Chemical Geology 518 (2019) 32–44

Contents lists available at ScienceDirect

Chemical Geology

journal homepage: www.elsevier.com/locate/chemgeo

Chemical and isotopic evolution of groundwater through the active Andean arc of Northern Chile T ⁎ L.V. Godfreya, , C. Herrerab, C. Gamboab, R. Mathurc a Department of Earth and Planetary Sciences, Rutgers University, Piscataway, NJ 08854, USA b Departamento de Ciencias Geológicas, Universidad Católica del Norte, Antofagasta, Chile c Department of Geology, Juniata College, Huntingdon, PA, USA

ARTICLE INFO ABSTRACT

Editor: Donald Porcelli The role of rock-water reactions, leakage of sedimentary brine, evaporite dissolution and variations in lithology Keywords: on the chemical and isotopic evolution of groundwater draining the active arc and Preandean basins in the Lithium isotopes Antofagasta Region of northern Chile has been investigated using Li, B and Sr isotopes. The studied basin is that Boron isotopes of the upper Loa river and its sub-basins, which is adjacent to the basin occupied by the Salar de Atacama where Strontium isotopes Li-rich brines are commercially exploited. Due to the arid climate, water draining this part of the Andes does so Water-rock interactions primarily through the subsurface. Water from a thermal source is a common component of water sampled, but its Northern Chile geochemical signature evolves through multiple processes including dilution, cold rock-water reactions, influxes Volcanic arc of sedimentary brine and evaporite mineral dissolution. These processes lead to a range of Li and B con- centrations in groundwater (0.01–14 mg/L and 0.18–20 mg/L) and isotope compositions (−1 to +12‰ and −6 to +14‰). The effect of rock dissolution on δ7Li and δ11B within aquifers is apparent from their low values, but even in streams these values only undergo small increases relative to other rivers worldwide. This may be due to the weathering of ignimbrites producing water with less Al and limited clay formation, the large reservoir of thermally-sourced Li and B of which only a small component is removed, and because the streams acquire groundwater through their beds. The wider range in δ11B is a result of variable lithology and dissolution of evaporite minerals, which have lesser effects on δ7Li. Our multi-isotope approach highlights the role of lithology and different processes in defining groundwater chemistry. Isotopic signatures derived from marine sediments are recognizable in some springs even though this type of rock has nearly no surface exposure due to extensive volcanic rock cover, and are acquired directly or via incorporation of these sediments into volcanic systems. These high-Cl rocks can weather to produce water with a chemical composition like that of sedimentary brine but is distinguishable through δ11B composition. Combination of 87Sr/86Sr, δ7Li and δ11B also indicate that mixed salts which contribute to groundwater draining the ignimbrite covered areas of the basin did not form following cold temperature weathering, but from water where thermal contributions dominated solute budgets keeping δ7Li low, and Li/Cl and B/Cl high compared to other fluid reservoirs such as seawater. We propose that these saline systems lay within the confines of one of the multiple late-Miocene-Pliocene calderas following the active stage of volcanism and while heat gradients re- mained high.

1. Introduction Herrera et al., 2016). Chemical and isotopic characteristics combined with volcanic activity and high heat input means that many ground and The objective of this study is to determine the source and evolution surface waters can be identified as thermal or mix with thermal water of solutes in groundwater draining the volcanically active Andean arc of (Aravena and Suzuki, 1990; Risacher et al., 2011; Rissmann et al., northern Chile where climate is arid. Weathering of volcanic and se- 2015). The arid climate and dominance of closed basin morphology has dimentary-evaporitic rocks contributes the majority of the solute load, produced an abundance of salt pans (salars) which contain evaporitic while inputs of airborne desert dust only affects the most dilute waters salts and brine, which can be enriched in lithium and boron. Of the (Risacher et al., 2003; Boschetti et al., 2007; Rissmann et al., 2015; evaporitic salts which have formed in the past, only gypsum is recycled,

⁎ Corresponding author. E-mail address: [email protected] (L.V. Godfrey). https://doi.org/10.1016/j.chemgeo.2019.04.011 Received 6 July 2018; Received in revised form 1 April 2019; Accepted 8 April 2019 Available online 15 April 2019 0009-2541/ © 2019 Elsevier B.V. All rights reserved. L.V. Godfrey, et al. Chemical Geology 518 (2019) 32–44 suggesting that brine has leaked from these basins into regional Desert (Risacher et al., 2003). The solute load of surface and ground- groundwater flow (Risacher et al., 2003; Rissmann et al., 2015; Herrera water may originate from low temperature weathering and recycling et al., 2016). The thick (~1 km) halite fill of the Salar de Atacama through salars (Risacher and Fritz, 2009) or high temperature weath- (Jordan et al., 2002), suggests this basin is the discharge point of many ering and volcanic sources (Rissmann et al., 2015). We aim to address groundwater flow paths, and here Li concentrations in the brine can these different mechanisms using isotope tracers. exceed 5000 mg/L (Munk et al., 2018). With the increase in global demand for lithium, a better understanding of the processes which 2. Geological and hydrological description of the study area occur in the weathering and transport of solutes is needed. Lithium and boron are elements whose elevated concentrations in 2.1. Geology of Western Cordillera of the Andes groundwater have been used as tracers of hydrothermal activity (Ellis and Mahon, 1964; Giggenbach and Soto, 1992; Fouillac and Michard, The study area focuses on a pre-Andean basin which occupies the 1981). However, Li and B are also considered to be somewhat con- area between the crest of the Western Cordillera of the Andes and the servative during evaporation leading to high concentrations in sedi- Domeyko range which forms the Precordillera. In the upland area, the mentary brine (McCaffrey et al., 1987). While these two mechanisms climate is arid to semi-arid, but at lower elevations to the east it tran- for raising Li and B concentrations are very different, both can operate sitions to hyper-arid. Mean annual precipitation varies from < 50 mm in volcanically active areas under dry climate regimes, such as the ac- at elevations below 2500 masl to > 150 mm above 4000 masl (Houston, tive Andean arc of northern Chile. Use of isotopes (δ7Li and δ11B) has 2007); mean annual temperature ranges are > 15 °C and < 4 °C at the done much to elucidate their behavior during high temperature water- same elevations. The basin of interest is that of the upper Loa River, rock reactions and evaporation. The temperature dependence of isotope which drains to the ocean. To its south, and formed within virtually fractionation causes the extraction of Li and B from rocks at high geologically identical units, is the internally drained Salar de Atacama temperature with transfer of the rock isotopic composition to solution basin, the host of lithium rich brine. (Millot and Négrel, 2007; Millot et al., 2007). While changes in the Li The elevation of the magmatically active Western Cordillera isotope composition of thermal waters was found to be related to re- averages 3800–4500 masl and locally reaches 6350 masl. The volcanoes servoir temperatures based on other chemical geothermometers (Millot unconformably overlie a complex basement consisting of early and Négrel, 2007; Millot et al., 2007), they found that the isotopic Paleozoic crystalline basement rocks and Mesozoic sedimentary, vol- heterogeneity of lithology made temperature a secondary effect for canic and plutonic rocks. Volcanic rocks consist of andesitic and dacitic defining δ11B. At the low temperatures typical of most surface en- lavas and voluminous, but compositionally homogenous calc-alkaline vironments, Li and B are incorporated into secondary minerals which dacitic and minor rhyolitic and andesitic ignimbrites which erupted involves an increase in solution phase δ7Li and δ11B. The global dis- between the late Miocene and mid-Pleistocene to form one of the lar- charge-weighted average riverine flux of Li has a concentration of gest and youngest silicic volcanic fields worldwide. These eruptions 1.5 μg/L and δ7Li of +23‰ (Huh et al., 1998), while that of B is 10 μg/ occurred within the Altiplano-Puna Volcanic Complex (APVC) which L with δ11Bof+10‰ (Lemarchand et al., 2002). overlies the Altiplano-Puna Magmatic Body, some 200 km in diameter The influence of evaporation on solution δ11B and δ7Li has been less and 11 km thick (Ward et al., 2014). There are numerous endorheic well studied. The assemblage of minerals formed during evaporation basins in the Central Andes filled with alluvial material and at their can have a strong influence on the isotope fractionation of the eva- topographic low points, salars. porating water. Precipitation of halite appears to occur without isotope Immediately to the west of the Western Cordillera of the Andes are fractionation for both Li and B (Godfrey et al., 2013; Paris et al., 2010), the Preandean basins followed by the Precordillera, the Cordillera but minerals formed earlier in the evaporation sequence do. During the Domeyko. The Cordillera Domeyko is a N-S trending basement block evaporation of seawater and continental water increases up to 30‰ comprised of fault bounded Paleozoic, Mesozoic and Cenozoic crystal- have been reported for δ11B(Vengosh et al., 1992) due to the pre- line rocks and sediments, and the host of Eocene-aged giant copper ferential occupation of 10B in tetrahedral sites and incorporation into porphyry deposits. Sedimentation in the Preandean basins over much of carbonate and certain borates (Palmer and Helvaci, 1997). The role of the Cenozoic has consisted of siliciclastics, evaporites and minor car- these early precipitating minerals on fractionating Li isotopes has yet to bonate units (Dingman, 1967; Blanco and Tomlinson, 2009; Jordan be rigorously determined. While inorganically formed calcite frac- et al., 2007; Muñoz et al., 2002; Tomlinson et al., 2010; Evanstar et al., tionates Li by −2to−5‰ and aragonite by −11‰ (Marriott et al., 2015). Differences between basins reflect their drainage: the Salar de 2004) these minerals have low distribution coefficients towards Li and Atacama Basin has been internally drainage since the Paleogene and increases in solution δ7Li are expected to be minimal. contains a thick accumulation of halite, but halite is absent in the Upper We aim, by using a multi-isotope approach, to distinguish between Loa groundwater basin which drains to the Pacific(May et al., 1999; the different processes that weather rocks of variable lithology and that Jordan et al., 2007). combine to influence groundwater chemistry in this region. These in- Paleozoic sediments and volcanic/plutonic rocks outcrop to the clude low temperature silicate weathering, hydrothermal reactions, west of the upper Loa groundwater basin in the Precordillera and in a evaporation and evaporite dissolution (Ericksen, 1981; Risacher et al., few localities within the arc itself. In the southern part of the Upper Loa 2003, 2011; Risacher and Fritz, 2009; Rissmann et al., 2015). basin are Permo-Triassic plutonic rocks as well as siliciclastic sedi- Our study area is the Western Cordillera of the Andes in the ments, sometimes intercalated with volcanic rocks of Triassic and Antofagasta Region of Chile on the eastern edge of the Atacama Desert. Cretaceous age (Marinović and Lahsen, 1984). These rocks form part of The low mean annual precipitation limits dilution by meteoric water so the topographic high which separates the Loa drainage basin and the that the geochemical and isotopic compositions of groundwater and Salar de Atacama basin (Marinović and Lahsen, 1984). The Cretaceous groundwater fed streams strongly reflect rock-water interactions in the Lomas Negras Formation is mostly comprised of siliciclastic sediments subsurface. Aridity also limits the development of vegetation, so that but bears carbonate sands with marine fossils (Marinović and Lahsen, meteoric water infiltrating from the surface is not made as acidic by 1984). Its outcrop is limited to the eastern end of the Turi Basin im- respired CO2 or organic acids, reducing the aggressiveness of meteoric mediately west of the El Tatio geothermal field. East of the western waters towards silicate weathering. The aridity, persistent for much of crest of the Andes are Early Late Cretaceous continental sediments of the Neogene, and particularly severe since the early Holocene, has the Salta Rift System in Argentina and southern Bolivia, separated from produced many evaporite beds and brine which occupy salt pans contemporaneous eastern Chilean depocenters by the Huaytiquina High (salars). By dissolution or dilution, they contribute substantially to so- basement massif (Mpodozis et al., 2005). During the Late Cretaceous- lute loads draining the segment of the Andes adjacent to the Atacama early Cenozoic, transgression of the El Molino – Yacoraite Sea over

