Seawater-basalt interaction taking place at mid-ocean ridges was studied using numerical modeling to determine the compositional evolution of hydrothermal fluids and associated alteration mineralogy forming within newly emplaced crustal material. Geochemical modeling was carried out in a closed seawater-basalt system at discrete temperature intervals between 2 and 400°C at 500 bars, varying fluid/rock ratios, and secondary mineral assemblages representative of basalt alteration in natural systems. In addition to temperature, the fluid/rock ratio has a fundamental control on the resulting system chemistry. At rock-buffered conditions (low fluid/rock ratios), the mineral-solution equilibrium was characterized by high cation to proton activity ratios for aCa2+/(aH+)2 and aNa+/aH+ and very low dissolved Mg concentrations due to the precipitation of smectites and chlorite. A complex secondary mineral alteration assemblage dominated by Ca- and Na-bearing minerals including zeolites, calcite, epidote, prehnite, clinozoisite, and albite was predicted to form. The resulting fluid composition was alkaline and reduced relative to ambient seawater, with Eh values ranging between −0.2 and −0.6 V. In contrast, seawater-buffered conditions (high fluid/rock ratios) resulted in lower cation to proton activity ratios for aCa2+/(aH+)2 and aNa+/aH+ and higher dissolved Mg concentrations comparable to the value of this element in ambient seawater. A more simple mineral assemblage was predicted to form at these conditions with the predominance of Al-Si- and Mg-bearing minerals including kaolinite, quartz, and talc in addition to large amounts of anhydrite. The resulting fluid composition was mildly acidic and oxidized relative to seawater with Eh values ranging between −0.2 and 0 V. These modeling results were compared to a compilation of submarine hydrothermal vent fluid compositions from mid-ocean ridge settings and analogous basalt-dominated environments. The agreement obtained between the simulations and the compiled fluid data indicates that mid-ocean ridge hydrothermal processes can be closely reproduced by mineral-solution equilibria for a broad range of temperatures and fluid/rock ratios.

1. Introduction

Mid-ocean ridges account for more than 60 percent of the Earth’s global magmatic budget [1] and generate a hydrothermal energy output of 1.8 ± 0.3 TW [2] via both low- (<200°C) and high- (>200°C) temperature fluid fluxes. Since the first discoveries of active high-temperature hydrothermal vents in the late 1970s [3, 4], it has been recognized that seawater circulation within the oceanic crust represents one of the principal mechanisms of heat transfer. During circulation, seawater is heated to a maximum temperature of ~400°C [5, 6] and is chemically modified due to interaction with rocks along the fluid flow pathways [7, 8]. Because the permeability of the oceanic crust varies with depth and from one vent field to another, various regimes of seawater circulation are expected to occur, affecting the chemistry of the hydrothermal fluids [9, 10]. In particular, the fluid/rock (F/R) ratio has a pronounced effect on the composition of both the fluids and alteration mineralogy produced during fluid/rock interaction [11, 12].

In a pioneering work, Bowers and Taylor [13] simulated the evolution of hydrothermal fluids at the 21°N East Pacific Rise vent field at temperatures between 100 and 350°C and an average F/R ratio of 0.5. Berndt et al. [14] and Seyfried et al. [15] showed that temperature-dependent fluid/rock equilibria between dissolved Ca, Na, and Si and their corresponding mineral buffers could explain the variation of these elements in mid-ocean ridge hydrothermal fluids. Wetzel and Shock [16] extended this observation to a broader fluid-mineral compositional range at temperatures between 350 and 400°C and an F/R ratio of 1. Although realistic hydrothermal fluid compositions were obtained in these studies, the applicability of thermodynamic modeling was limited by the choice of a fixed F/R ratio. Bowers and Taylor [13] and Bowers et al. [17] determined that the overall F/R ratio from the 21°N East Pacific Rise vent field is 0.5 based on a δ18OH2O value of +2.0‰ in a 350°C hydrothermal fluid. However, more recent vent data have reported isotopic values ranging from 0 to +2.5‰ [1820], suggesting that the F/R ratio in mid-ocean ridge hydrothermal systems can vary significantly. Fluid/rock ratios have been estimated to range from ~1 to ~100 based on various methods including boron isotopic and energy balance calculations [2, 9, 2124].

