Geological CO2 sequestration is proposed as one method for mitigating the effects of fossil fuel burning on global climate [e.g., IPCC, 2007]. One of the outstanding scientific questions relates to the potential for injected CO2 to leak and contaminate overlying freshwater aquifers [Gale, 2004; Smyth et al., 2009]. CO2, injected into geological formations in a gaseous or supercritical state, is likely to dissolve into formation fluids at the gas-water interface, forming acid brines [Bickle, 2009; Dubacq et al., 2012]. These brines can mobilize trace metals via desorption of metals bound to mineral surfaces, or by dissolution of trace metal-bearing phases [Bradl, 2004; Yin et al., 1996]. If these brines were to migrate into overlying freshwater aquifers, then there is concern that drinking water supplies could be affected. Models [Wilkin and DiGiulio, 2010; Zheng et al., 2009] and experiments [Little and Jackson, 2010; Lu et al., 2010] have shown that CO2 can mobilize trace metals; however, field experiments have not shown hazardous levels of contamination [Keating et al., 2010; Kharaka et al., 2010]. Of critical importance is whether the mobilized metals remain in solution or are redeposited locally.
 CO2 has been escaping along the Little Grand Wash and Salt Wash grabens for at least 400,000 years, based on U-Th dating of surface carbonate deposits. [Kampman et al., 2012; Dockrill and Shipton, 2010]. Near the town of Green River, Utah, the red-bed Entrada sandstone, exposed by the Salt Wash graben, exhibits spectacular bleaching in which hematite grain coatings are removed and trace metals released from the grain coatings are concentrated at the bleaching reaction front (Figure 1). Wigley et al.  argue that the bleaching results from inputs of CO2-charged brines along the Salt Wash fault, distinguishing the site from hydrocarbon-related bleached sandstones elsewhere on the Colorado Plateau [Wigley et al., 2012; Loope et al., 2010; Beitler et al., 2005, 2003; Garden et al., 2001; Chan et al., 2000]. The CO2-promoted reaction fronts are used as an analogue for processes occurring where CO2 leaks from injection sites into overlying aquifers. Here we quantify trace metal enrichments and corresponding time-integrated fluid fluxes [Bickle and Baker, 1990] at the bleaching reaction fronts, and use these to calculate the fraction of mobilized metals which have been redeposited. This allows an estimate to be made of the concentration of metals in the solution leaving the CO2-promoted reaction front.
1.1 Green River Natural Analogue Site
 Within the field area, the Jurassic Entrada sandstone is a well-sorted sediment of uniform grain size and mineralogy. Regional burial diagenesis includes the development of Fe-oxide grain coatings [Cullers, 1995; Trimble, 1978], giving the rock its red color. Along the axis of the Green River anticline, the Entrada sandstone has been bleached by diagenetic fluids which have dissolved the Fe-oxide grain coatings [Wigley et al., 2012]. Bleaching occurs at the base of the formation in a broad domal structure and exposed over a 1–2 km east-west section along the Salt Wash graben (Figure 2a).
 Bleaching of red bed sandstones is a common worldwide occurrence, with the bleaching being a result of acid-reductive dissolution of hematite grain coatings, and mobilization of otherwise insoluble Fe+3 to soluble Fe+2. Such fluid-rock reaction may be promoted by a wide range of common volatile species (CH4, CO2, H2S, hydrocarbons, and organic acids) present in geological fluids that modify fluid Eh-pH. Several lines of evidence suggest that the fluid causing the bleaching of the Entrada sandstone in the Salt Wash graben was a CO2-charged brine with minor methane.
 Calcite cements and veins associated with the Green River bleaching have a heavier δ13C than hydrocarbon-related bleaching, consistent with deposition from isotopically heavy C-rich fluids, similar to those in the modern, actively leaking CO2 system at Green River that underlies the study site [Assayag et al., 2009; Kampman et al., 2009], but inconsistent with isotopically light methane-saturated fluids. Mobilization of low pH-soluble trace metals is inconsistent with methane-rich fluids which have pH ≈7.6 when saturated with Fe-oxides (Figure S1 in the supporting information).1 Regionally, such bleaching is typically localized to structural highs, implicating a low density methane-saturated bleaching fluid. At Green River, bleaching is restricted to the base of the formation, and the morphology of bleached zone is consistent with the flow of a dense fluid as a gravity current, discrediting the role of buoyant methane-saturated bleaching fluids. If the bleaching fluids contained significant H2S, then hematite would be immediately reprecipitated in situ as pyrite and would not result in the observed Fe depletion. While some pyrite is observed in fractures, the lack of pyrite (or its Fe-bearing weathering products) throughout the bleached sandstones precludes involvement of significant H2S.
