1 Introduction

The Aqua de Ney spring is located in the Shasta Valley, between the Eastern Klamath Mountains and the southern end of the Cascade Range (Siskiyou County, Northern California). The spring is brackish water (TDScalc. ≈ 33 mg/l), with a Na-Alk.(Cl) composition and various extreme characteristics, e.g., pHaverage = 12.0 ± 0.3,Footnote 1 a significant non-carbonate alkalinity (Alk.non-carbonate > Alk.carbonate), B = 250 mg/l, Li = 4 mg/l, SiO2 = 4200 mg/l, and outflows from ultramafic and dacitic rocks (Barnes et al. 1972; Berkstresser 1968; Cullen 2013; Feth et al. 1961; García-Ruiz et al. 2017; Mariner et al. 2003, 2006; White et al. 1963) (Supplementary File S1). Despite the increasing interest in this spring due to the prokaryotic diversity in hyperalkaline fluids (Souza et al. 1974), the presence of cyclic C–S organic compounds (McCollom et al. 2015) and the mimicking of primordial biomorphs (García-Ruiz et al. 2017), its origin is still unclear. The salinity, the relatively high concentration of Na and Cl, the high deuterium value of the water molecule, and the high boron concentration led the first researchers to hypothesize a connection with a connate seawater entrapped in the sediments (Feth et al. 1961). One decade later, new water isotope data were interpreted as “non-meteoric, probably metamorphic origin with analogies with oilfield brine” (Barnes et al. 1972). More recently, the chemical and δ37Cl isotope composition were believed to be controlled by the strong interaction with an underlying serpentinized body, a process that is different than that affecting the thermal springs in the Cascadia Range (Cullen 2013). Finally, the local ultramafic lithology, the very high pH, the bubbling gas composition (CH4 > 80%; Mariner et al. 2003) and the serpentine supersaturation would suggest that serpentinization was the main process that determined the fluid composition (Barnes et al. 1972; Boschetti et al. 2013b; Liakhovitch et al. 2005; Mottl 2009). However, the high boron and silica concentration of the water are unusual for this kind of water–rock process, unless to invoke other sources (Boschetti et al. 2013a; Boschetti and Toscani 2008). In this work, we attempted to better pinpoint the origin of the water using a new analysis of water isotopes, as well as lithium and boron isotopes, which would allow a better distinction between serpentinization process and other potential sources (Boschetti et al. 2013a; Vils et al. 2009). The previous chemical analyses listed above were also taken into account for elemental ratios/diagrams and activity calculations.

2 Methods

The hydrogen (2H/1H) and oxygen (18O/16O) isotope ratios of the water molecule were determined in the distilled sample to avoid the salt effect (Boschetti et al. 2011).

Before distillation, the Aqua de Ney water sample was treated with Zn-acetate to remove the sulfide anions (~ 500 mg/l of total reduced sulfur as H2S), followed by 2 min in a vortexer and an ultrasonic bath to promote the reaction and stripping of the gases from the water (Lu 2016). After a 0.45-μm filtration to remove the precipitated ZnS, the water sample was distilled in a vacuum line and condensed in a liquid nitrogen, cryogenic trap (Boschetti et al. 2011). Finally, the condensed water sample was transferred to an automatic H2(g)–H2O/CO2(g)–H2O equilibration device on line with a dual inlet mass spectrometer (Delta Plus Finnigan; University of Parma) (Boschetti et al. 2011). Lithium (7Li/6Li) and boron (11B/10B) isotope ratios were obtained by MC-ICP-MS measurement (Neptune, Thermo Scientific) at ALS Scandinavia AB and University of Calgary, respectively. Analyses were performed on 0.45-μm filtered samples and after column pretreatment to avoid an analytical background and taking into account the concentration of the analytes (Boschetti et al. 2013a; Tomascak et al. 2016). Delta values in permil (δ‰) of the isotope ratios were calculated against V-SMOW (δ18O, δ2H), LSVEC NIST 8545 RM (δ7Li) and NIST SRM 951 (δ11B). The relative uncertainty on analyses, expressed as standard deviation on replicates (1σ), was ± 0.3‰ on δ7Li and ± 0.4‰ on δ11B. For water isotopes, it will be discussed and compared with previous results in the next session.

The activity of the dissolved species and mineral equilibria at different T–P conditions were calculated using different thermodynamic datasets and program codes (Boschetti et al. 2016). PHREEQCI (Parkhurst and Appelo 2013) was used to estimate the logPCO2 and minerals saturation indexes value at P = 1 bar (Supplementary File S1), whereas EQ3/6 (Wolery and Jarek 2003) to perform calculation at higher P–T values. DBCreate (Kong et al. 2013) and The Geochemist’s Workbench® (Bethke and Yeakel 2008) were used to draw mineral equilibria in the activity diagrams.