To better understand the processes of fluid/rock interaction taking place during seawater circulation through the oceanic crust, the present contribution examines the factors controlling the chemistry of hydrothermal fluids generated at mid-ocean ridges. Numerical modeling was conducted using the GEMS code package [2528] based on the Gibbs energy minimization (GEM) method to constrain the influence of the rock composition, F/R ratio, and temperature on the fluid composition and corresponding alteration mineralogy formed during seawater circulation in the oceanic crust. Modeling results are compared to the fluid and mineral chemistry of basalt-dominated subaerial and submarine hydrothermal systems to demonstrate the validity of the simulations. The agreement between modeling results and the composition of natural waters indicates that mid-ocean ridge hydrothermal systems share similar patterns of fluid evolution, which can now be effectively modeled using the GEM method.

2. Methods

2.1. Geochemical Modeling

Numerical simulations were conducted in the Na-K-Mg-Ca-Ti-Fe-Al-C-Si-S-F-Cl-H-O system using the GEM-Selektor v.3 code package (http://gems.web.psi.ch). This program evaluates chemical equilibria among fluid and minerals using the GEM method, which provides advantages for the calculation of pH and redox in complex hydrothermal fluids in comparison to the more conventionally used law of mass action (LMA) type codes [2528]. Thermodynamic data used for the minerals and aqueous species were obtained from the MINES 17 database (http://tdb.mines.edu) [29]. Data for rock-forming minerals were taken from Holland and Powell [30] and Gysi and Stefánsson [31], and data for aqueous species were taken from Supcrt92 [3234]. This dataset was updated with experimental data for the dissociation constant of HCl [35] and the internally consistent dataset of Miron et al. [36] for Na-K-Al-Si obtained using the GEMSFITS code package [37], extended to Ca-Mg-CO2 by Miron et al. [38]. A list of minerals included in the simulations is given in Figure 1.

The standard state adopted for minerals was that of the pure phase at any temperature and pressure. An ideal solid solution model was used for sheet silicates and zeolites, and the estimated thermodynamic properties for their endmembers can be found in Gysi and Stefánsson [31] and Gysi [29]. The standard state adopted for pure H2O and for the aqueous species was unit activity in a hypothetical one molal standard solution referenced at infinite dilution at any pressure and temperature. The extended Debye-Hückel activity model was used for aqueous species assuming NaCl as the background electrolyte [52, 53]. All activity models and equations of state were calculated using the TSolMod library class implemented in GEM-Selektor [54].

2.2. Geological Parameters and Simulation Setup

Seawater-rock interaction was simulated under closed-system conditions using a titration model for temperatures ranging from 2 to 400°C in steps of 2°C, a constant pressure of 500 bars, and F/R ratios ranging from 1 to 100. The fluid/rock equilibrium was calculated at each temperature step to monitor the dissolution of primary minerals, precipitation-dissolution of secondary minerals, and associated changes in the composition of seawater. The model is aimed at approximating the circulation and equilibration of fresh seawater in a young mid-ocean ridge crust, where seawater reacts with unaltered basalts along a prograde temperature path covering a vertical section from the seafloor to the top of the subaxial magma chamber. The calculations were repeated for different F/R ratios (i.e., one section per F/R ratio), assuming that the value of the F/R ratio represents the overall fluid/rock equilibria attained in a section. The F/R ratio can be related to the natural circulation of fluids within a given volume of rock that yields a given amount of mineral surfaces accessible for reaction. Natural open systems involve complex dynamic fluid flow in time and space, largely due to variations in rock porosity and permeability between various rock units and depending on fracture networks and faulting. The F/R ratio is a concept that can be seen as the effective result of fluid/rock interaction in the open hydrothermal system, where local chemical equilibrium has been achieved.

The compositions of mid-ocean ridge basalt and seawater used in the numerical model are listed in Table 1. The basalt composition was calculated by averaging the major element composition of eight fresh basaltic glasses and their reported Fe3+/∑Fe ratios (data from Kelley and Cottrell [55]), which is an important parameter to fix the initial redox potential of the system. Major element analyses of the basaltic glasses were normalized to 100 wt.%, without taking into account the low concentrations in MnO and P2O5. The composition of seawater was taken from the reference composition of Millero et al. [56] and normalized to the components of the system. The minor concentration of Sr2+ in seawater was converted to Ca2+ to maintain charge balance. Similarly, Br was converted to Cl and B(OH)4 to OH.

The upper temperature limit of the model corresponds to the highest vent temperature observed in mid-ocean ridge hydrothermal systems (405–407°C) [19, 57]. A fixed pressure of 500 bars was used in the simulations to take into account the hydrostatic pressure prevailing in the reaction zone of mid-ocean ridge hydrothermal systems, assuming an average water depth of 2500 m [58] and a distance of 2500 m between the seafloor and the top of the axial magma chamber [59, 60]. Because previous studies have established that the F/R ratio varies from approximately 1 to 100 by mass in mid-ocean ridge hydrothermal systems [2, 9, 2124], the numerical simulations were carried out using F/R mass ratios of 1, 5, 20, 50, 70, and 100. These ratios were achieved by reacting 1 kg of rock with the required amount of seawater and are thus defined as the total weight of seawater that will equilibrate in the system to the weight of unreacted rock.