 Microthermometry of single-phase aqueous and two-phase liquid-vapor fluid inclusions hosted in quartz overgrowths and gypsum, petrographically linked to the bleaching, suggests that the bleaching fluid was a low temperature (≈27°C) brine (salinity 2.5–7 wt%) [Wigley et al., 2012]. Raman spectroscopic analysis of successfully targeted vapor bubbles reveals gas-phase compositions dominated by CO2, with minor methane (11–27 vol%) present in three inclusions (Figure 3) [Wigley et al., 2012]. The presence of a free-phase CO2 bubble suggests the aqueous phase is saturated with CO2 at the inclusion pressure. The magnitude of the Fermi-diad separation in the Raman spectra of CO2 is sensitive to the CO2 density, and this separation has been experimentally calibrated for CO2 density in the vapor phase, for gaseous CO2 to liquid CO2, in the density range of 0.01–1.20 g/cm3 [Rosso and Bodnar, 1995]. Similarly, the Raman peak position of the symmetric C–H stretching mode in CH4 is dependent on pressure, in CH4 bearing inclusions [Lu et al., 2007]. Measurements of the Fermi-diad separation in CO2 spectra reveals CO2 densities of 0.04–0.09 ±0.02 g/cm 3 which equates to CO2 pressures of 2.28–4.56 ±1.03 MPa and formation depths of 202–413 ±97 m, assuming hydrostatic pressures. Inclusion pressures based on Raman peak position in CH4 bearing inclusions [Lu et al., 2007] reveal inclusion pressures of 2.59–3.98 ±1.34 MPa. At these inclusion, pressures dissolved CO2 concentrations in equilibrium with a CO2 vapor phase are 0.65–1.30 mol/L, based on the solubility models of Duan et al. . Based on burial curves for the Entrada Sandstone in the area surrounding Green River [Nuccio, 2000], modified for local variations in burial depth due to the structural high formed by the Green River anticline, the interval from which these samples were taken passed through the burial depth range 200–400 ±97 m between 2.1 and 5.5 ±1.5 Ma, and this gives an approximate age of the bleaching event.
 The upper bleached-red contact is sharp and subhorizontal except in regions of fractures and granulation seams, where differences in porosity and tortuosity enhance or retard vertical transport of the reaction front, respectively (Figures 2b–2c). To the east, the bleached zone terminates as a series of meter scale fingers into red sandstones, while the western extent is not exposed. It is thought that the CO2-rich brine migrated as a gravity current, with the main components of flow to the east, down dip and in the direction of regional groundwater flow [Kampman et al., 2009], with vertical transport dominated by diffusion except in areas containing faults and open fractures (Figure 2b) [Wigley et al. 2012].
 The contact between bleached and unbleached sandstone is characterized by stepped profiles in bulk rock Fe-concentration, hematite and K-feldspar abundances, which occur over 10–15 cm, and demarc the final position of the acid-reductive front. A fine intergrowth of oxide and carbonate is deposited in a 2–3 cm region spanning the reaction front [Wigley et al., 2012]. Bulk rock trace metal enrichments occur systematically in a narrow zone within 5–10 cm of the upper subhorizontal contact (Figure 4a–4h), due to release of trace metals as grain coating Fe-oxides were dissolved, with subsequent redeposition at the reaction front [Wigley et al., 2012]. The enriched metals are characteristic in having high solubilities at low pH [e.g., Benjamin and Leckie, 1981], and this mobilization-redeposition is thought to be fundamentally driven by acid-reductive dissolution of the Fe-oxide grain coatings at low pH, with subsequent transport of the trace metals in solution and redeposition at the reaction front as pH rises due to a combination of pH buffering silicate dissolution reactions and secondary mineralization. The scale of enrichments is determined by the volume of fluid that interacted with the reaction front, which had previously reacted with unaltered rock.
 Bleaching also occurs around open fractures (Figure 2c), which are mineralized with carbonate and Fe-oxides, and exhibits the same geochemical and petrological signatures as the subhorizontal bleaching contacts, indicating a common parent fluid and continuity in reactive-transport processes. The fracture-associated bleaching reaction front initiated at the plane of the fracture and migrated laterally by diffusion and advection of the fluid. The asymmetry of the bleached halos is due to the transport away from the fractures being imposed on unidirectional horizontal regional flow in the aquifer [Wigley et al., 2013]. Spikes in trace-metal concentration are observed within 5–10 cm of the red-bleached transition adjacent to the fractures (Figures 4i–4l), similar to those across subhorizontal contacts. These are assumed to be related to deposition of metals released by the bleaching reaction front as it propagates from the fracture, in addition to any metals present in the bleaching fluid entering the fracture, due to the increase in pH across the reaction front.
 Key questions include the following: (1) What was the original distribution of the metals? (2) How, and in which phases, are the metals redeposited? (3) What fraction of the mobilized metals is redeposited? (4) Does the concentration of metals in the fluid exceed drinking water guidelines?