3 Results and Discussion

The results obtained in this study are shown in Supplementary File S1 along with previous published chemical and isotopic data and thermodynamic calculation at sampling conditions.

3.1 Isotope Ratios: Serpentinization Versus Slab Dehydration

Previous published results concerning the stable isotope composition of water molecule show contrasting values, particularly on oxygen ratio (Supplementary File S1). It is difficult to understand the reason of these differences because the authors give no explanation about the preparation method adopted to overcome the analytic effect due to dissolved salts and gases. Despite this, our results δ18O(H2O) = + 5.29 ± 0.06‰ and δ2H(H2O) = − 15.4 ± 0.2‰ (mean ± std. dev. of three replicates), are very similar to the values of + 6.02 and − 15.6‰ obtained by Barnes et al. (1972). Therefore, hereafter we will refer and plot only these latter values and our results. Figure 1a shows how the enriched water isotope values of Aqua de Ney are quite different from other local (e.g., Stewart Spring; Mariner et al. 2006) and abroad hyperalkaline waters from ophiolites. The latter are generally Ca-OH, fresh and of meteoric origin falling close to the Global Meteoric Water Line (GMWL; Fig. 1b) (Meyer-Dombard et al. 2015; Monnin et al. 2014; Morrill et al. 2013; Sánchez-Murillo et al. 2014), whereas Aqua de Ney is Na-Alk(Cl) and brackish. In particular, the comparison of local thermal/non-thermal groundwater of meteoric origin (Cullen 2013; Mariner et al. 1998, 2006; Nathenson et al. 2003) with hyperalkaline water influenced by connate fluids (Complexion Spring) shows the peculiar water isotope composition of Aqua de Ney (Fig. 1b). Its δ2H value is comparable with that of Boswell Spring (Fig. 1a), a supposed slab-derived fluid from Eocene sediments located above the serpentinized body of the Cascadia Subduction Zone (CSZ) in the forearc of the Central Oregon Cascade Range (Hurwitz et al. 2005). Figure 1a also reveals how both the springs have deuterium values falling between the hydrothermal vents from Juan De Fuca–Gorda Ridges (Böhlke and Shanks III 1994; Shanks III 2001) and the theoretical composition of the fluids from a dehydrating slab (Martin et al. 2011; Walowski et al. 2015).

Fig. 1
figure 1

δ2H(H2O) versus δ18O(H2O) diagrams. a comparison of Aqua de Ney Spring (N-diamond, this study; Barnes et al. 1972) with Boswell Spring (square, Cullen 2013; Hurwitz et al. 2005), hydrothermal vents (+) from Juan de Fuca–Gorda Ridges (Böhlke and Shanks III 1994; Shanks III 2001; Von Damm et al. 2005) and water in equilibrium with local seafloor hydrothermal smectite (Decitre et al. 2004). Dotted paths represent the evolution of seawater and Boswell Spring (a supposed Eocene fossil water; Hurwitz et al. 2005) after serpentinization at decreasing water/rock ratios (numbers; Supplementary File S2). b expanded view of the “A” diagram, showing the comparison of Aqua de Ney Spring with: hyperalkaline waters from Complexion Spring, TC-2 (Trinity County) and Stewart Spring (Barnes et al. 1972; Goff et al. 2001; Mariner et al. 2006), fluid in equilibrium with oceanic serpentinites and crust (Alt and Shanks III 2006), the theoretical composition of the water expelled by the dehydration of the Cascadian slab with depth in km (Martin et al. 2011; McCrory et al. 2012; Walowski et al. 2015), and local waters of meteoric origin (Cullen 2013; Mariner et al. 1998, 2006; Nathenson et al. 2003). In both, GMWL = Global Meteoric Water Line (Gourcy et al. 2007)

Full size image

The Aqua de Ney showed serpentine supersaturation at sampling conditions (Supplementary File S1). However, its enriched oxygen isotope composition is also consistent with fluids from a serpentinite wedge (Fig. 1b) or a serpentinization process at T > 200 °C (Sturchio et al. 1989). Both slab dehydration and serpentinization by non-meteoric water processes were invoked to explain the deuterium and oxygen isotope composition of local serpentine minerals (Liakhovitch et al. 2005; Peacock 1987). Indeed, the water–rock isotope modeling of the serpentinization process from Boswell spring, supposed as initial fluid, reveals a path toward Aqua de Ney at lower W/R ratios (Fig. 1a; Supplementary File S2). However, according to Vils et al. (2009), at low W/R ratios the fluid contribution from dehydrating slab could became important in these geodynamic settings (Fig. 1a).