33 L.V. Godfrey, et al. Chemical Geology 518 (2019) 32–44 northwestern Argentina left limestone and sandstone sediments saturation states were calculated using the GWB software (Bethke, (Mpodozis et al., 2005). Any connection between the Cretaceous – early 2011) which incorporates major ion concentrations into the calculation. Cenozoic basins of northern Chile and the shallow marine systems to Cations were measured at Rutgers University. Major cations, Si, Li the east are obscured by volcanism of the western Cordillera. The and Sr were measured using a Varian ICP-OES; Br, B and Rb con- Lomas Negras Formation, as well as a marine limestone protolith of centrations were measured using a Thermo iCAP-Q at Rutgers xenoliths in the Lascar volcano bordering the Salar de Atacama University using a Teflon introduction system with trace HF in the (Matthews et al., 1996) suggest that these marine sediments may reach carrier acid. Concentrations were determined using synthetic standard as far west as the crest of the active Western Cordillera. mixtures and accuracy checked with certified standards and seawater. Volcanism in the Western Cordillera has produced stratovolcanoes Chloride and sulfate were measured at Juniata College by ion chro- and massive calderas, the source of extensive and thick ignimbrite units matography with each reported value representing the mean of 3 dif- which create the ignimbrite plateau of the central Andes. Orthogonal to ferent measurements (the 2σ variation of the samples is < 3% of the the roughly N-S strike of the Andean crest are lines of volcanoes which reported value). form sub basins filled with volcano-clastic deposits, ignimbrites and Strontium isotopes were measured following chromatographic se- tuff, as well as recent salar deposits and mud (Houston, 2007). These paration using AGW50x8 or Sr-spec resin using either a VG Sector 54 basins have surface drainage to the Loa River, whereas east of the Salar TIMS at Cornell University or a ThermoScientific Neptune Plus MC- de Atacama basin they are small and endorheic. Although topo- ICPMS at Rutgers University. Measurement of NBS 987 was included graphically separated, these endorheic basins may be hydraulically with all analyses, and average 0.710245 ± 0.000006 (2σ, n = 40) by connected at depth, and leakage from these basins can profoundly in- TIMS or 0.710270 ± 0.000004 (2σ, n = 13) by MC-ICPMS. Analyses fluence the geochemistry of groundwater and may obscure geochemical by MC-ICPMS were then corrected based on those values of NBS 987 signatures of geothermal activity and primary low temperature silicate values to match those of the TIMS instrument. Lithium isotopes were all weathering (Risacher et al., 2003, 2011; Risacher and Fritz, 2009). analyzed at Rutgers University by MC-ICPMS following separation The surface consists of gavels and alluvial material where clays using two cation exchange (AGW50x12) columns 200–400 mesh (7 mL minerals can be notably absent (Sun et al., 2018). Where soils exist they resin bed followed by 0.2 mL resin bed) using 0.5 N HCl. A sample of are gypsic in the drier low elevation areas, but older paleosols are calcic seawater was included with all sample batches and separated using Vertisols or calcic Argillisols (Rech et al., 2006). All soils contain different column pairs to ensure that yields were quantitative. Lithium montmorrilonite and illite in addition to oxides, calcite or gypsum. The solutions were analyzed in solutions of 20 ppb concentration, which, volcanic units have alteration zones with smectite, illite and chlorite utilizing a Teledyne Cetac Aridus II and X-type Ni skimmer cone, (Maza et al., 2018). Erosion of the volcanic summits has exposed hy- yielded a signal exceeding 18 V on 7Li with an acid blank of < 150 mV. drothermal alteration including kaolinite plus elemental sulfur. The high sensitivity of the instrument allows small samples to be pro- cessed though ion chromatography which enables column calibrations 2.2. Groundwater flow to be stable. Standard-sample-standard (L-SVEC) bracketing was used, matching signals to within 5%. A solution of IRMM-016 was included in Information regarding groundwater flow has been compiled from each batch of analyses, and the long-term δ7Li average of that, and data published by the largest mining companies and inferred between seawater, were 0.1‰ ± 0.07 (2σ, n = 25) and +30.6‰ ± 0.12 (2σ, these points because large parts of the basin aquifers cannot be accessed n = 18). Measurement of boron isotopes followed separation using the for observation. Additional understanding is gained from the hydro- micro-sublimation technique described by Wang et al. (2010). Samples geology of the springs in the Western Cordillera and Precordillera to were run in duplicate and on the rare occasion (1 in 40 samples) when a determine the geological units though which groundwater flows. duplicate gave δ11B which differed by > 0.5‰, the sample was re- Overall, groundwater flow is to the west through fractured volcanic analyzed in duplicate. A seawater sample was included in every sub- aquifers, unconsolidated sediments and alluvial fans. Streams are limation batch. Measurement of δ11B was made with the Thermo groundwater fed. Within the arc, shallow volcanic aquifers overlie an Neptune Plus at Rutgers University using a PFA-barrel or PFA micro- ancient erosion surface which may direct flow. Groundwater flow is concentric spray chamber (ESI) and diluted in a solution of 2% HNO3 also determined by tectonic structures, which can act as barriers or and 0.1% HF. The signal on 11B was about 350 mV for a 30 ppb solu- focus it. Examples of structure-related barriers are horsts and reverse tion, the acid blank was < 2.5 mV. Sample-standard bracketing was faults which form wetlands such as the Vegas de Turi on the north side used with NIST 951a. A sample of Sargasso seawater was included in of the Salado River and parts of the San Pedro basin (Houston, 2007). the micro-sublimation step with each sample batch, and solutions of the Folds also provide barriers and redirect flow, for example the N-S ridges BAM standards AE120 and AE122, as well as an acid blank, were run of the Peine Formation (Kuhn, 2002) prevent direct east to west flow of periodically throughout each analysis. Standards AE120 and AE122 groundwater towards the southwest sector of the Salar de Atacama. yielded −19.9‰ ± 0.08 (2σ, n = 13); +39.5‰ ± 0.14 (2σ, n = 13), Within the Calama Basin groundwater flow has been described in detail and Sargasso seawater, +40.3‰ ± 0.14 (n = 11). by Jordan et al. (2015) and is predominantly to the southwest. Where sedimentary units thin across the ChiuChiu monocline, formed by dis- 4. Results solution of underlying gypsic beds (Blanco and Tomlinson, 2009), the resulting constriction may limit water from the Salado catchment Locations, field measurement, concentration and isotope data are flowing directly into the basin aquifers (Jordan et al., 2015). listed in Table 1.

3. Samples and methods 4.1. pH, major and minor ion concentrations

Samples were collected from springs, groundwater fed rivers and The pH range is between 6 and 8.5, but most samples are neutral to wells (Fig. 1). Sampling employed plastic syringes, peristaltic pumps slightly alkaline, averaging 7.7 ± 0.6. Dissolved inorganic carbon and silicone tubing or in-situ submersible pumps. All samples were (DIC) is high and varies between 1 and 10 meq/L. The range of partial –3.5 filtered to < 0.4 μm in the field and stored in new, acid cleaned poly- pressures of CO2 with which water is in equilibrium is wider, 10 to –0.8 ethylene bottles. Samples were not acidified. Temperature and pH were 10 atm. Samples with the highest pCO2 and lowest pH were col- measured in the field, and alkalinity on the same day by acid titration lected from a spring system near the crest of the Altiplano and from after being stored in glass bottles or by a field checker (Hanna Instru- flowing wells and springs at the eastern end of the Calama basin. ments). The distribution of different carbon species and mineral Important contributions of magmatic CO2 are expected from samples

34 L.V. Godfrey, et al. Chemical Geology 518 (2019) 32–44

a

b

Fig. 1. (A) Oblique topographic map of the Loa Basin showing the association of the PreAndean basins occupied by the Upper Loa River and the Salar de Atacama and their position relative to the Altiplano and western Cordillera of the Andes. Semi-transparent ovals show the approximate locations of late Miocene – Pliocene calderas. The inset map shows country borders. (B) Geological map with sample locations. located close to the volcanic front but carbonate dissolution by water member ALT has a higher proportion of carbonate than AND. The Silala acidified by reactions with volcanic gases, such as CO2,SO2 and HCl is stream plots closest to end member ‘ALT’, and the Paniri spring to end another important way in which CO2 becomes supersaturated relative member ‘AND’. The samples which plot between ‘IG’ and ‘ALT’ are two to the atmosphere. of the wells in the eastern Calama Basin, spring water at Chitor and The major cation and anion compositions of ground and surface Linzor 1. Samples which to plot towards ‘AND’ are Loa river water near water produce two trends on a Piper diagram (Fig. 2, which includes the Mino volcano, Ojo San Pedro, one well in the western Calama Basin, data from the DGA (Direccíon General de Aguas) for nearby basins; and Baños de Turi. Stratovolcanoes are dominant in the landscape Risacher et al., 1999). End-member (IG) is dominated by (Na + K) and around these sample locations. Two outlying samples are the spring at – (Cl + SO4), with Cl being the dominant anion. Geothermal fluids from the base of Volcan Chela and Linzor 2. Chela has low HCO3 and Mg, – El Tatio are coincident with this point, and data from the Salado and and Linzor 2 has high HCO3 and is distinguished from end-member Caspana rivers, and the Chitor spring plot close by. These samples come ALT by high Na + K. from areas of the drainage basin dominated by ignimbrites. The other Silica concentrations are between 27 and 50 mg/L. The sample two end-members (AND and ALT) get 60–70% of their positive charge obtained from the Tatio geothermal field has a much higher value, from (Ca + Mg) but differ in their carbonate - sulfate ratio. End- 94 mg/L. Bromide concentrations vary between 0.01 mg/L and 12 mg/

35 ..Gdry tal. et Godfrey, L.V.

