as twice the standard error (2 SE), calculated using the standard deviation of either the sample or standards, whichever was larger. The CRM-145 standard was run every third analysis to correct for small amounts of instrument drift. The reproducibility of replicate measurements of the U isotope composition of CRM-145 is within ±0.08‰ (2 SD, n = 24). The accuracy of the U isotope measurement was assessed by measuring the standard CRM-129a. The measured average d238U of CRM-129a is 1.70 ± 0.10‰ (2 SD, n = 20, Table S8 in Supplementary Information), which agrees with previously published values (Weyer et al., 2008; Shiel et al., 2013; Wang et al., 2015). The blank from the U purification procedure was 0.0 ± 0.04 ng (2 SD, n = 6, Table S8 in Supplementary Information), which was negligible relative to the 500 ng sample U. 2.8. U speciation calculations The speciation of U(VI) over the pH range 7–9 was modeled with the software package PHREEQC (Parkhurst and Appelo, 2004), using database sit.dat with updated U thermodynamic data from Grenthe et al. (1992) and Guillaumont et al. (2003). Three new aqueous uranyl complexes Ca2UO2(CO3)3(aq), CaUO2(CO3)2 3 , and MgUO2(CO3)2 were also added to the database 3 (Dong and Brooks, 2006). Specific ion interaction theory (SIT) was used for aqueous U speciation calculations because it is effective for estimating single-ion activity coefficients for solutions of up to 3 M ionic strength in which NaCl is the dominant electrolyte (Bethke, 2008).

Fig. 2. Variations of pH in calcium carbonate coprecipitation experiments at pH 8.5 ± 0.1 (A1 and C1) and 7.5 ± 0.1 (A2 and C2). The variations of pH in other experiments A*1, A3 and C3 (pH 8.5) were similar to those of A1 and C1.

in experiment C1 and C2. In experiment C3, which was modified to increase the amount of coprecipitation, approximately 90% of the total U was incorporated. 3.2. XRD analysis The XRD spectra of the synthetic carbonates (C1, C2, C3, A1 and A2) were compared to those of calcite and aragonite standards. Spectra of synthetic carbonates matched perfectly with their corresponding standards (Fig. 3). Although no XRD analysis was done for precipitates in experiments A3 and A*1, the higher Mg2+/Ca2+ ratios in these experiments (>10) relative to that in aragonite experiment A1 (9) indicate that the polymorph of CaCO3 in A3 and A*1 should also be aragonite (Morse et al., 1997; Choudens-Sa´nchez and Gonza´lez, 2009).

3. RESULTS 3.1. U coprecipitation with aragonite and calcite U coprecipitation experiments with aragonite and calcite were conducted under different conditions as shown in Table 1. The pH of these experiments increased from 7.0 to a maximum point (8.5 or 9.2) until the first calcium carbonate nucleated on the wall of the glass reactor (Fig. 2). Subsequently, the pH dropped quickly and stabilized within ±0.1 pH unit when the amount of calcium pumped into the reactor was balanced via precipitation as CaCO3. The measured U concentrations in aragonite and calcite precipitates were about 2300 ppm and 300 ppm, respectively, and were at most only weakly dependent on pumping speed or pH (Table 1). Precise determination of the isotopic fractionation factor requires that most of the U is incorporated into the solid, thereby amplifying the isotopic fractionation in the residual solution. More than 94% of the total U was incorporated into aragonite in experiments A1 and A2, whereas less than 18% of U was incorporated into calcite

Fig. 3. XRD spectra of calcium carbonate. A1 (aragonite 1), A2 (aragonite 2), Astd (aragonite standard), C1 (calcite 1), C2 (calcite 2), C3 (calcite 3), Cstd (calcite standard).

