The oceanic composition on Saturn's moon Enceladus is evaluated through calculations of thermochemical equilibria at hydrothermal and freezing settings. Conditions of rock alteration are constrained from assumptions and models for the moon's interior composition and thermal evolution, and from the composition of Enceladus' plume. Results show that an early ocean was an alkaline Na+-Cl−-HCO3− solution. Underlying altered rocks consisted of Mg-phyllosilicates, magnetite, Fe and Ni sulfides, and carbonates. Subsequent freezing of oceanic water caused the deposition of a NaCl hydrate, Na, K and Ca carbonates, and the formation of a salt-free ice shell. If an aqueous phase exists on today's Enceladus, it could consist of eutectic Na-Cl-HCO3− brine that at least locally decouples the ice shell and facilitates tidal heating. A lack of firm detection of Na and Cl at Enceladus is consistent with the accumulation of salts at the ice-rock boundary and implies the plume formation via sublimation in the ice shell.
 According to a widely accepted model, decay of short-lived radionuclides on homogeneously accreted Enceladus caused melting of water ice, accumulation of rocks in a core, and the formation of an ice-covered ocean [Schubert et al., 2007; Castillo-Rogez et al., 2007]. The composition of the ocean should have been affected by low-temperature (T) water-rock interactions (e.g., mineral dissolution, secondary precipitation) during ice melting. In addition, water-soluble components of melted ice (trapped gases, their hydrates, organic species) contributed to the oceanic composition. Despite low temperatures, kinetic models reveal rapid hydration of fine-grained Mg silicates and aqueous oxidation of Fe0 metal leading to H2 production [cf. Zolotov and Mironenko, 2007]. Subsequent cycling of oceanic water thorough the rocky core changed the composition of the ocean and rocks. At some point, high temperature could have caused some dehydration and thermal metamorphism of rocks [Castillo-Rogez et al., 2007; Schubert et al., 2007] and organic compounds, and also limited deep water circulation. Large volume changes caused by hydration/dehydration and redox processes affected the progress, rate, and style of hydrothermal activity. Exhaustion of short-lived radionuclides within a few Myr led to cooling of fluids in the core, some re-hydration of solids, and a substantial (or complete) freezing of the ocean. Subsequent low-T water-rock interactions could have been driven by the radioactive decay of 238U, 235U, 232Th, and 40K [Schubert et al., 2007], and tidal heating. These concepts of Enceladus' evolution are internally consistent, reasonable scenarios. However, other possibilities, such as late formation of Saturn and its satellites (after decay of short-lived radionuclides) or heterogeneous accretion would have other consequences and might lead to no ocean at all, or to an ocean of different composition than modeled. Late accretion of aqueously oxidized planetesimals (similar to CI/CM chondrites) could have delivered easily leachable chlorides and Mg-sulfates [cf. Kargel et al., 2000]. The plume composition may, in part, represent cometary-type volatiles. Notwithstanding these uncertainties, and fully consistent with the prior body of published theoretical work cited above, this letter is devoted to theoretical physical-chemical modeling of water-rock interactions and fluid compositions on Enceladus.
2. Modeling of Water-Rock Interactions
 It is assumed that an oceanic composition on early Enceladus was governed by the composition of upwelling hydrothermal fluids, cooling of the fluids at the ocean-rock interface, low-T ocean-rock reactions, and freezing from above. The model assumes that slow water circulation at low gravity together with rapid dissolution of mineral grains led to local water-rock equilibration. Therefore, fluids that entered the ocean were equilibrated with rocks near the oceanic floor. The composition of aqueous solution and secondary mineralogy was modeled through calculation of thermochemical equilibria in the water-rock-gas type system H-O-C-S-Cl-Si-Al-Mg-Fe-Ca-Na-K-Mn-Cr-Co-Ni-P open with respect to H2. On the basis of assumptions and models of Glein et al. , the fugacity (f) of H2 was set to match the CO2/CH4 mole ratio in the plume [Waite et al., 2006]. Note that assumed oxidation state (fH2) represents oxidized fluids/rocks and implies significant H2 escape from earlier reduced fluids. The assumption that plume gases represent fH2 in the early body's interior [Glein et al., 2007] is crucial for redox-sensitive (Fe-, S-, C-beading) species but has a minor effect on models for rock hydration and pH of fluids. The maximum T of fluids that enter the ocean was chosen as ∼310°C, which corresponds to the water-gas saturation at pressure of ∼102 bars at the rock-water boundary. The nominal calculations were performed for the water to rock mass ratio (W/R) of 1, in accordance with the bulk numbers evaluated from the moon's interior models (0.72–1.18 [Schubert et al., 2007]; 0.78 [Nimmo et al., 2007]). Note that an effect of sequestering of oceanic water into the ice shell on W/R ratio and a limited water circulation in the rocky core would compensate each other. Limited water circulation could have been caused by volume expansion during rock's hydration and by filling of pore spaces through precipitation of minerals and organic compounds. The bulk water-free elemental composition of the Orgueil CI carbonaceous chondrite [Jarosewich, 1990] together with Cl abundance in CI chondrites were selected to exemplify a rock of solar composition that has not been altered in a parent body. Taking into account a limited reactivity of kerogen-like organic polymer observed in chondrites, nominal calculations were performed for 10% of rock's carbon. This amount may in part represent cometary-like C-bearing substances that could have accreted together with water ice.