2.3. Model Assumptions

In the numerical simulations, alteration minerals were allowed to precipitate during fluid/rock interaction upon reaching saturation and partial equilibrium with the aqueous fluid. This model is similar to the classical irreversible reaction of feldspar hydrolysis by Helgeson [61] and Helgeson et al. [62], where a series of partial equilibria take place upon continued reaction progress of feldspar dissolution until the mineral-fluid system reaches an overall equilibrium. However, complex multicomponent-multiphase systems are more difficult to predict, as saturation may be reached with several of the primary igneous minerals depending on temperature [63], as well as with secondary minerals upon rock alteration. This implies that certain primary minerals such as olivine will only dissolve until they reach saturation but will not reprecipitate, whereas others such as feldspar may dissolve and reprecipitate to form secondary feldspar with a different composition, as observed in natural geothermal systems [64].

The precipitation of certain low-temperature alteration minerals, particularly sheet silicates and zeolites, may be inhibited by slower reaction kinetics. As a result, the simulations may predict the formation of stable mineral phases that are not observed in nature for a given temperature. For instance, the numerical simulations predict trioctahedral smectites and chlorite to be stable across the entire temperature range considered in the model. This contradicts observations made in natural systems, where smectite represents a predominant alteration mineral in basaltic environments at temperatures <150–200°C, whereas chlorite is known to form at higher temperatures [49, 65]. For this reason, upper and/or lower temperature metastability constraints were added in the code for the precipitation of minerals according to the alteration mineralogy observed in natural systems. The primary and secondary minerals considered in the simulations are listed in Figure 1.

3. Results

3.1. Simulated Alteration Mineralogy

A summary of the alteration mineralogy obtained from the simulation of seawater-basalt interaction is presented in Figure 2. Both the temperature and the F/R ratio have a principal control on the predicted mineral assemblages. The overall reaction path can be exemplified by the results obtained at an F/R ratio of 1. At low temperature (<200°C), Na-Ca-bearing zeolites, trioctahedral (Ca)-Mg-Fe-bearing smectites, and calcite were stable, whereas at high temperature (>200°C), these minerals were replaced by chlorite-(Mg-Fe), albite, quartz, and Ca-bearing aluminosilicates, including epidote, wairakite, prehnite, and clinozoisite. Trioctahedral (Ca)-Mg-Fe-bearing smectites were replaced by chlorite-(Mg-Fe) at high temperature, illustrating a competition for Mg and Fe between these minerals. A sharp discontinuity can be observed in the model at 200°C due to the temperature constraints of mineral precipitation used for the sheet silicates and zeolites. Scolecite, thomsonite, and calcite constitute the stable alteration minerals competing for Ca at low temperature, followed by their progressive replacement by epidote, wairakite, and prehnite with increased temperature. The Ca-bearing mineral clinozoisite was stabilized at high temperature, between 370 and 380°C. In the group of Na-bearing minerals, natrolite was stable at low temperature and replaced by analcime above 100°C, followed by albite above 200°C. Smectite-(K) was the only K-bearing mineral formed in the simulations. Hematite was stable up to 200°C and pyrite up to 310°C, where it became replaced by pyrrhotite. Titanite was the only stable Ti-bearing mineral present at all temperatures.

The number of minerals formed at a given temperature varied with the F/R ratio. At low F/R ratios (F/R < 50), representing rock-buffered conditions, a larger number of alteration minerals were formed with complex competing reactions among low-temperature zeolites and high-temperature Ca-Al-Si-bearing minerals (epidote, prehnite, and clinozoisite). At high F/R ratios (F/R > 50), representing seawater-buffered conditions, the total number of alteration minerals decreased and relatively simple mineral alteration assemblages were formed. Albite progressively decreased towards higher F/R ratios until disappearing completely at a F/R value of 100 above 325°C. Kaolinite appeared towards higher F/R ratios at the expense of the Na-Ca-bearing zeolites such as natrolite and scolecite. Increasing amounts of anhydrite were formed towards higher F/R ratios in competition with other Ca-bearing minerals such as epidote and wairakite until F/R values of >50 were reached, where all the Ca was consumed by the precipitation of anhydrite. Talc was present at relatively high F/R ratios (F/R of 50 and 70) and was accompanied at low temperature by the replacement of calcite by ankerite. Hematite and pyrite increased progressively towards higher F/R ratios at low temperature, while at high temperature, hematite was absent and pyrite decreased until disappearing at a F/R value of 20. Pyrrhotite was replaced by magnetite at a F/R value of 5 and rutile replaced titanite at a F/R value of 50. The amounts of chlorite, smectites, and quartz were not significantly influenced by the F/R ratio.