X. Chen et al. / Geochimica et Cosmochimica Acta 188 (2016) 189–207

3.3. U speciation The speciation of U(VI) over the pH range of 7–9 at different experimental conditions are shown in Fig. 4. U speciation varies significantly between pH 7.5 and 8.5 in all calcite and aragonite experiments. U speciation also varies between aragonite and calcite experiments (Fig. 4a vs. c) due to the presence of Mg–U species (e.g., MgUO2 (CO3)2 3 ) in the aragonite experiments. U speciation is dominated by CaUO2(CO3)2 3 (33%) in experiments A1 and A*1 at pH 8.5 (Fig. 4a). However, in experiment A2, conducted at lower pH (7.5 vs. 8.5), (UO2)2(CO3)(OH) 3 is the dominant U species (50%). The Mg2+ concentration in experiment A3 is about twice that in experiments A1 and A*1, and MgUO2(CO3)2 becomes 3 the dominant U species, accounting for 42% of the total U species at pH 8.5. Fig. 4c shows that UO2(CO3)4 (50%) is the predomi3 nant U species in experiment C1 at pH 8.5 and that (UO2)2(CO3)(OH) 3 (55%) becomes another important

195

U species in experiment C2 at lower pH (7.5). In experiment C3 (Fig. 4d), which is characterized by higher ionic strength, UO2(CO3)4 (75%) is the dominant U species. 3 Although the U concentrations in the solution decreased during the course of these calcium carbonate coprecipitation experiments, thermodynamics calculations using PHREEQC shows that U speciation was constant throughout each experiment. 3.4. CaCO3 crystal growth rates To compare the crystal growth rates of calcium carbonates in our experiments to those in nature, the CaCO3 crystal growth rates were estimated using following two methods. One is estimation of crystal growth rate from the surface area measured by BET. The specific surface areas for calcite C1 and C2 were 3500 cm2/g and 3184 cm2/g, respectively. Given the total mass of the calcite precipitated in these experiments, the respective total surface area of C1 and

Fig. 4. Aqueous U(VI) speciation as a function of pH (7–9) for aragonite and calcite experiments at PCO2 = 103.5 atm. The vertical dashed lines are pHs of different aragonite and calcite experiments. (a) U(VI) speciation in aragonite experiments A1, A2 and A*1 with ionic strength I = 0.65 M, total Ca2+ concentration 5 mM, Mg2+ concentration 50 mM. U speciation in A*1 is slightly different from that in A1 due to the lower concentration of Ca2+ (4.8 mM); (b) U(VI) speciation in aragonite experiment A3 with ionic strength I = 0.8 M, total Ca2+ concentration 4 mM, Mg2+ concentration 100 mM; (c) U(VI) calcite experiments C1 and C2 with ionic strength I = 0.5 M, total Ca2+ concentration 2 mM; (d) U(VI) speciation in calcite experiment C3 with ionic strength I = 2.0 M, total Ca2+ concentration 3.5 mM.

196

X. Chen et al. / Geochimica et Cosmochimica Acta 188 (2016) 189–207

C2 were 6685 cm2 and 5221 cm2. The carbonate precipitation rate (in g CaCO3/s) divided by the total surface area (cm2) and the density of calcite (g/cm3) yields estimated crystal growth rates of 0.002 nm/s and 0.020 nm/s for C1 and C2. Alternatively, the calcite crystal growth rate can be estimated using the following empirical equation (Wolthers et al., 2012): R ¼ I 0:004 pH10:71 r0:35 ðS  1Þ2 aq

ð1Þ

where R is the overall growth rate (m/s), raq is the ionic activity ratio of Ca2+ to CO2 3 , and S is the saturation ratio. Based on this empirical equation, the crystal growth rates for calcite experiments C1 and C2 were 0.008 nm/s and 0.024 nm/s, respectively. These values are very close to those estimated in the first method. Because the ionic strength in calcite experiment C3 (2.0 M) was far beyond the range of the appropriate ionic strength in the empirical equation (0–0.7 M), a calcite growth rate of 0.02–0.08 nm/s was estimated for C3, which precipitated 10 times as much CaCO3 as in C1 during the same time period. Although we do not have direct estimates of the crystal growth rates for the aragonite experiments, the overall durations and masses