 Cooling of ocean-entering fluids and related mineral precipitation at oceanic-rock interface were modeled by re-equilibration of high-T fluid composition at 0°C without consideration of rocks. For simplicity, mixing of hydrothermal fluids with oceanic water has not been considered. Equilibrium calculations were performed with the GEOCHEQ code that considers nonideality of aqueous, gas, and solid solutions [e.g., Mironenko and Zolotov, 2005]. Freezing of oceanic water was modeled with the FREZCHEM codes [e.g., Marion et al., 2005] that use Pitzer parameters to calculate activity coefficients of solutes and water activity.
3. Modeled Composition of Ocean-Entering Fluids and Oceanic Water
 Throughout history of early Enceladus' ocean, temperature of ocean-entering fluids could have evolved from ∼0°C to ∼300°C and then decreased below 0°C. The ocean consisted of a mixture of cooled high-T and low-T fluids with different oxidation state and degree of equilibration. Since the input of each fluid type is unknown, a series of T-W/R conditions was considered for ocean-forming solutions. However, a rapid rock alteration during core formation [cf. Zolotov and Mironenko, 2007] and ocean-rock interactions away from hydrothermal systems imply a major contribution from low-T water-rock reactions at elevated W/R ratios.
 Equilibrium calculations performed in a range of W/R ratios (Figure 1), temperatures (Figure 2), and bulk C abundances demonstrate that ocean-forming fluids are alkaline solutions with Na+, Cl− and HCO3− as major species. Less abundant species are K+, CO32−, OH−, HS−, and neutral solutes NaCl, CO2, and CH4. The solutions are depleted in Mg, Ca, Fe, Mn, Ni, Al, P, and sulfate sulfur. Higher-T fluids are characterized by elevated concentrations of Na+, K+, HCO3−, HS−, and formate (HCOO−), as well as dissolved (aq) neutral species: NaCl, CO2, CH4, H2, H2S, and SiO2. In low-T solutions, carbonate ion (CO32−) becomes abundant, while CO2 (aq) content is negligible (Figure 2a). Low calculated abundances of CO2 and CH4 in low-T solutions may reveal a necessity of high-T reactions to account for the plume composition [cf. Glein et al., 2007]. Although N speciation is not considered, N2 should predominate over NH3 in high-T aqueous environments [Glein et al., 2007] and probably formed via oxidation of N in accreted organic matter. The solution pH ranges from 8 to 11 (Figures 1 and 2b). The salinity (2–20 g/kg H2O) and ionic strength (0.04–0.1 molal) are higher at high-T (Figure 2c) and low-W/R conditions, but less than in Earth's seawater (∼35 g kg−1 and 0.7 molal).
 Results show that water-rock equilibration causes pervasive alteration of sub-oceanic rocks (Figure 2d). Major secondary minerals are Mg-rich phyllosilicates (saponite, serpentine, chlorite), calcite, magnetite, pyrrhotite, and Ni sulfide(s). Goethite and pyrite form at lower temperatures. Saponite is Na-rich (Na/Na + Ca + K > 0.7), and serpentine and chlorite are Mg-rich (chrysotile, clinochlore). Less abundant solids are phosphates, chromite, and Mg and Mn carbonates (dolomite, rhodochrosite). Higher abundances of carbon involved in hydrothermal reactions lead to larger amounts of carbonates, which include siderite and magnesite.
 Rock's elements are unevenly partition into solution (Figure 2e). Although Cl, Na, and K are not abundant in original rocks, the predominance of these elements in aqueous solutions is accounted for by the complete aqueous extraction of Cl and a limited incorporation of alkalis in the clay (saponite). Variations in Na/Cl and K/Cl ratios in solution are affected by the Na and K abundances in saponite (decrease with increasing T and W/R). Low solubilities of Mg-phyllosilicates and Fe-oxides (oxyhydroxides) limit concentrations of Mg and Fe in fluids. In turn, the deficiency of the cations does not favor precipitation of corresponding carbonates and is consistent with elevated concentrations of C-bearing solutes.