3.2. Simulated Fluid Evolution

Both the temperature and the F/R ratio also affect the calculated seawater chemistry during interaction with basaltic rocks. Figure 3 shows the modeled fluid composition for dissolved Mg, Ca, Na, and K concentrations and pH and Eh (redox potential) in comparison to the initial composition of unreacted seawater (Sw2°C). The discontinuity occurring at 200°C is analogous to that obtained in the secondary mineralogy and results from the mineral metastability constraints used for smectites and chlorite, as described above.

The Mg concentration of the simulated fluids was very low at low temperature and low F/R ratios (F/R of <50) and increased with higher temperature and F/R ratio (F/R of >50). At high F/R ratios (F/R value of 100), the Mg concentration of the fluids was several orders of magnitude higher and approached the concentration of unreacted seawater. The mobility of Mg was controlled by the stability of trioctahedral smectite-(Mg) at low temperature and by chlorite-(Mg) at high temperature. The increase of Mg in the fluids at high temperature and low F/R ratios relates to the lower amounts of chlorite precipitation at these conditions in comparison to the amounts of smectites formed at low temperature (Figures 2 and 3). At higher F/R ratios, the large volumes of reacted seawater involved in the system preserved high Mg concentrations, while large quantities of smectites and chlorite were predicted to precipitate.

The mobility of Ca was controlled by the stability of carbonates, zeolites, and smectites at low temperature and by anhydrite, epidote, wairakite, prehnite, and clinozoisite at high temperature. The highest concentrations of dissolved Ca were achieved at low F/R ratios for all temperatures except between ~50 and 200°C (Figures 2 and 3). The diverging fluid evolution for dissolved Ca between ~50 and 200°C can be related to the excessive uptake of Ca by scolecite. The decrease of dissolved Ca with increasing F/R ratios at temperatures above 120°C was caused by the precipitation of anhydrite at these conditions (Figures 2 and 3).

The mobility of Na was controlled by the stability of zeolites and smectites at low temperature and by albite at high temperature (Figures 2 and 3). The concentration of Na of the simulated fluid did not vary significantly with temperature and remained within the range of the unreacted seawater value after seawater-basalt interaction. At low F/R ratios, a slight enrichment of Na was observed in the simulated fluid below 200°C, while a slight decrease in Na was observed at temperatures above 200°C due to the formation of albite over Na zeolites.

The mobility of K was controlled by the stability of smectite-(K) at low temperature, while K was not transferred from the primary minerals to the alteration assemblage at high temperature (Figures 2 and 3). The concentrations of dissolved K at high F/R ratios are comparable to the value of this element in unreacted seawater. However, at low F/R ratios and temperatures below 120°C, the dissolved concentrations of K were lower, whereas above 120°C, these values exceeded unreacted seawater concentrations.

The simulated pH varied significantly in the simulations and is characterized by an overall decrease with increasing temperature and fluid/rock ratios (Figure 3). At low temperature and low F/R values, the fluid is characterized by alkaline conditions reaching pH values of ~10 to 11, consistent with the formation of zeolites and calcite as major alteration minerals. Increasing F/R values to >50 caused a moderate shift of the pH to values reaching between ~8.5 and 9. Increasing temperatures to 400°C led to a considerable pH decrease with values reaching close to 6 at low F/R ratios and values close to 4 at high F/R ratios. These lower pH values are in agreement with the stability of kaolinite at these conditions. The overall decreasing trend of pH with temperature can be related to the increased endothermic ionization of pure water into H+ and OH, which translates into an increased dissociation constant () towards higher temperatures. This process shifts the neutral pH value from 7 at room temperature to ~5.5 at 200°C and back to ~6.1 at 400°C. In addition to the shift of neutral pH for pure water, the increase in temperature causes seawater to become more acidic due to the precipitation of anhydrite and magnesium hydroxide sulfate hydrate [66] between 180 and 400°C and to the increased association of OH-bearing aqueous complexes. The pH curve of seawater, shown in Figure 3, is similar to the one obtained by Bischoff and Seyfried [5]. The simulated pH resulting from fluid/rock interaction follows a similar trend, but either plots above/below this curve, yielding more alkaline/acidic conditions, respectively. These changes can result from a combination of ionization of the acids, and the overall reaction progress of fluid/rock interaction, which involves proton (H+) consumption and cation release upon dissolution of primary minerals and dissolution-precipitation of secondary minerals.