Table 1 Concentration and isotope data. UTM coordinates WGS84-19K. Concentration units are mg/L. Sr isotope data normalized to NIST 987 = 0.71024, Li isotope data relative to LSVEC, B isotope data relative to NIST 951.

87 86 7 11 UTM (E) UTM (S) Elevation (masl) pH HCO3 Ca K Mg Na Cl SO4 Br Si Sr Rb Sr/ Sr Li δ Li B δ B

Upper Loa groundwater basin Loa at V Miño 535399 7658658 3810 8.20 142 45 14 20 103 91 186 0.43 35.5 0.05 0.70636 0.6 5.1 3.7 6.1 Loa at Taira 541964 7583565 3270 7.05 351 51 25 56 316 554 252 1.41 31.8 0.07 0.70591 0.7 5.3 13.8 10.6 Loa at Santa Barbara 540206 7569821 3070 8.00 37 20 48 233 423 352 1.02 36.1 0.90 0.06 0.70589 0.8 5.9 9.7 9.1 Loa at Conchi 540492 7568845 3030 8.00 40 27 83 356 614 407 1.04 39.0 1.85 0.09 0.70606 0.7 7.9 9.3 9.4 Chela 543891 7639944 3880 7.60 59.4 81 12 10 91 179 101 0.29 29.3 0.78 0.07 0.70512 0.1 6.4 4.5 7.7 Loa at Angostura 527107 7516404 2470 8.02 114 75 98 1071 1845 361 0.8 5.38 0.33 0.70743 4.0 1.5 16.8 −1.0 Salado water 545347 7530138 2630 8.20 400 56 68 54 784 1492 40 1.4 43.9 3.89 0.33 0.70763 4.2 1.2 11.9 −4.0 El Tatio 601638 7529871 4280 7.06 84 145 233 4 2180 3886 40 3.2 94.3 2.30 0.70900 20.0 −0.5 88.0 −5.4 −

36 V d. Barro stream 601631 7525750 4370 8.22 49 9 19 19 65 139 0.012 30.8 0.30 0.04 0.70897 0.0 10.1 0.2 5.3 Silala at Inacaliri 596622 7563893 4140 8.50 113 18 5 8 23 39 11 0.025 27.5 0.10 0.02 0.70816 0.1 8.8 0.3 −3.1 Baños de Turi 574139 7540704 3090 7.90 401 117 45 46 432 733 251 0.126 46.2 3.47 0.21 0.70779 1.3 0.0 3.7 −2.9 Paniri 575126 7550635 3265 8.20 243 51 3 14 33 135 12 0.238 39.7 0.41 0.00 0.70716 0.0 1.8 0.4 13.4 Caspana 578953 7533247 3170 8.20 383.3 41 36 41 789 1229 312 0.126 27.3 1.62 0.17 0.70791 0.5 4.1 1.7 2.6 Linzor 1 600667 7541446 4105 7.53 70.4 10 11 5 70 148 29 0.069 33.0 0.10 0.04 0.70891 0.4 0.6 1.6 −6.1 Linzor 2 " " 4105 7.56 346.5 11 10 5 66 101 49 0.06 34.6 0.08 0.04 0.70888 0.3 0.7 1.5 −5.6 Chitor spring 585204 7520822 3770 6.01 603.7 124 16 66 803 948 270 0.031 53.6 1.19 0.10 0.70827 0.2 2.5 0.8 −6.9 Chitor stream " " 3740 8.28 550.5 117 15 64 815 1160 96 0.034 49.9 1.18 0.10 0.70826 0.1 3.0 0.8 −7.0 Angostura spring 527578 7516518 2475 6.62 557 118 61 95 1058 1940 201 0.8 39.3 5.17 0.30 0.70743 3.7 1.1 14.9 −1.0 Ojo San Pedro 566626 7569569 3800 7.15 329 82 17 54 171 185 291 0.15 48.5 0.58 0.07 0.70783 0.3 2.1 2.9 1.0 PCAR7C 508504 7520800 2640 6.86 564 112 30 35 340 448 143 0.3 40.5 1.27 0.09 0.70625 1.4 5.1 3.9 −0.9 TB05 533062 7534705 2630 6.65 567 83 22 58 212 323 77 43.2 1.10 0.01 0.70642 0.3 2.8 1.7 Si37E 533000 7532628 2610 6.90 618 184 20 57 270 654 77 0.59 27.6 1.27 0.02 0.70587 0.7 6.2 2.5 2.2 Precordillera ChugChug 484593 7554130 2170 7.71 77.2 587 234 1919 16,227 18,222 18,312 11.2 0.7 1.68 0.70822 1.0 10.3 76.1 22.5 Chemical Geology518(2019)32–44 L.V. Godfrey, et al. Chemical Geology 518 (2019) 32–44

Fig. 2. Piper diagram. Additional samples (Giggenbach, 1978; Risacher et al., 1999; Tassi et al., 2010; Rissmann et al., 2015) plotted for Monturaqui-Negrillar-Tilopozo (MNT) system gray triangles; Tuyaito (TUY), Capur (CPR) and Agua Calientes 3 (AC3), gray hourglass, Laguna Miscanti (MISC) and Laguna Lejía (LL), gray diamond; Torta Tocorpuri, Pampa Apacheta Puritama, open gray star, El Tatio, black star; Tatio brine black open circle.

– 2– L, the lowest concentrations were measured in samples collected in the HCO3 +CO3 is virtually zero. Brackish and saline water collected Turi Basin and at Chitor, the highest measured at ChugChug. The lowest close to salt pans plot along this axis because carbonates are one of the 2− − Cl/Br occur in the headwaters of the Loa, and the highest in some of the first minerals to form as water evaporates, while SO4 and Cl ac- springs in the upper Salado basin. The concentration of Li varies be- cumulate. Fluctuating lake levels can allow meteoric and shallow tween 0.01 mg/L and 20 mg/L, though it is only El Tatio water which groundwater to wash over and dissolve sulfates and halite formed 2− − contains > 4.2 mg/L Li. The concentration of B varies between 0.2 mg/ earlier, contributing SO4 and Cl . Whether this type of evaporated or L and 88 mg/L, only El Tatio water contains more, 17 mg/L B. salt-bearing water characterizes all water draining the arc is less clear. Data for groundwater in the MNT (Monturaqui-Negrillar-Tilopozo) aquifer at the south end of the Salar de Atacama collected from wells 4.2. Sr, Li and B Isotopes belonging to BHP-Billiton (Rissmann et al., 2015) plot between AND and ALT, unlike data for nearby surface water reported by Risacher 87Sr/86Sr varies between 0.705 and 0.709. The samples which have et al. (2003) which trend towards endmember IG. The differences could the lowest 87Sr/86Sr (< 0.706) are upstream of the confluence with the reflect different sampling between studies, with a water that has dis- outlet from San Pedro basin plus one of the flowing wells in the east end solved evaporites being more predominant in the Risacher et al. (1999) of the Calama basin. The samples with the most radiogenic Sr are close study. However, the role of geothermal water in defining water to the crest of the Altiplano: El Tatio, the stream near Volcan de Barro, chemistry is also unclear because the three geothermal fields within the and Linzor. Groundwater at ChugChug, which drains Upper Jurassic study area are chemically diverse. In the case of El Tatio, its very high marine limestone has 87Sr/86Sr of 0.70822, slightly more radiogenic Cl content could reflect a magmatic brine, or its composition is influ- than the seawater Sr isotope curve for this period, indicating minor enced by leaching of evaporite beds (Giggenbach, 1978; Cortecci et al., influence of more radiogenic silicate rocks of the Cordillera Domeyko in 2005). In comparison, water is dominated by SO4 at Pampa Apacheta weathering processes within its catchment. – and water from Torta de Tocorpuri is HCO rich (Tassi et al., 2010). δ7Li varies between −0.5‰ and +10.1‰. El Tatio has the lowest 3 Consequently, the contribution of evaporated saline water versus geo- δ7Li followed by samples located near it or associated with the San thermal water to groundwater is not resolved. Pedro-Linzor volcanic chain. Samples where δ7Li exceeds +6‰, and which tend to have the lowest Li concentration are surface water and groundwater in the eastern Calama Basin. The range in δ11Bis−7.0‰ 5.1. A model of water sources based on soluble elements Li, B and Br to +13.4‰. The lowest values occur in samples close to the Altiplano, and the highest in the Río Loa upstream of its union with the Río Li and B are elements which are enriched in hydrothermal water Salado, and the Paniri spring. The Precordillera spring ChugChug has and evaporitic brine, and this makes it difficult to assign their enrich- δ7Li of +10.3‰, and δ11B of +22.5‰. ment to a single process (Risacher et al., 2011). Ternary diagrams of Cl, B and Li have been used to distinguish geothermal water from per- 5. Discussion ipheral groundwater, and to assess their maturity (Giggenbach and Soto, 1992; Cortecci et al., 2005), but the presence of Cl from even The importance of evaporation in affecting the major element small contributions by inputs of sedimentary brine or salt dissolution 2— geochemistry is clear when brine data for nearby closed basins reported makes this type of plot very complex. Even on the Cl-SO4 tAlk plot by Risacher et al. (1999) are included in the Piper diagram (Fig. 2). (Fig. 2), the field for mature Na-Cl hot springs (Giggenbach, 1988), and These data plot between mixing line IG-AND and the axis where the acidic thermal Tatio brine in particular (Giggenbach, 1978), is