The measured value δ11B = + 0.5 ± 0.4‰ and the mean historical B/Cl = 0.11 ± 0.01, Br/Cl = 7.3 × 10−4 ± 0.3 × 10−4 molar ratios suggest a main “hydrothermal” origin for Aqua de Ney (Fig. 2a, b). In particular, its position near to MORB field and local rocks in those diagrams suggests a water–rock interaction occurred at high temperature. This would exclude the involvement of seawater or other fluids from sedimentary basin; as alternative explanation, the mixing with dehydrated fluid and the high-temperature water–rock interaction deeply modify the early water composition.

Fig. 2
figure 2

δ11B versus B/Cl (a) and B/Cl versus Br/Cl (b). In (a), Aqua de Ney Spring (N-diamond, this study for δ11B, mean historical data for B/Cl: see text for complete references) is compared with the hydrothermal vents from Juan de Fuca–Gorda Ridges (Campbell et al. 1994; James et al. 2003) and Complexion Spring (Peters 1993). The paths and open fields from Vengosh (2014) and Boschetti et al. (2015), melt inclusions from Mt. Shasta basaltic andesites (Le Voyer et al. 2010), AOC = altered oceanic crust (Le Voyer et al. 2010; Sano et al. 2008) and S = sediments (Boschetti et al. 2015; Campbell et al. 1994; Leeman et al. 2004) are also shown for comparison. In (b), open fields and paths from Vengosh (2014); local thermal/non-thermal groundwaters (Cullen et al. 2015) and hydrothermal vents are also shown (Butterfield et al. 1994; Campbell et al. 1994; Fouquet et al. 1998; Gieskes et al. 2000; Von Damm et al. 2005) are also shown. In both, MORB (Le Roux et al. 2003; White 2015; Awaleh et al. 2017) and rocks from Cascadia Range (X, OIB in bold with error bars: Cullen 2013; Leeman et al. 2004) are shown

Full size image

The serpentinization model produced a δ11B–δ7Li path that not intercepted the Aqua de Ney sample point (Fig. 3a). In particular, the δ11B values obtained by the model are very high, even if taking into account the lowest published Eocene seawater composition, δ11BEocene_SW = + 32‰ (Paris et al. 2010) instead of δ11BPresent-Day_SW = + 39.6‰ (White 2015), and fractionation factors, Δ11Bwater–serpentine = + 1.57‰ (Hansen et al. 2017) instead of Δ11Bwater–silicates = + 17.5‰ (Wunder et al. 2005) (Fig. 3a; Supplementary File S2). Therefore, a supplementary source, or fractionation effect, is necessary to explain the isotope composition of Aqua de Ney.

Fig. 3
figure 3

δ11B versus δ7Li diagrams. In (a), dotted paths depict the evolution of the present-day and Eocene seawater compositions obtained from the serpentinization models at decreasing water/rock ratios (numbers; Supplementary File S2). In (b), the values detected in the water from Aqua de Ney Spring (N-diamond) are compared with the Juan de Fuca–Gorda Ridges hydrothermal vents (+) (James et al. 2003). The altered oceanic crust (AOC), mantle (M) and sediments (SED) melts and Cascadian slab fluid (SF) open fields are from Leeman et al. (2004). The values of the residual olivine and pyroxenes after serpentinization of the Trinity peridotite are from Lundstrom et al. (2005). “Low-T” slab fluid model represents the δ7Li mean ± error bars of fluids in equilibrium with the average and highly altered AOC (Marschall et al. 2007), in addition to the δ11B of fluids from a diagenetic model (Boschetti et al. 2015). The values of serpentinites minerals and fluids in convergent margins (e.g., South Chamorro–Mariana Forearc) are also shown for comparison (Benton et al. 2001; Magna et al. 2006; Marschall et al. 2007; Zack et al. 2003)

Full size image

Comparing the boron isotope ratio and concentration with the known discrete fields proposed for the CSZ models (Le Voyer et al. 2010; Leeman et al. 2004; Martin et al. 2011), Aqua de Ney could derive from a mixing between dehydrating altered oceanic crust (AOC), δ11B = + 5.5‰, and sediments δ11B = −3 ± 1‰ (Campbell et al. 1994; Leeman et al. 2004). However, the models published until now consider the high-temperature melting of these sources (> 700 °C) because these were addressed to explain the isotope composition of rocks or their fluid inclusions. In the first approximation, no boron isotopic fractionation would occur at a hyperalkaline pH (> 9). Therefore, the fluid could have preserved the source signature and the deep temperature imprint during fluid upwelling. For example, a temperature of 234 °C is inferred by the δ11B-temperature relationship of Boschetti et al. (2015). The geothermometric equation in that work was produced to explain the boron release from clay minerals in diagenetic processes. However, similar to the initial AOC and sediment values, that model was checked at the starting values of + 5 and − 5‰ of smectite and illite, respectively, because clays are the main boron sources in these reservoirs, at least at the low-temperature dehydration of the slab (Rosner et al. 2003). Similar δ11B values of the expelled fluids could be obtained also by the model of Peacock and Hervig (1999), which further supports the viability of both the approaches.