of CaCO3 precipitated in the calcite and aragonite experiments were very similar. Thus, while acknowledging that the specific surface area of calcite and aragonite may be dissimilar, we assume that the CaCO3 crystal growth rate of experiments A1 and A2 were within an order of magnitude of those in C1 and C2, respectively. Thus, the CaCO3 crystal growth rates in our experiments ranged from 0.0002 to 0.24 nm/s, which is well within the range of natural CaCO3 crystal growth rates, 0.0000033–3.30 nm/s (Ter Kuile and Erez, 1984; Genty et al., 2001; Owen et al., 2002; Stoll et al., 2002; Wingrad et al., 2006). 3.5. U isotope fractionation in aragonite and calcite experiments 3.5.1. Aragonite experiments U isotopic compositions in the aqueous and bulk solid samples in aragonite coprecipitation experiments are displayed in Fig. 5. Based on repeated analysis, d238U of the starting stock (ICP standard PU1KN, Ricca Chemical) is 0.23 ± 0.10‰ (2 SD, n = 12). Thus, U isotope compositions at the beginning of aragonite and calcite coprecipitation experiments were 0.23 ± 0.10‰. To verify mass

Fig. 5. U isotopic compositions of aqueous and solid samples versus fraction of U remaining in the aqueous solution for aragonite experiments A1, A*1, A2 and A3. The open triangles and filled rectangles represent the U isotopic compositions of aqueous samples and bulk solid samples, respectively. The black curves are Rayleigh fractionation curves in all aragonite experiments. The gray areas in (a)–(d) are confidence envelopes of the isotope fractionation factor a (95% confidence interval).

X. Chen et al. / Geochimica et Cosmochimica Acta 188 (2016) 189–207

balance, we compared d238U in the bulk aragonite at the end of the experiments, when >94% of the U in the solution was incorporated into the aragonite, to that of the initial solution. These values are identical to the starting stock within the uncertainties (A1: 0.22 ± 0.06‰, A*1: 0.23 ± 0.06‰, A2: 0.30 ± 06‰, and A3: 0.29 ± 06‰), consistent with expectations. To estimate the magnitude of the U isotope fractionation factor in experiments, the data in each experiment were fitted using the Rayleigh fractionation model, which assumes that coprecipitated U was continuously incorporated into the solid phase and isolated from further isotope exchange with the aqueous solution. The instantaneous isotope fractionation factor, a, is defined as: a¼

197

Fig. 6. The U isotope data in calcite coprecipitation experiments were also fitted using the Rayleigh fractionation model (Table 2). The Rayleigh fractionation factors in calcite experiments C1 (0.99979 + 0.00021/0.00027) and C2 (1.00043 + 0.00011/0.00050) have large uncertainties due to the limited extent of U incorporation (8.2 was incorporated wholesale into aragonite without a change in U coordination (Reeder et al., 2000). In contrast, coordination number and bond length both decrease during UO2(CO3)4 3 incorporation into calcite (Reeder et al., 2000, 2001; Kelly et al., 2003, 2006). Thus, if coordination change upon incorporation into the mineral causes resolvable isotope fractionation, it is reasonable to expect that heavier U isotopes will be preferentially incorporated into calcite, and that no U isotope fractionation will be observed in aragonite. If the U isotope fractionation driven by coordination change is too small to be resolved, no U isotope fractionation should be observed in either the aragonite or the calcite experiments. Since our data do not conform to either expectation, our results are not consistent with coordination change during U incorporation as the driver of the observed fractionation. 4.1.2. Differences in aqueous U speciation prior to incorporation Differences in aqueous U speciation prior to incorporation might be another mechanism driving the U isotope fractionation. U speciation changes significantly with pH (Dong and Brooks, 2006). Our speciation calculations indicate that the distribution of U species was different in our experiments at pH 7.5 and 8.5. In aragonite experiment A1, A*1, and A3, at pH 8.5, four U(VI) species, 2 CaUO2(CO3)2 3 , MgUO2(CO3)3 , Ca2UO2(CO3)3(aq), and 4 UO2(CO3)3 , should be present in significant amounts