37 L.V. Godfrey, et al. Chemical Geology 518 (2019) 32–44 consistent with any cold water which might have encountered eva- porite minerals, and halite in particular. Steam heated thermal water is either enriched in S or C, for example some water at El Tatio, Apacheta and possibly Torta de Tocporpuri, which places these systems close to the axis linking the SO4 and tAlk end-members. Lithium concentrations have been noted to increase with water temperatures, with concentration and Na/Li ratio reflecting the rock reservoir. Lithium concentrations relative to major elements are slightly lower than the rock because Li is taken into secondary quartz while temperatures remain above ~200 °C (Giggenbach and Soto, 1992). As waters cool further, Li is incorporated into secondary minerals and especially into clays with octahedral sites (Fouillac and Michard, 1981). Boron is enriched in young geothermal systems by dissolution of magmatic vapor (Smith et al., 1987) but is depleted through boiling, adsorption to clay or precipitation of borate minerals (Giggenbach and Soto, 1992; Chong et al., 2000). As a result of these removal processes, mature geothermal systems become Cl dominated and reach neutral pH as they equilibrate with surrounding rock (Giggenbach, 1988). As geothermal water migrates away from the heat source it mixes with meteoric dominated groundwater which reacted with rock at low temperatures. In this area of the Andes, groundwater may also be brackish to saline because surface water has dissolved surface eva- poritic salts before infiltrating or contains a component of brine which leaked through the bottom of saline lakes (Risacher and Fritz, 2009). These processes can dramatically alter compositions and obscure che- mical indicators of geothermal systems, especially those which may exist at depth with no surface expression. Li and B concentrations increase with Cl concentration, but because of potential contributions from salts or brine and processes that remove Li and B to clay minerals, there is considerable variation in their Li/Cl and B/Cl ratios (Fig. 3A and B). Bromine is a conservative element which has been used to identify salt and brine contributions to water in the central Andes among other places (Holser, 1966; Risacher et al., 2003). It has also been used with Cl to identify halite-bearing marine sediments as halide sources to the Precordillera porphyries of the Loa district (Arcuri and Brimhall, 2003). In water draining the arc, in the upper Loa drainage basin, the variation in Cl/Br (Fig. 3C) is more than double than for Cl/Li or Cl/B (a factor of 150 compared to 40 and 60). Like Br, Li is concentrated in the brine relative to halite by a factor of 10 (Holser, 1979; Godfrey et al., 2013), so precipitation or dissolution of halite should not redistribute Li relative to Br. The difference in the relationships of these three elements with Cl could reflect greater variability in Cl/Br ratios of rocks, emissions to the atmosphere from water specific to Br and not Li or B (Risacher et al., 2006) or additions of Li and B which reduce their concentration variations relative to Cl and Br. Thermal water at El Tatio, Apacheta and Torta de Tocorpuri have high Li and B compared to Br when normalized against Cl, so it is possible that inputs of thermal water to groundwater reduces the var- iation in Li/Cl and B/Cl, but not Br/Cl. Low Br relative to Li and B when Fig. 3. Lithium, boon and bromide concentrations relative to chloride. The lines Cl concentrations are high providing evidence that sedimentary brines indicate elemental:chloride ratios for seawater and mixed salts from Olaroz are the source of Cl because Br is lost to the atmosphere during eva- salar (Garcia pers. comm, 2019) and for Br/Cl, sedimentary halite identified by poration at the elevations of the Altiplano (Risacher et al., 2006) re- Risacher et al. (2011) as a source of Cl. Additional thermal water data is from quires further study. Tassi et al. (2010). Steam heated water has higher B relative to Cl. Li and B concentrations are high relative to seawater, which receives large influxes of In a geochemically complex environment of the arid western foot- non-thermal water. hills of the volcanically active Andes, and similar basins worldwide, there are too many processes that will influence elemental distributions. Even using elements that are often considered soluble and conservative, which sit atop widespread thick ignimbrite sheets erupted during the fl chemical compositions alone do not distinguish whether groundwater late Miocene-Pliocene volcanic are-up in the Central Andes. These carries a signature of geothermal systems, or is affected by dissolution ignimbrites spread westward into the Preandean basins where they of evaporites, leakage of brine from other basins, or weathering of overlie clastic and evaporitic sediments as well as older volcanic se- marine sediments. The use of isotopes is explored to distinguish be- quences (Mpodozis et al., 2005; Blanco and Tomlinson, 2009; Jordan tween these different scenarios. et al., 2007, 2015). Interbedded in the older sequences are Mesozoic marine carbonate and siliciclastic sediments, possible equivalents to the 5.2. Isotopic composition of solute sources to surface and groundwater Yacoraite Formation of Argentina (Muñoz et al., 2002; Marquillas et al., 2007), and Paleozoic sedimentary sequences (Bahlburg and Bretkreuz, The western slope of the Andes is dominated by stratovolcanoes 1991). There is limited surface exposure of these sediments east of the

38 L.V. Godfrey, et al. Chemical Geology 518 (2019) 32–44

Table 2 from Risacher et al. (1999) and B and Li isotope data reported by End-member solute sources. Concentrations in mg/kg. Schmitt et al. (2002) and Lagos (2017). Evaporite minerals which form

Na Cl Br Li δ7Li B δ11B within the salars consist of halite with minor calcite and aragonite, gypsum, sylvite and borates, primarily ulexite, but less common mi- Thermal water 3240 5400 20 40.0 −0.5 140.0 −5.0 nerals such as uralborite have also been recorded (Risacher et al., 1999; Sedimentary brine 72,000 134,000 23 68.0 7.4 225.0 6.0 Lagos, 2017; Garcia, pers. Comm., 2019). While there is almost no Mixed salts 497,000 607,000 5 63.0 2.0 135.0 −10.0 Cold dilute water NA 554 1 4.0 NA 13.8 −5.0 isotope fractionation of Li or B during halite precipitation (Paris et al., 2010; Godfrey et al., 2013) there is significant fractionation during the precipitation of earlier forming evaporite minerals. Due to the greater West Fault, but these rocks can directly interact with groundwater or be proportion of B that exists in tetrahedral coordination in the observed incorporated into volcanic systems influencing their geochemistry, and natural borate assemblages (Grice et al., 1999) and the partitioning of ultimately groundwater as well. 10B into the borate anion (Oi et al., 1989) these borates' δ11B are lower The rocks of this area are predominantly volcanic. Volcanic rocks than the solution phase or borates with equal proportions of B(OH)3 − and their clastic erosional derivatives and other silicate rocks have a and B(OH)4 (Vengosh et al., 1995; Palmer and Helvaci, 1997). Ana- limited δ7Li and δ11B variation and vary within a few per mil of the lyses of an evaporite assemblage from Tuyaito and Olaroz salars con- upper continental crust or MORB (Xiao et al., 2013; Penniston-Dorland, firm this expectation, with δ11Bof−10.3‰ and −12.4‰ ± 1.0 re- 2017). Strontium has greater variability because of the mixing between spectively (Lagos, 2017; Garcia, pers. Comm., 2019). Changes in δ7Li crystalline basement rocks and young volcanic rocks, both in terms of during evaporation are less studied, but preliminary analyses of eva- clastic sediments and magmatic mixing as expressed by ignimbrites porating seawater (Godfrey, unpubl. data) as well as inorganically (Lucassen et al., 1999; Kay et al., 2010). However, the presence of formed calcite (Marriott et al., 2004) indicates that δ7Li might be lower marine rocks will impart greater variability in terms of B and Li isotopes in non-halite evaporite minerals than Li in solution. However, the than to Sr, because seawater is very enriched in 7Li and 11B compared to amount of Li that is incorporated into early forming evaporite minerals crustal rocks (Xiao et al., 2013; Penniston-Dorland, 2017). The end- is small, because the δ7Li composition of mixed surface salts in Olaroz is member sources of Li and B are described below and listed in Table 2. within error of the surface co-existing brine (Garcia pers. Comm., The two most important types of volcanic rock which are weath- 2019). ered, are rhyo-dacite ignimbrites and andesitic lavas of the composite volcanoes; these are labeled as IG and AND in isotope plots. The 5.3. The role of lithology on δ11B and δ7Li based on radiogenic Sr 87Sr/86Sr values range from 0.7050 to 0.7095 in lavas (Walker, 2011; Godoy et al., 2014) and 0.7083 to 0.7116 for ignimbrites reflecting The 87Sr/86Sr of water is consistent with Sr released from volcanic assimilation of crustal contaminants (Kay et al., 2010; de Silva et al., rocks, lavas and ignimbrites, according to their distribution: Sr is more 2006). The δ11B composition of nearby ignimbrites range from −9.7 to radiogenic in the southeastern part of the Upper Loa Basin because it is +2.2‰ and average −4.1 ± 3.3‰ (Schmitt et al., 2002). Lavas are closer to the ignimbrite plateau of the central Andes where ignimbrites slightly more enriched in 11B, and range between −6 and +2.4‰ like the Atana and Toconao have whole rock 87Sr/86Sr above 0.710. We (Rosner et al., 2003). There is little data for δ7Li of rocks in this area. observe negative correlations between δ11B, δ7Li and 87Sr/86Sr. The Ignimbrites analyzed to date vary from +1.9 to +4.4‰ (Godfrey, samples within the ignimbrite water group overlap the field for ig- unpubl. data), and we assume that since silicate reservoirs do not ex- nimbrites in both 87Sr/86Sr - δ11B and 87Sr/86Sr - δ7Li space (Fig. 4A and hibit large variations, the lavas will be similar. B), but the samples from the Loa group have Sr isotope values con- Mesozoic marine sediments are an important Cl source in the sistent with the lavas while their δ11B and δ7Li are higher than the Precordillera (Arcuri and Brimhall, 2003), and there is evidence that expected andesitic values (Rosner et al., 2003; Deegan et al., 2016; these rocks exist under the arc. If we assume that these sediments are Magna et al., 2006). There is also an offset between surface and mixed carbonate-siliciclastics, their isotope composition will be a mix groundwater which is most notable in δ11B. A few samples plot well between contemporary seawater and detrital material according to outside of these fields, Silala and V Barro's stream have high δ7Li, and mass balance. Strontium and B are more concentrated in the carbonate Paniri spring has high δ11B. component, so we assume that 87Sr/86Sr will be close to that of sea- The Paniri spring has much higher δ11B relative to the other samples water at the time of deposition, 0.707–0.7075 (Veizer et al., 1999), and and indicates additional B sources may be present. Modern seawater δ11B slightly lower than marine limestone, +9 to +12‰ (Paris et al., has δ11B around +40‰ and limestone forming today about +21‰. 2010). Carbonates are not an important reservoir for Li, so we expect Although seawater δ11B, and hence limestone, have been lower in the that δ7Li will only be slightly elevated by seawater from the marine past, the 11B enrichment of marine limestone will always be high re- detrital component of −2to+5‰ (Chan et al., 2006). A spring lative to silicate rocks (Ishikawa and Nakamura, 1993). The role of emanating from Jurassic marine carbonate-bearing sediments in limestone weathering is demonstrated by ChugChug spring water. The Quebrada Chug Chug was sampled. It's δ7Li and especially δ11B are high drainage basin of ChugChug in the Cordillera Domeyko includes large for the overall area (+10.3‰ and +22.5‰), while 87Sr/86Sr is more volumes of Jurassic - Cretaceous limestone and its water has δ11Bof radiogenic than contemporary seawater 0.70822, probably due to +22.5‰, well within the range of Mesozoic marine limestone weathering of radiogenic basement rocks of the Precordillera. (Joachimski et al., 2005; Paris et al., 2010). ChugChug Sr is more Geothermal water, sedimentary brine and evaporites are important radiogenic than Mesozoic seawater which can be explained by the in defining groundwater chemistry in this region. Isotopic fractionation weathering of radiogenic silicate rocks in the Cordillera Domeyko. Al- of Li and B of geothermal water compared to the host rock is minimal though ChugChug lies on a mixing line between the headwaters of the due to the temperature dependence of isotope fractionation during Loa and marine rain (Chetelat et al., 2005), weathering of marine secondary mineral formation and rock dissolution being the dominant limestone releasing B and Sr derived from carbonate with a minor process. Analysis of water at El Tatio by various investigators contribution of radiogenic Sr from the basement is a more plausible (Giggenbach, 1978; Cortecci et al., 2005; Tassi et al., 2010) shows its explanation for groundwater in this particularly dry part of the Desert. composition has been consistent over time, and we use the isotope Furthermore, Cl in supergene atacamite in multiple copper deposits compositions determine in this study differences to characterize it. The within the Atacama is sourced from basement marine Jurassic rocks in composition of sedimentary brine which may infiltrate into regional the Domeyko Basin (Arcuri and Brimhall, 2003; Cameron et al., 2007) groundwater is based on the chemical data of the most saline water and not rain. The extent of marine limestone is unknown east of the Precordillera due to the coverage of volcanic rocks. The Paniri spring