In contrast, the mean value δ7Li = + 16.4 ± 0.3‰ obtained for Aqua de Ney could be obtained by different sources or processes. For example, the slab serpentinites have δ7Li values between approximately + 3 and + 14‰ (Marschall et al. 2007), whereas the fluid from serpentinite dehydration in CSZ was expected to be δ7Li > 10‰ (Magna et al. 2006). Values higher than + 40‰ detected in the pore waters circulating in the serpentinite mud of the Mariana Forearc were explained as 6Li sequestration by the precipitating mineral (Lui-Heung et al. 2007; Zack et al. 2003). Our value corresponds to that of a inferred fluid obtained by (slab-)dehydration of an average or highly altered AOC at 200 and 290 °C, respectively (Marschall et al. 2007). Therefore, by combining the δ11B and δ7Li results of the Aqua de Ney sample and comparing them with the “low-temperature slab dehydration models,” a similar temperature of approximatively 240 °C was obtained (Fig. 3b). The same value of δ7Li = + 16.4‰ was calculated for a hydrothermal fluid in equilibrium at 140 °C with the smectite from the Juan De Fuca Ridge sediment (Decitre et al. 2004). However, (i) the value of that fluid should be δ7Li = + 13.6‰ considering an up-to-date clay-water fractionation factor (Tomascak et al. 2016; Vigier et al. 2008); and (ii) the effect of the water–clay interaction should produce lower values at increasing temperatures, as shown by the path of the hydrothermal vents from that ridge (Fig. 3b). This further confirms that the deep dehydrating slab contributed to the Aqua de Ney fluid genesis.

3.2 Cation Geothermometry and Activity Calculations

Despite the very high dissolved silica concentration at Aqua de Ney, the extreme pH and the concomitant formation of silicate anions and polymeric species poses a problem regarding the use of silica geothermometers, which assume the presence of dissolved orthosilicic acid, i.e., H4SiO4 or SiO2(aq). In this case, the most useful chemical geothermometric approach to infer the temperature reached by the waters in depth could be the cation exchange based reactions. Considering the local mean annual air temperature at the sampling height (T a = 10 °C at 965 m) and the mean temperature of the water (T w = 13 ± 3 °C), the springs can be classified as “orthothermal” (T a < T w < T a + 4 °C; Schoeller 1962). Using the mean composition of the Aqua de Ney, a temperature of 96 ± 10 °C was obtained by the Na–K geothermometric equation, supposing an equilibration between low-temperature albite and K-feldspar (Verma and Santoyo 1997). However, Na–Li equations produce very different values (Sanjuan et al. 2014): from 209 °C based on the relationship for hydrothermal vents to 24 °C using the equation for geothermal waters with Cl < 0.3 M. Undoubtedly, the cooling to supergenic conditions may have caused the dissolved constituents to re-equilibrate to the formation of Na-bearing authigenic minerals, e.g., kenyaite (Barnes et al. 1972). The recalculation of the activities of the dissolved species at increasing temperatures and pressures shows the proximity of the sample to the albite/K-feldspar equilibrium line but shifted toward albite and jadeite fields at high T–P conditions (Fig. 4). The water isotope composition and the mean temperature of 240 °C obtained by δ7Li and δ11B low-T slab model agree with this hypothesis and are similar to the albite forming fluids from primary jadeite in subduction zones (Johnson and Harlow 1999; Morishita et al. 2007). Finally, the comparison between the constituent concentration of Aqua de Ney and H2O-rich fluids from the dehydrating slab (Manning 2004) shows that the former falls between the natural samples from the serpentinite mud volcanoes and the composition of fluids from deeper P–T settings obtained by calculations or lab-modelings (Supplementary File S3). Establishing the exact mineral assemblage buffering the pH at depth and after cooling is difficult. However, our calculated free pH at high T–P conditions (Fig. 4) agrees with the value obtained by the serpentinization model of McCollom and Bach (2009). Moreover, according to other published modeling and experimental results concerning serpentinizing fluids (Frost and Beard 2007; Sekine et al. 2015), the high silica activity of Aqua de Ney could be buffered by serpentine-talc instead of serpentine–brucite pair. This is exhibited in the study area by: (i) the presence of talc-rich serpentinite formed along the thrust as a result of the infiltration of deep silica-bearing metasomatic fluids (Peacock 1987) and (ii) their similar isotope composition (Liakhovitch et al. 2005). Comparing the logK of the general reaction:

$${\text{serpentine }} + \, 2{\text{SiO}}_{2} \left( {\text{aq}} \right) \, = {\text{ talc }} + {\text{ H}}_{2} {\text{O}}$$
(1)

with the log(activity) of the dissolved species in Aqua de Ney at a high pressure (5 kbar), the mean equilibrium temperature is quite similar to the temperature obtained by the slab model (Supplementary File S4). Therefore, most probably the equilibrium (1) buffered the silica concentration at depth, whereas according to Barnes (1972) and to the low saturation indexes of the silica phases (Supplementary File S1), at shallower conditions the silicate anions concentration gradually increased due to the high pH and dacite dissolution.

Fig. 4
figure 4

Na–K activity diagrams. The activity of the dissolved species in Aqua de Ney (N-diamond) and pH were recalculated at different T–P conditions by EQ3/6 software (Wolery and Jarek 2003). T = 200 °C and P = 2000 bar (2 × 108 Pa) conditions agree with the local depth of Trinity Ophiolite (Fuis et al. 1987)

Full size image

4 Comparison of the New Results with Previous Gas Analysis

Examining the composition of the gases bubbling in the spring water, Mariner et al. (2003) evidenced, in the ternary N2/100-10He-Ar diagram, a similarity between Aqua de Ney and the fumaroles at the summit of neighboring Mt. Shasta (Cascade Range). Nathenson et al. (2003) inferred the presence of a geothermal system at 210 °C beneath Mt. Shasta, which is similar to our calculated temperature. The free gas composition and the temperature similarity of the Mt. Shasta system and Aqua de Ney could be a mere coincidence considering: (i) the near neutral to acid pH of the Shasta springs and their meteoric composition; and (ii) the location of the Aqua de Ney spring between Mt. Shasta/Cascade Range and the Trinity Ophiolite/Eastern Klamath Province (Supplementary File S5). However, we cannot neglect that the N2/He volume ratio between 8985-4264 and the Ar = 0.02 vol% of Aqua de Ney fall within the typical composition of the “andesitic” gases (Giggenbach 1996). In contrast, the N2/Ar between 1123-1066 and δ15N = + 3.5‰ of Aqua de Ney (Mariner et al. 2003) could testify an origin of these free gases from organic matter in clay minerals, e.g., from the metasedimentary rocks of the Franciscan Complex (Zhu et al. 2000). These rocks, whose gases offer a window into the effects of metamorphism in the relatively “cool” subduction zone beneath the forearc (Mitchell et al. 2010; Sadofsky and Bebout 2004), have outcrops in fault contacts with the rocks of the Klamath Province in the western sector, whereas in the area of the spring the metasedimentary rocks are buried beneath the Trinity Ophiolite with stratigraphic contact at approximately 7–8 km (Fuis et al. 1987). Moreover, combining the N2/He and δ15N values, a similarity between the free gas from Aqua de Ney and the gases from geothermal fluids in volcanic arcs transpires (Fischer et al. 2002). This interdigitation between hydrothermal and metamorphic environments is not a surprising feature in a convergent margin, particularly in this area where a forearc-arc switch occurs. Finally, the isotopic composition and temperature obtained (240 °C) by the dehydrating slab model of Aqua de Ney from this study agree with the deep fluid path in the area presented by Elder and Cashman (1992) (Supplementary file S5). The attached section clearly shows how the local deep tectonic structure could collect the fluids from the deep dehydrating slab but consequently also intercept and carry the gases from the metasedimentary rocks or other units toward a more shallow level.

5 Conclusion

Notwithstanding the hyperalkaline composition and the local ultramafic lithology, this study shows how serpentinization was not the only processes that determined the O–H–Li–B isotope ratios of Aqua de Ney. Most probably, it was a fossil fluid of seawater origin, comparable in terms of initial composition and age to Boswell Spring. The serpentinization and slab dehydration processes had deeply modified the chemical and isotope composition of the early fluid. In particular, the high pH (hyperalkaline) and the serpentine supersaturation agree with the serpentinization, whereas the slab dehydration explains the high boron concentration and the low δ11B value. Both processes occurred at high temperature and concurred to determine the 18O-enrichment of this water.