X. Chen et al. / Geochimica et Cosmochimica Acta 188 (2016) 189–207

(Fig. 4a). However, in calcite experiments C1 and C3, at pH 8.5, only three U species, UO2(CO3)4 3 , Ca2UO2(CO3)3 (aq) and CaUO2(CO3)2 3 should coexist. Finally, in experiments A2 and C2 at pH 7.5, we expect that (UO2)2(CO3) (OH) 3 became the dominant U species. Isotope fractionation of U between different aqueous species is controlled by the local chemical bonding environments, including the coordination number and bond length (Schauble, 2004). The coordination number of equatorial O atoms and bond lengths between U and axial O and between U and equatorial O for these U species are shown in Table 3. Differences in coordination and bond lengths suggest that there may be significant isotope fractionation between various species. Table 3 organizes U species with similar bonding environments into several groups. For example, Group 1, containing UO2(CO3)4 3 and Ca2UO2(CO3)3(aq), shares a similar 6-fold coordination, with axial ˚ and equatorial bond lengths bond length (U–Oax) of 1.80 A ˚ (U–Oeq) of 2.43 ± 0.01 A. Because of the similarity in chemical bonding environments, these species are expected to have similar U isotopic composition. Group 2, containand MgUO2(CO3)2 ing CaUO2(CO3)2 3 3 , differs in the ˚ . Group 3 [UO2(equatorial bond lengths are 2.39 ± 0.01 A  CO3)2 2 ] and Group 4 [(UO2)2(CO3)(OH)3 ] contain the remaining species, which have variable bonding environments, but are major U species only below pH 8, there are only four significant U species (in Group 1 and Group 2), whose abundances sum to unity:

202

X. Chen et al. / Geochimica et Cosmochimica Acta 188 (2016) 189–207

1 ¼ f neutral þ f UO2 ðCO3 Þ4 þ f CaUO2 ðCO3 Þ2 þ f MgUO2 ðCO3 Þ2 3

3

3

ð18Þ Combine Eqs. 13–18, we can get the fraction of neutral U species as: 0 K1 f neutral ¼ @1 þ  2 K 2 cUO2 ðCO3 Þ4 cCa2þ ½Ca2þ  3

þ

þ

K3 K 2 cCaUO2 ðCO3 Þ2 cCa2þ ½Ca2þ  3   K 4 cMg2þ ½Mg2þ  K 2 cMgUO2 ðCO3 Þ2 ðcCa2þ ½Ca2þ Þ

11 2

A

ð19Þ

3

This equation shows that, at pH >8, constant ionic strength, and fixed [Mg2+], fneural varies only as a function of [Ca2+]. Thus, combining Eqs. (17) and (19), the isotope fractionation factor (a) as a function of the fraction of neutral U species (fneutral) can be described. To predict the relationship between fneutral and the U isotope fractionation factor (a), the following experimental conditions are assumed: (1) [Mg2+] = 50 mM and ionic strength = 0.65 M (A1 and A*1); (2) [Mg2+] = 100 mM, and ionic strength = 0.80 M (A3); (3) [Mg2+] = 0 mM and ionic strength = 2.0 M (C3). The activity coefficients of all chemical species are calculated using the software PHREEQC (Parkhurst and Appelo, 2004). With all these conditions, and using equilibrium constants (K1, K2, K3 and K4) from Grenthe et al. (1992) and Dong and Brooks (2006), the U isotope fractionation factor (a) as a function of fneutral at different [Mg2+] is calculated and displayed in Fig. 8. Fig. 8 predicts that a always increases as the fraction of neutral U species increases at different Mg concentrations. This prediction is consistent with our experimental results at pH 8.5. Larger fractionation was observed in experiments A1 (1.00007 + 0.00002/0.00003) and A*1 (1.00005 ± 0.00001), in which 27% and 24%, respectively, of the dissolved U was present as Ca2UO2(CO3)3(aq), while smaller U isotope fractionation was seen in experiment A3 (1.00003 ± 0.00001), which had a fraction of 9% Ca2UO2 (CO3)3(aq). Calcite experiment C3 showed unresolvable U isotope fractionation with a fraction of 5% Ca2UO2 (CO3)3(aq) in the aqueous solution. All these isotope fractionation factors lie on their corresponding isotope fractionation factor lines (Fig. 8). The agreement between the model and data in Figs. 7 and 8 support the hypothesis that differences in aqueous U speciation prior to incorporation is a likely mechanism to explain our results in these experiments. One limitation of the model discussed above is that it is limited to pH >8, where only the four U species in Group 1 and 2 are present. At lower pH (i.e.,