39 L.V. Godfrey, et al. Chemical Geology 518 (2019) 32–44

Fig. 5. Weathering causes an increase in silica but minimal change in δ7Li in groundwater. Surface water in diluted by meteoric water causing a decrease in silica with adsorption increasing δ7Li. Groundwater input to the Loa detracts from this pattern, causing Si and δ7Li to increase. Cooling of water is indicated by data for El Tatio and Baños de Turi.

shown). Silica and δ7Li then increase in groundwater as residence time increases (Wanner et al., 2014). The increase in δ7Li is from 0 to 2.5‰ while Si increases from 1 to 2 mmol/kg, reaching the highest values at Chitor, the only sample, apart from El Tatio where water is in equili- brium or saturated with amorphous silica (Table 1). Over a similar increase in Si, but with a granitic mineral assemblage, Wanner et al. (2014) predicted an increase of nearly 10‰ in δ7Li. Pogge von Fig. 4. The role of lithology on boron (a) and lithium (b) isotopes indicated by Strandmann et al. (2014) measured an increase of 7‰ in the sandstone their correlation with 87Sr/86Sr. Ignimbrite δ11B data Schmitt et al. (2002), hosted Great Artesian Basin, much more than our measured increase of lava: Rosner et al. (2003); rain: Chetelat et al. (2005). 2.5‰. Our small increase in δ7Li may be a consequence of the dom- inance of silica-rich rhyolitic glass of the ignimbrites and lack of Al, has around 15‰ higher δ11B relative to its Sr isotope composition limiting the formation secondary minerals (e.g., Declercq et al., 2013), based on the trend described by other samples, and it plots within the B- indeed, dissolution rates of andesitic glass is lower than other volcanic Sr isotope field for Mesozoic limestone. Limestone could be present at aluminosilicates (Rowe and Brantley, 1993; Wolff-Boenisch et al., depth where water might interact with it directly, or decarbonation 2004). We did not measure Al, but as an example, water in the MNT 11 reactions of limestone could release B(OH)3 which is B enriched re- aquifer which runs between the Altiplano and the southeastern end of lative to the carbonate itself (Deegan et al., 2016). Limestone-magma the Salar de Atacama in very similar lithology, has one hundred times interaction supports similar conclusions of marine rocks underlying lower Al than in the Great Artesian Aquifer, but over 40 times more Li parts of the arc based on rare earth element data from the scoria cone of (Pogge von Strandmann et al., 2014; Rissmann et al., 2015). The large El Rojo III adjacent to Paniri (Godoy et al., 2014), and wollastonite amount of Li derived from thermal sources undergoes a minor relative xenolith rare earth element data from Lascar volcano (Matthews et al., reduction in amount, rock continues to release Li and consequently we 7 1996). measure only minimal increases in δ Li. 7 The Silala and V. de Barros stream plot well above the 87Sr/86Sr - When groundwater reaches the surface δ Li can rapidly increase δ7Li array, especially compared to other members of the ignimbrite because the formation of clays dominates over rock dissolution, and at group. While they plot close to ChugChug we do not think their δ7Li the same time Si concentrations decrease through dilution with me- compositions are directly related to limestone lithology, as high δ11Bin teoric water (Wanner et al., 2014). Water from the Salado plots with 7 Paniri was, because Li is not enriched in limestone although its δ7Li is groundwater on the δ Li-Si plot, while water from the Loa show in- 7 high relative to silicates (Chan et al., 2006). There is no evidence from creasing Si and δ Li. The Salado has been recognized as a stream which δ11B that limestone is important in their catchments and a more likely is heavily impacted by water from El Tatio (Romero et al., 2003) and explanation is that secondary minerals have removed 6Li since water displays a reservoir effect. The Río Loa, a gaining stream along this 7 reached the surface. section (Jordan et al., 2015), shows increases in Si and δ Li over 50 km before being joined by the Salado, because of groundwater inputs. Ex- 5.4. The effect of silicate weathering and the formation of clays on δ7Li and change between surface and groundwater also accounts for the high δ7 δ11B Li of the eastern Calama Basin samples where it has been hypothe- sized that water which enters these aquifers to the west of the Chiu Chiu Silica, B and Li are released from primary minerals during weath- monocline originates from the Loa River (Jordan et al., 2015), a theory ering. These elements are more soluble at high versus low tempera- we can also support from the similarity of the Sr isotope composition. tures, and their dissolved concentrations decrease by formation or up- The incorporation of Li and B into secondary minerals, especially take into secondary minerals (Vigier et al., 2008). When thermal waters clays, increases Na/Li, Na/B, and their isotope composition. For Li, “ ” δ7 – 87 86 cool, Si precipitates as sinters and, with Al, as clays which fractionate samples which formed the Loa group in Li Sr/ Sr space form a aqueous δ7Li and δ11B(Pistiner and Henderson, 2003; Vigier et al., roughly linear array which is parallel to a second roughly linear array fi “ ” 2008; Meredith et al., 2013; Pogge von Strandmann et al., 2014). de ned by samples from the Ignimbrite group (Fig. 6A). Using a Precipitation of silica sinters during cooling, represented by El Tatio logarithmic Na/Li axis, mixing and open system isotope fractionation and Baños de Turi, produces little change in δ7Li (Fig. 5)orδ11B (not produce curves, and Rayleigh fractionation produces lines. For

40 L.V. Godfrey, et al. Chemical Geology 518 (2019) 32–44

systems (Vigier et al., 2009). Kaolinite occurs in the eroded and exposed volcanic summits, but little is known how common kaolinite is in the surrounding subsurface, adsorption on the other hand is common-place. The large fractionation effects measured on smectite do not appear to apply here, even though smectite is present. For surface waters, mixing requires an endmember with elevated δ7Li and high Na/Li. One can- didate is sedimentary brine or the very dilute water draining V. Barro. However, the concave-down curve that describes mixing of such water with thermal water does not encompass the surface water compositions we have and mixing with other endmembers is equally unsatisfactory. In contrast, removal of Li from a thermal water source following a Rayleigh model and using the same isotope separation between clay and water of −3.5‰ can explain the data. For most samples, 60 to 95% of the Li must be removed from solution, but these percentages may be reduced if the initial water composition reflects a mixed thermal-cold water with a higher Na/Li ratio. In such a geologically complex area, it is surprising that simple re- moval models quite successfully describe solution Li isotope systema- tics. It is possible that this stems from the high concentration of Li in thermal water which makes weathering reactions at low temperatures somewhat uninfluential. While the only sample which plots near to the mixing line between thermal water and salt is Baños de Turi, several of the water samples in the ignimbrite group plot near a mixing line be- tween thermal water and sedimentary brine. The sedimentary brine accounts for 30% of the mixture. The role silicate weathering and the formation of clays on the B isotope composition is less apparent (Fig. 6B). Mixing between thermal water and sedimentary brine, and between thermal water and salt which could in part explain groundwater Na/Li - δ7Li does not describe measured B. Only a few groundwater samples plot close to open system fractionation with thermal water as the starting composition. Further- Δ11 − Fig. 6. The formation of secondary minerals increases δ7Li relative to Na/Li (A) more, it requires an isotope separation ( Bc-w) on the order of 10 to δ11B relative to B/Na (B). Mixing (dashed) lines describe can describe some of −8‰ which is consistent with adsorption of only tetrahedral B (Palmer the groundwater samples when thermal and sedimentary brine is used. Open et al., 1987). Since pH indicates that both B species are present, and system removal processes describes water with high Na/Li better than mixing that the trigonal form will be more abundant, such limited fractionation between thermal and mixed salts, but for high Na/B the opposite holds. Once is problematic. water reaches the surface rock buffering is minimized relative to adsorption to Unlike Li, there is a decrease in the Na/B ratio of the Loa surface secondary phases, and Rayleigh fractionation processes can describe surface samples compared to thermal and groundwater which further high- water samples. Samples from the Salar de Atacama (Munk et al., 2018) are lights the inconsistency of removal processes in controlling B isotopes. included, and indicate that before the water evaporates to form Li-rich brine, A second difference between Li and B is that thermal water had the the inflow water has strong characteristics of a thermal water mix. lowest δ7Li but for B, there are several samples with lower δ11B, and these plot towards the mixed salt end-member. groundwater, Rayleigh fractionation is not an appropriate model to use, because the reactant, in this case dissolved Li, is constantly replenished 5.5. Evaluating the influence of evaporite mineral dissolution using δ11B by rock dissolution and the Rayleigh model assumes that the reactant's and Cl/Br isotope or chemical composition changes as the reaction proceeds. For surface water, Rayleigh conditions are more likely to be met, unless the In terms of δ11B and Cl/Br, our samples fall into the two groups, fl addition of groundwater to stream ow is large and persistent along the “Loa” and “Ignimbrite”, but the Paniri spring plots with the “Loa” group rivers course. (Fig. 7). The “Loa” group has δ11B above +6.1‰ and Cl/Br below The applicability of mixing, open system and Rayleigh fractionation 1400; the “ignimbrite plateau” group has δ11B below +2.6‰ and Cl/Br models in describing the data have been investigated for Li using the above 2400. Based on the data for mixed salts in the region, dissolving end-members from Table 2 following Henchiri et al. (2014). Open salts add Cl and 10B-rich B to water, and no Br. δ7 δ7 – Δ7 system fractionation follows Liw = Lii (1-f). Lic-w and Rayleigh The samples in the ignimbrite group most affected by salt inputs are δ7 δ7 – Δ7 δ7 fractionation Liw = Lii Lic-w. ln(f), where Liw is the isotope Chitor followed by Linzor (Fig. 7). The location of this salt source is not δ7 composition of the evolving water, Lii its initial isotope composition, obvious since the closest salt lake or salar are Laguna Colorada and Δ c-w the isotope separation between clay and water, and f the fraction of Salar de Chalviri 40–60 km to the east in Bolivia. However, the Cl/Br Li remaining in solution. We assume that Na is released from rock or ratio of El Tatio is greater than subduction related andesite or rhyolite dissolved from mixed salts similarly to Li, but once in solution, it is not (45–64, 225–8451; Aiuppa et al., 2009, though we caution that the removed to clays while Li is. The initial water composition used for the halide data base for rocks is small) or other geothermal systems, in- model is the thermal endmember, El Tatio. While mixing between cluding Apacheta and Torte de Tocorpuri (Tassi et al., 2010). Although thermal water and salt or brine produces a line that is close to many of Giggenbach (1978) argued against dissolution of buried evaporites in the groundwater samples, open system fractionation also gives a good forming the Cl-rich Tatio brine and favored magmatic HCl due to its low fi Δ7 − ‰ t if we assign a value for Lic-w of 3.5 . This value is small for cold pH, its near unity Cl/Na strongly suggests that halite dissolution is in- temperatures if Li is taken into octahedral sites of Mg-smectite (Vigier volved. Leakage of water into which mixed salt assemblages containing et al., 2008) but consistent with either surface adsorption or uptake into borates have dissolved, and characterized by Chitor, can explain the kaolinite (Vigier et al., 2010) and is in the range found for some basaltic complexity of the mixing relationships involving δ11B and Na/B, and