 Note that a low-T (<150°C–200°C) inhibition of aqueous redox reactions among sulfides and sulfates [Ohmoto and Lasaga, 1982] and between CH4 and oxidized C species [e.g., Seewald et al., 2006] makes the equilibrium model inadequate for corresponding redox pairs. It follows that upwelling fluids may contain elevated amounts of CH4, H2, and sulfates, which correspond to high-T reactions occurred deeper in the core. This notion also applies to all upwelling hydrothermal fluids that cool and mix at the ocean-rock interface. Note that mineralogy of CI and CM carbonaceous chondrites reveal redox disequilibria in low-T aqueous systems. The likely lack of equilibria among H2, CH4, CO2, HS−, SO42− in low-T fluids and oceanic water provided potential sources of metabolic energy on early Enceladus. A preferential separation of low-soluble gases (H2, CH4, N2) could have also contributed to redox disequilibria.
 Calculated re-equilibration of hydrothermal solutions at the ocean-rock interface (0°C, 100 bar) shows that hydrothermal fluids generally preserve their speciation after cooling (Figure 3). As an example, a cooled 100°C fluid is characterized by the following molal (mol/kg H2O) concentrations: Na+, 0.052; K+, 3.6 × 10−3; Ca, 4 × 10−6; Cl−, 0.034; HCO3−, 0.012; CO3−2, 4.3 × 10−3; HS−, 2.2 × 10−4; OH−, 1.1 × 10−5; SO4−2, 3 × 10−6; NaCl, 1.7 × 10−4; NaHSiO3, 7 × 10−5; SiO2, 8 × 10−6; CaCO3, 4 × 10−6; CO2, 5 × 10−6; CH4, ∼2 × 10−6; H2S, 6 × 10−7, has pH 9.9, salinity of 3.3 g/kg H2O, and ionic strength of 0.06 molal. Observed minor precipitation of pyrite, carbonates, and amorphous silica is limited by concentrations of cations (Fe2+, Ca2+, Mg2+) and SiO2(aq), and have only a minor effect on solution composition (compare Figures 2a and 3). The range of concentrations shown in Figure 3 could be considered as a proxy for oceanic composition.
 Separation and upward migration of gases from rock's fluids and the ocean should have also affected aqueous compositions. In our models, CO2-CH4 rich gas forms at T above ∼270°C in C-rich hydrothermal systems (at 100 bar). If only 10% of original C is involved, the gas may separate only at upper (low pressure) parts of the ocean covered by a thin ice shell. In a high-pH ocean, separation of CO2(g) was strongly limited by the conversion of aqueous CO2 to HCO3− and CO32− ions. Gas accumulated below the ice shell could have trapped in solid clathrates (CO2·6H2O, CH4·6H2O [cf. Kieffer et al., 2006]), especially upon a substantial ocean freezing. Although CI-based C content (∼3 wt. %) can potentially account for a clathrate-made shell, the actual amount of the clathrates is limited by restricted reactivity of primary organic polymer, graphite formation in deep interior, substantial trapping of C in solid/aqueous carbonate species and secondary organic compounds, and escape of C-bearing gases.
4. Where Are Aqueous Solutions Today?
 Substantial freezing of oceanic water caused precipitation of hydrohalite (NaCl·2H2O) and smaller amounts of Na, K, and Ca carbonates, as exemplified in Figure 4. Assuming a slow downward freezing on early Enceladus, salts and brine pockets should have been expelled from ice. Before complete freezing, an eutectic Na+-K+-HCO3−-Cl− brine accumulated at the ice-rock (salt) interface. A uniform downward freezing would have led to a stratified global salt layer of 0.4–0.6 km thick. A localized freezing and salt precipitation could have caused uneven heat generation through radioactive decay of 40K in salts/brines. This scenario may explain the south pore thermal anomaly [Spencer et al., 2006]. Throughout history or episodically, tidal motions of the low-viscosity eutectic slush (ice-salt-brine) could have kept the brine unfrozen well below the freezing point of H2O. An accumulation of condensed organic compounds (oils) at the bottom of the ice shell also lowered the viscosity at the ice-salt interface and eased generation of tidal heat [Zolotov, 2007]. Brines and/or oils caused at least local decoupling of the ice shell. If a transient aqueous phase on Enceladus exists in the present epoch, it could be the part of salt-ice-brine (+organic) eutectic mixture at the ice-salt (rock) boundary, preferably below the south pole region. However, the lack of firm detection of Na [Schneider et al., 2007] and Cl at Enceladus implies generation of gases in the ice shell and is not consistent with an aqueous (oceanic) origin the plume. The lack of detection of Na and Cl does not exclude Na-Cl-rich brines at the bottom of the ice shell on today's Enceladus.
 This work has benefited from conversations with Mikhail Mironenko, Christopher Glein, Everett Shock, and review comments from Jeffery Kargel. I thank Giles Marion and Mikhail Mironenko for sharing their FREZCHEM codes. This work is supported by the NASA Outer Solar Planet Research and Cosmochemistry programs.