41 L.V. Godfrey, et al. Chemical Geology 518 (2019) 32–44

covered by ignimbrite, but far from pervasive, and barely present in the upper reaches of the Loa River. The area most strongly impacted by brine or evaporites dissolving in the subsurface occurs close to El Tatio and is particularly apparent at Chitor. These salts have 87Sr/86Sr con- sistent with ignimbrite, and Chitor, with low δ7Li indicates that the salts formed from water that acquired its chemical composition from water- rock reactions at above surface temperatures because higher δ7Li would be expected if temperatures had been low. If the salts are from ancient and large salars now buried under ignimbrites, then they formed during or shortly after the eruptions of the Miocene-Pliocene ignimbrite flare- up when the geothermal gradient was high (de Silva, 1989; Kay et al., 2010). A modern example is the lake currently occupying the Cerro Galan caldera in Argentina where δ7Li is indistinguishable from the Cerro Galan ignimbrites (Godfrey et al., 2013). Groundwater flow under the western rim of the Pastos Grandes caldera has been suggested as a source of spring water to the east of the Ascotán salar (Keller and Soto, 1998), and westward dipping ignimbrites erupted from these calderas (such as the Sifon and Puripicar) may provide additional flow paths for groundwater to the west. Drainage through the Chitor spring, and other places, is more than probable. We expect groundwater flowing towards the Salar de Atacama in its northern part is very similar to groundwater in the Upper Loa Basin since they share a long boundary and their geology is similar. Based on water from Chitor and Caspana, the δ7Li composition of the San Pedro and Vilama rivers are probably < 5‰, lower than the +5 to +10‰ values for water entering the southern end of the Salar through the MNT aquifer (Munk et al., 2018). The isotopic and Na/Li ratios of water entering the SE sector of the Salar de Atacama are indistinguishable from the values for sections of the Loa, rather than rivers draining ig- nimbrite or the El Tatio area, or sedimentary brines. East of the Salar de Atacama is la Pacana caldera, the origin of thick and extensive ignim- brites (de Silva, 1989; Lindsay et al., 2001). Within La Pacana's confines Fig. 7. The presence of salts, including borates, with high Cl/Br and low δ11B, are numerous salars (Lindsay et al., 2001), which similar groundwater has a strong effect on the B isotope systematics (A), 2-component mixing in- dicated by the dashed line. These salts originate from the weathering of ig- levels around its southern end suggest are connected by a single shallow nimbrites followed by evaporation from their 87Sr/86Sr composition (B). aquifer system connected through fractures in the volcanic rocks (Herrera et al., 2016). The final outflow of this large groundwater system is unknown since La Pacana's western rim for much of its length the samples most strongly affected are those of the ignimbrite group. acts as a barrier to groundwater flow and retains several lakes which While Caspana plots near the sedimentary brine in terms of δ11B and would otherwise drain to the salar and hence not exist. Na/B it's lower δ11B and δ7Li compared to saline lakes on the Altiplano Rissmann et al. (2015) found that water within of the MNT aquifer (Schmitt et al., 2002; Godfrey et al., 2013; Orberger et al., 2015; Lagos acquires thermal water from Socompa volcano before it eventually et al., 2015) suggest that it is influenced by both evaporite and brine reaches the Salar de Atacama. The water in the MNT aquifer could be sources. compatible with a mixture of thermal and sedimentary brine, though We can elucidate the source of the salts from 87Sr/86Sr (Fig. 7B). El additional samples would be needed to test this. Most of the data pre- Tatio and Linzor have the most radiogenic, ignimbrite-like composi- sented by Munk et al. (2018) are shallow groundwater (or close to tions, but as salt contributions increase Cl/Br ratios, 87Sr/86Sr increases springs), and have δ7Li values suggesting that significant removal of 6Li too suggesting that the salt itself has a Sr isotope composition like that to clays has occurred, probably within the clay-rich margin of the salar. of the ignimbrite. Considering the closed basin morphology of the Al- At the same time, evaporation has allowed halite to precipitate leaving tiplano plateau which is partly due to the presence of multiple calderas, Li in solution and decreasing Na/Li. it is not surprising that these calderas would have hosted saline lakes In conclusion, groundwater draining the volcanic arc in the and salt plans following their eruption and that the lakes would acquire Antofagasta region of northern Chile has a composition which is the Sr isotope signature of the surrounding ignimbrites. The samples strongly buffered by rock dissolution because very low rainfall does not from the Loa group are least affected by salt dissolution, though their produce enough flow through shallow aquifers or rivers to overprint its proximity to Ascotan makes it possible that windblown material from signature. A common component of groundwater is hydrothermal this borate-rich salar washes into the river. Boron precipitates earlier water. To account for high Cl and Na, dissolution of mixed salts, in- during evaporation than Li, so Li will be largely unaffected by this cluding borates, is required, but isotopes indicate even these salts process, but we cannot exclude the possibility that the rivers δ11B formed from evaporation of water derived from weathering at tem- might have been lowered. However, we think wind-blown salts do not peratures higher than air temperatures. Explosive volcanism along this contribute as much material to the river as rock sources. segment of the arc left a series of calderas in which lakes and salars formed. Groundwater with particularly high Cl and Na drains from 6. A regional, isotope-driven model them into regional groundwater flow and can be distinguished from other water when their δ7Li and δ11B compositions are considered. Multi-isotope analyses are used to highlight the role of lithology, Given the similarity of the geology and volcanic landscape of the high and low temperature weathering, inputs of sedimentary brine and western side of the Andes, this hydrologic model is applicable to water associated evaporite minerals. The impact of sedimentary brine and entering other pre-Andean basins located next to geologically recently evaporite dissolution in the Upper Loa Basin is important in the area active calderas. Those basins high Li and B content also highlight the

42 L.V. Godfrey, et al. Chemical Geology 518 (2019) 32–44 role of high temperature silicate weathering in forming Li and B rich chain, Central Andes, Northern Chile. J. S. Am. Earth Sci. 52, 24–42. water. Grice, J.D., Burns, P.C., Hawthorne, F.C., 1999. Borate minerals. II. A hierarchy of structures based upon the borate fundamental building block. Can. Mineral. 37 (731- 716). Acknowledgements Henchiri, S., Clergue, C., Dellinger, M., Gaillardet, J., Louvat, P., Bouchez, J., 2014. The influence of hydrothermal activity on the Li isotopic signature of rivers draining volcanic areas. Procedia Earth Planet. Sci. 10, 223–230. We would like to acknowledge useful discussions with Dr. T.E. Herrera, C., Custodio, E., Chong, G., Lambán, L.J., Riquelme, R., Wilke, H., Jódar, J., Jordan and other members of the Geology Department at UNC in Urrutia, J., Urqueta, H., Sarmiento, A., Gamboa, C., Lictevou, E., 2016. Groundwater Antofagasta. The reviews of two anonymous reviewers greatly im- flow in a closed basin with a saline shallow lake in a volcanic area: Laguna Tuyajto, – proved this contribution. This research was supported by National northern Chilean Altiplano of the Andes. Sci. Total Environ. 541, 303 318. Holser, W.T., 1966. Bromide geochemistry of salt rocks. In: Rau, J.L. (Ed.), Second Geographic Society award #800006, NSF OISE-1037929 and EAR- Symposium on Salt, Northern Ohio Geol. Soc. 2. pp. 248–275. 1117496, the Anillo Project ACT1203 of the CONICYT of Chile and by a Holser, W.T., 1979. Trace elements and isotopes in evaporites. In: Burns, R.G. (Ed.), Ž Ph.D. grant from CONICYT-PCHA/Doctorado Nacional, number 2016- Marine Minerals, Review in Mineralogy, . vol. 6. Mineralogical Society of America, pp. 295–346. 21160152 for C. Gamboa. Houston, J., 2007. Recharge to groundwater in the Turi Basin, northern Chile: an eva- luation based on tritium and chloride mass balance techniques. J. Hydrol. 334, – References 534 544. https://doi.org/10.1016/j.jhydrol.2006.10.030. Huh, Y., Chan, L.-H., Zhang, L., Edmond, J.M., 1998. Lithium and its isotopes in major world rivers: implications for weathering and the oceanic budget. Geochim. Aiuppa, A., Baker, D.R., Webster, J.J., 2009. Halides in volcanic systems. Chem. Geol. Cosmochim. Acta 62, 2039–2051. 263, 1–18. Ishikawa, T., Nakamura, E., 1993. Boron isotope systematics of marine sediments. Earth Aravena, R., Suzuki, O., 1990. Isotopic evolution of river water in Northern Chile Region. Planet. Sci. Lett. 117, 567–580. Water Resour. Res. 26, 2887–2895. Joachimski, M.M., Simon, L., van Geldern, R., Lécuyer, C., 2005. Boron isotope geo- Arcuri, T., Brimhall, G., 2003. The chloride source for atacamite mineralization at the chemistry of Paleozoic brachiopod calcite: implications for a secular change in the Radomiro Tomic porphyry copper deposit, northern Chile. Econ. Geol. 98, boron isotope geochemistry of seawater over the Phanerozoic. Geochim. Cosmochim. 1667–1681. Acta 69, 4035–4044. Bahlburg, H., Bretkreuz, C., 1991. Paleozoic evolution of active margin basins in the Jordan, T.E., Muñoz, N., Hein, M., Lowenstein, T., Godfrey, L., Yu, J., 2002. Active southern Central Andes (northwestern Argentina and northern Chile). J. S. Am. Earth faulting and folding without topographic expression in an evaporite basin, Chile. GSA Sci. 4, 171–188. Bull. 114, 1406–1421. Bethke, C.M., 2011. The Geochemist's Workbench. University of Illinois. Jordan, T.E., Mpodozis, C., Muñoz, N., Blanco, N., Pananont, P., Gardeweg, M., 2007. Blanco, N., Tomlinson, A., 2009. Carta Chiu Chiu, Región de Antofagasta: Servicio Cenozoic subsurface stratigraphy and structure of the Salar de Atacama Basin, Nacional de Geología y Minería. In: Carta Geológica de Chile no. 117, (54 pp., scale northern Chile. J. S. Am. Earth Sci. 23, 122–146. https://doi.org/10.1016/j.jsames. 1:50.000). 2006. 09.024. Boschetti, T., Cortecci, G., Barbieri, M., Mussi, M., 2007. New and past geochemical data Jordan, T., Herrera, L., Kirk-Lawlor, C., Godfrey, L., 2015. Architecture of the aquifers of on fresh to brine waters of the Salar de Atacama and Andean Altiplano, northern the Calama Basin, Loa catchment basin, northern Chile. Geosphere 11, 1438–1474. Chile. Geofluids 7, 33–50. Kay, S.M., Coira, B.L., Café, P.J., Chen, C.-H., 2010. Regional chemical diversity, crustal Cameron, E.M., Leybourne, M.I., Palacios, C., 2007. Atacamite in the oxide zone of copper and mantle sources and evolution of central Andean Puna plateau ignimbrites. J. deposits in northern Chile: involvement of deep formation waters? Mineral. Deposita Volcanol. Geotherm. Res. 81–111. 42, 205–218. Keller, B., Soto, D., 1998. Hydrogeologic influences on the preservation of Orestias Chan, L.-H., Leeman, W.P., Plank, T., 2006. Lithium isotopic composition of marine se- Ascotánensis (Teleostei: Cyprinodontidae), in Salar de Ascotán, northern Chile. Rev. diments. Geochem. Geophys. Geosyst. 7, Q06005. https://doi.org/10.1029/ Chil. Hist. Nat. 71, 147–156. 2005GC001202. Kuhn, D., 2002. Fold and thrust belt structures and strike-slip faulting at the SE margin of Chetelat, B., Gaillardet, J., Freydier, R., Négrel, P., 2005. Boron isotopes in precipitation: the Salar de Atacama basin, Chilean Andes. Tectonics 21, 1026. https://doi.org/10. experimental constraints and field evidence from French Guiana. Earth Planet. Sci. 1029/2001TC901042. Lett. 235, 16–30. Lagos, L.V., 2017. Hidrogeoquímica de fuentes termales en ambientes salinos re- Chong, G., Pueyo, J.J., Demargasso, C., 2000. Los yacimientos de boratos de Chile. Rev. lacionados con salares en los Andes del norte de Chile (Master Thesis). University of Geol. Chile 27, 99–119. Chile (285 pp.). Cortecci, G., Boschetti, T., Mussi, M., Herrera Lameli, C., Mucchino, C., Barbieri, M., Lagos, L.V., Godfrey, L., Reich, M., Achurra, L., Morata, D., Aguirre, I., Mayorga, C., 2015. 2005. New chemical and original isotopic data on waters from El Tatio geothermal Disentangling the impact of evaporitic deposits and saline lakes in thermal waters field, northern Chile. Geochem. J. 39, 547–571. from the Chilean Altiplano using δ7Li, δ11B, 87Sr/86Sr, and trace element geochem- de Silva, S.L., 1989. Altiplano-Puna volcanic complex of the central Andes. Geology 17, istry. In: Goldschmidt Abstract. 1102–1106. Lemarchand, D., Gaillardet, J., Lewin, E., Allegre, C.J., 2002. Boron isotope systematics in de Silva, S.L., Zandt, G., Trumbull, R., Viramonte, J., 2006. Large scale silicic volcanism e large rivers: implications for the marine boron budget and paleo-pH reconstruction the result of thermal maturation of the crust. In: Chen, Y.-T. (Ed.), Advances in over the Cenozoic. Chem. Geol. 190, 123–140. Geosciences. World Scientific Press, pp. 215–230. Lindsay, J.M., de Silva, S., Trumbull, R., Emmermann, R., Wemmer, K., 2001. La Pacana Declercq, J., Diedrich, T., Perrot, M., Gislason, S.R., Oelkers, E.H., 2013. Experimental caldera, N. Chile: a re-evaluation of the stratigraphy and volcanology of one of the determination of rhyolitic glass dissolution rates at 40–200 °C and 2 < pH < 10.1. world's largest resurgent calderas. J. Volcanol. Geotherm. Res. 106, 145–173. Geochim. Cosmochim. Acta 100, 251–263. Lucassen, F., Franz, G., Thirlwall, M.F., Mezger, K., 1999. Crustal recycling of meta- Deegan, F., Troll, V.R., Whitehouse, M.J., Jolis, E.M., Freda, C., 2016. Boron isotope morphic basement: Late Paleozoic granites of the Chilean Coast Range and fractionation in magma via crustal carbonate dissolution. Nat. Sci. Rep. https://doi. Precordillera at, 228S. J. Petrol. 40, 1527–1551. org/10.1038/srep30774. Magna, T., Wiechert, U., Grove, T.L., Halliday, A.N., 2006. Lithium isotope fractionation Dingman, R., 1967. Geology and ground-water resources of the northern part of the Salar in the southern Cascadia subduction zone. Earth Planet. Sci. Lett. 250, 428–443. de Atacama, Antofagasta Province, Chile. Geol. Den. Surv. Bull. 1219, 49. Marinović, N., Lahsen, A., 1984. Carta Geológica de Chile,1:250,000, Hoja Calama. Ellis, A.J., Mahon, W.A.J., 1964. Natural hydrothermal systems and experimental hot- Servico Nacional de Geología y Minería, Santiago, Chile. water/rock interactions. Geochim. Cosmochim. Acta 28, 1323–1357. Marquillas, R., Sabino, I., Sial, A.N., del Papa, C., Ferreira, V., Matthews, S., 2007. Carbon Ericksen, G.E., 1981. Geology and Origin of the. Chilean Nitrate Deposits. Professional and oxygen isotopes of Maastrichtian–Danian shallow marine carbonates: Yacoraite Paper 1188. vol. 37 USGS. Formation, northwestern Argentina. J. S. Am. Earth Sci. 23, 304–320. Evanstar, L.A., Hartley, A.J., Archert, S.G., Neilson, J.E., 2015. Climatic and halokinetic Marriott, C.S., Henderson, G.M., Crompton, R., Staubwasserm, M., Shaw, S., 2004. Effect controls on alluvial-lacustrine sedimentation during compressional deformation, of mineralogy, salinity, and temperature on Li/Cl and Li isotope composition of Andean forearc, northern Chile. Basin Res. 1–24. https://doi.org/10.1111/bre12124. calcium carbonate. Chem. Geol. 212, 5–15. Fouillac, C., Michard, G., 1981. Sodium/lithium ratio in water applied to geothermometry Matthews, S.J., Marquillas, R.A., Kemp, A.J., Grange, F.K., Gardeweg, M.C., 1996. Active of geothermal reservoirs. Geothermics 10, 55–70. skarn formation beneath Lascar Volcano, northern Chile: a petrographic and geo- Giggenbach, W.F., 1978. The isotopic composition of waters from the El Tatio geothermal chemical study of xenolith in eruption products. J. Metamorph. Geol. 14, 509–530. field, Northern Chile. Geochim. Cosmochim. Acta 42, 979–988. May, G., Hartley, A.J., Stuart, F.M., Chong, G., 1999. Tectonic signatures in arid con- Giggenbach, W.F., 1988. Geothermal solute equilibria, derivation of Na-K-Mg-Ca geoin- tinental basins: an example from the upper Miocene–Pleistocene, Calama basin, dicators. Geochim. Cosmochim. Acta 52, 2749–2765. Andean forearc, northern Chile. Palaeogeogr. Palaeoclimatol. Palaeoecol. 151, Giggenbach, W.F., Soto, R.C., 1992. Isotopic and chemical composition of water and 55–77. https://doi.org/10.1016/S0031-0182(99)00016-4. steam discharges from volcanic-magmatic-hydrothermal systems of the Guanacaste Maza, S.N., Collo, G., Morata, D., Lizana, C., 2018. Clay mineral associations in the clay Geothermal Province, Costa Rica. Appl. Geochem. 7, 309–332. cap from the Cerro Pabellón blind geothermal system, Andean Cordillera, Northern Godfrey, L.V., Chan, L.H., Alonso, R.N., Lowenstein, T.K., McDonough, W.F., Houston, J., Chile. Clay Miner. 53, 117–141. Li, J., Bobst, A., Jordan, T.E., 2013. The role of climate in the accumulation of li- McCaffrey, M.A., Lazar, B., Holland, H.D., 1987. The evaporation path of seawater and − thium-rich brine in the central Andes. Appl. Geochem. 38, 92–102. the coprecipitation of Br and K+ with halite. J. Sediment. Petrol. 57, 928–938. Godoy, B., Wörner, G., Kojima, S., Aguilera, F., Simon, K., Hartmann, G., 2014. Low- Meredith, K., Moriguti, T., TOmascak, P., Hollins, S., Nakamura, E., 2013. The lithium, pressure evolution of arc magmas in thickened crust: the San Pedro-Linzor volcanic boron and strontium isotopic systematics of groundwaters from an arid aquifer

43 L.V. Godfrey, et al. Chemical Geology 518 (2019) 32–44

system: implications for recharge and weathering processes. Geochim. Cosmochim. Rio Loa (Second Region, Chile). Appl. Geochem. 18, 1399–1416. Acta 112, 20–31. Rosner, M., Erzinger, J., Franz, G., Trumbull, R., 2003. Slab-derived boron isotope sig- Millot, R., Négrel, Ph., 2007. Multi-isotopic tracing (δ7Li, δ11B, 87Sr/86Sr) and chemical natures in arc volcanic rocks from the Central Andes and evidence for boron isotope geothermometry: evidence from hydro-geothermal systems in France. Chem. Geol. fractionation during progressive slab dehydration. Geochem. Geophys. Geosyst. 4, 244, 664–678. 1–25. Millot, R., Négrel, Ph., Petelet-Giraud, E., 2007. Multi-isotopic (Li, B, Sr, Nd) approach for Rowe, G.L., Brantley, S.L., 1993. Estimation of the dissolution rates of andesitic glass, geothermal reservoir characterization in the Limagne Basin (Massif Central, France). plagioclase and pyroxene in a flank aquifer of Poás Volcano, Costa Rica. Chem. Geol. Appl. Geochem. 22, 2307–2325. 105, 71–87. Mpodozis, C., Arriagada, C., Basso, M., Roperch, P., Cobbold, P., Reich, M., 2005. Late Schmitt, A.K., Kasemann, S., Meixner, A., Rhede, D., 2002. Boron in central Andean ig- Mesozoic to Paleogene stratigraphy of the Salar de Atacama Basin, Antofagasta, nimbrites: implications for crustal boron cycles in an active continental margin. northern Chile: implications for the tectonic evolution of the central Andes. Chem. Geol. 183, 333–347. Tectonophysics 399, 125–154. https://doi.org/10.1016/j. tecto.2004.12.019. Smith, C.L., Ficklin, W.H., Thompson, J.M., 1987. Concentrations of arsenic, antimony, Munk, L.A., Boutt, D.F., Hyneck, S., Moran, B., 2018. Hydrogeochemical fluxes and and boron in steam and steam condensate at the Geysers, California. J. Volcanol. processes contributing to the formation of lithium-enriched brines in a hyper-arid Geotherm. Res. 32, 329–341. continental basin. Chem. Geol. 493, 37–57. Sun, T., Bao, H., Reich, M., Hemming, S.R., 2018. More than ten million years of hyper- Muñoz, N., Charrier, R., Jordan, T., 2002. Interactions between basement and cover aridity recorded in the Atacama Gravels. Geochim. Cosmochim. Acta 227, 123–132. during the evolution of the Salar de Atacama Basin, northern Chile. Rev. Geol. Chile Tassi, F., Aguilera, F., Darrah, T., Vaselli, O., Capaccioni, B., Poreda, R.J., Delgado 29, 55–80. https://doi.org/10.4067/S0716- 02082002000100004. Huertas, A., 2010. Fluid geochemistry of hydrothermal systems in the Oi, T., Nomura, M., Musashi, M., Ossaka, T., Okamoto, M., Kakihana, H., 1989. Boron Arica–Parinacota, Tarapacá and Antofagasta regions (northern Chile). J. Volcanol. isotopic composition of some boron minerals. Geochim. Cosmochim. Acta 53, Geotherm. Res. 192, 1–15. 3189–3195. Tomlinson, A.J., Blanco, P.N., Dilles, J.H., 2010. Carta Calama, Región de Antofagasta: Orberger, B., Rojas, W., Millot, R., Flehoc, C., 2015. Stable isotopes (Li, O, H) combined Carta Geológica de Chile. Subdirección Nacional de Geología, Santiago, Chile with brine chemistry: powerful tracers for Li origins in Salar deposits from the Puna (SERNAGEOMIN Serie Preliminar no. 8, scale 1:50,000). region, Argentina. Procedia Earth Planet. Sci. 13, 307–311. Veizer, J., Ala, D., Azmy, K., Bruckschen, P., Buhl, D., Bruhn, F., Carden, G.A.F., Diener, Palmer, M.R., Helvaci, C., 1997. The boron isotope geochemistry of the Neogene borate A., Ebneth, S., Goddéris, Y., Jasper, T., Korte, C., Pawellek, F., Podlaha, O., Strauss, deposits of Western Turkey. Geochim. Cosmochim. Acta 61, 3161–3169. H., 1999. 87Sr/86Sr, δ13C and δ18O evolution of Phanerozoic seawater. Chem. Geol. Palmer, M.R., Spivack, A., Edmond, J.M., 1987. Temperature and pH controls over iso- 161, 59–88. topic fractionation during adsorption of boron on marine clay. Geochim. Cosmochim. Vengosh, A., Starinsky, A., Kolodny, Y., Chivas, A.R., Raab, M., 1992. Boron isotope Acta 51, 2319–2323. variations during fractional evaporation of seawater; new constraints on the marine Paris, G., Bartolini, A., Donnadieu, Y., Beaumont, V., Gaillardet, J., 2010. Investigating vs. nonmarine debate. Geology 20, 799–802. boron isotopes in a middle Jurassic micritic sequence: primary vs. diagenetic signal. Vengosh, A., Chivas, A.R., Starinsky, A., Kolodny, Y., Zhang, B.Z., Zhang, P.X., 1995. Chem. Geol. 275, 117–126. Chemical and boron isotope compositions of non-marine brines from the Qaidam Penniston-Dorland, S., 2017. Lithium isotope geochemistry. Rev. Mineral. Geochem. 82, Basin, Qinghai, China. Chem. Geol. 120, 135–154. 165–217 (Mineralogical Society of America). Vigier, N., Decarreau, A., Millot, R., Carignan, J., Petit, S., France-Lanord, C., 2008. Pistiner, J.S., Henderson, G.M., 2003. Lithium-isotope fractionation during continental Quantifying Li isotope fractionation during smectite formation and implications for weathering processes. Earth Planet. Sci. Lett. 214, 327–339. the Li cycle. Geochim. Cosmochim. Acta 72, 780−792. Pogge von Strandmann, P.A.E., Porcelli, D., James, R.H., van Calsteren, P., Schaefer, B., Vigier, N., Gislason, S.R., Burton, K.W., Millot, R., Mokadem, F., 2009. The relationship Cartwright, I., Reynolds, B.C., Burton, K.W., 2014. Chemical weathering processes in between riverine lithium isotope composition and silicate weathering rates in the Great Artesian Basin: evidence from lithium and silicon isotopes. Earth Planet. Iceland. Earth Planet. Sci. Lett. 287, 434–441. Sci. Lett. 406, 24−36. Vigier, N., Decarreau, A., Millot, R., Petit, S., 2010. Quantifying lithium isotope fractio- Rech, J.A., Currie, B.S., Michalski, G., Cowan, A.M., 2006. Neogene climate change and nation during clay formation at low temperatures. Geophys. Res. Abstr. 12 (EGU uplift in the Atacama Desert, Chile. Geology 34, 761–764. 11160). Risacher, F., Fritz, B., 2009. Origin of salts and brine evolution of Bolivian and Chilean Walker, B.A., 2011. The Geochemical Evolution of the Aucanquilcha Volcanic Cluster: salars. Aquat. Geochem. 15, 123–157. Prolonged Magmatism and its Crustal Consequences. Ph.D. thesis, 217 pp. Oregon Risacher, F., Alonso, H., Salazar, C., 1999. Geoquímica de aguas en cuencas cerradas, I, II, State University, Eugene, OR. III Regiones, Chile. Ministerio de Obras Públicas, Dirección General de Aguas. http:// Wang, B.S., You, C.F., Huang, K.F., Wu, S.F., Aggarwal, S.K., Chung, C.H., Lin, P.Y., 2010. www.chile.ird.fr/castellano/programmes/annexes/sp_geoquimica_salars.htm. Direct separation of boron from Na- and Ca-rich matrices by sublimation for stable http://horizon.documentation.ird.fr/exl-doc/pleins_textes/pleins_textes_7/divers2/ isotope measurement by MC-ICP-MS. Talanta 82, 1378–1384. 010019475.pdf (Technical Report S.I.T. n 51, Santiago, Chile). Wanner, C., Sonnenthal, E.L., Liu, X.M., 2014. Seawater δ7Li: a direct proxy for global Risacher, F., Alonso, H., Salazar, C., 2003. The origin of brines and salts in Chilean salars: CO2 consumption by continental silicate weathering? Chem. Geol. 381, 154−167. a hydrochemical review. Earth Sci. Rev. 63, 249–293. Ward, K.M., Zandt, G., Beck, S.L., Christensen, D.H., 2014. Seismic imaging of the mag- Risacher, F., Fritz, H., Alonso, H., 2006. Non-conservative behavior of bromide in surface matic underpinnings beneath the Altiplano-Puna volcanic complex from the joint waters and brines of Central Andes: a release into the atmosphere? Geochim. inversion of surface wave dispersion and receiver functions. Earth Planet. Sci. Lett. Cosmochim. Acta 70, 2143–2152. 404, 43–53. Risacher, F., Fritz, B., Hauser, A., 2011. Origin of components in Chilean thermal waters. Wolff-Boenisch, D., Gislason, S.R., Oelkers, E.H., Putnis, C.V., 2004. The dissolution rates J. S. Am. Earth Sci. 31, 153–170. of natural glasses as a function of their composition at pH 4 and 10.6, and tem- Rissmann, C., Leybourne, M., Benn, C., Christenson, B., 2015. The origin of solutes within peratures from 25 to 74 °C. Geochim. Cosmochim. Acta 68, 4843–4858. the groundwaters of a high Andean aquifer. Chem. Geol. 396, 164–181. Xiao, J., Xiao, Y.-K., Jin, Z.-D., He, M.-Y., Liu, C.-Q., 2013. Boron isotope variations and Romero, L., Alonso, H., Campano, P., Fanfani, L., Cidu, R., Dadea, C., Keegan, T., its geochemical application in nature. Aust. J. Earth Sci. 60, 431–447. Thornton, I., Farago, M., 2003. Arsenic enrichment in waters and sediments of the

44