The simulated redox potential (Eh) of seawater reacted with basalt is characterized by an overall increase with increasing temperatures and fluid/rock ratios (Figure 3) and by an overall shift towards negative Eh values in comparison to the initial value of unreacted seawater (EhSw, 2°C = −0.22 V). The simulated Eh values were observed to decrease from ~0 V at high fluid/rock ratios to almost −0.6 V at low fluid/rock ratios, indicating that increased rock buffering will cause more reduced conditions in the chemical system. The redox potential of the system is buffered by the initial Fe3+/∑Fe ratio of the rock and by the oxidation state of the secondary minerals formed in the simulations. The mobility of Fe is controlled by the stability of hematite, pyrite, pyrrhotite, magnetite, epidote, and ankerite. A decrease in Eh values is related to the precipitation of larger amounts of Fe(II)-bearing minerals such as pyrite, pyrrhotite, and magnetite relative to Fe(III)-bearing minerals such as hematite.

4. Discussion

4.1. Influence of Rock Crystallinity

The oceanic crust at mid-ocean ridges consists of a ~0.5 km thick upper succession of glassy-to-finely crystalline basalts forming pillows, massive flows and breccias, a ~1.5 km thick sheeted dike complex composed of diabase, and a several km thick basal succession of coarse-grained gabbroic rocks [9, 67]. These rocks are characterized by a similar major element composition, but vary in texture and mineralogy, which may affect the rate and stoichiometry of ionic exchange, particularly at low temperature where mineral dissolution may be controlled by reaction kinetics [29, 31, 6872]. Even if reaction kinetics is of key importance for basalt alteration at lower temperatures [7376], its relative importance with respect to fluid/rock equilibria processes is somewhat unclear. To answer these questions, the effects of temperature, fluid and rock compositions, and extent of reaction (F/R ratios) need to be studied separately.

Previous experimental and modeling work on the alteration of both glassy and crystalline mafic rocks suggests that although basaltic glass dissolves faster than more crystalline basaltic rocks [77], the crystallinity of the protolith has only a limited influence on the formation of secondary minerals when approaching fluid/rock equilibria at hydrothermal conditions [51, 71, 7882]. Similar observations have been made in Iceland, where low temperature alteration products of volcanic glasses have similar compositions to those formed during the alteration of crystalline rocks on longer geological timescales [83, 84]. This suggests that the relatively small volume of basaltic glass present at shallow depths beneath mid-ocean ridges may have a similar influence on the overall chemical mass balance associated to fluid/rock interaction as crystalline basalts of the same composition.

4.2. Comparison of the Simulated Alteration Mineralogy to Natural Systems

The nature and distribution of alteration minerals present in mid-ocean ridge hydrothermal systems is difficult to constrain as this would involve the deep drilling of active seafloor hydrothermal systems, where coexistence of hydrothermal fluids and rocks can be assured. However, it is well known that the alteration mineralogy and chemical composition of fluids in hydrothermal systems at close to equilibrium conditions are essentially determined by the initial chemical composition of the fluids, rock composition, and temperature [85, 86]. Therefore, similar alteration patterns can be expected to occur in similar petrogenetical settings, as indicated by previous studies [41, 48, 87]. Additional information on the nature of the minerals present in equilibrium with hydrothermal solutions in mid-ocean ridge hydrothermal systems can hence be obtained from active basalt-hosted geothermal systems where the rock composition and temperature range are analogous.

The formation of zeolites at low temperature in the simulations is consistent with the occurrence of these minerals as low grade alteration products in volcanic environments [88, 89]. Scolecite occurs in the early and intermediate stages of burial metamorphism of basaltic lavas in Iceland [88]. Natrolite, analcime, and thomsonite are typically found as alteration products of mafic rocks, where they develop within late stage alteration assemblages [47, 90]. The appearance of wairakite at high temperature in the model is supported by the presence of this mineral in active geothermal systems of New Zealand at temperatures exceeding 200°C [48, 91]. The formation of calcic zeolite is favored with increasing degrees of alteration, particularly in mafic rocks, due to the rise of Ca-Na activity ratios resulting from the progressive albitization of calcic plagioclase [92].

Trioctahedral smectites were also predicted to form in the low temperature section of the model. This is consistent with observations made in natural volcanic environments, where smectites commonly occur as an alteration product of volcanic glasses and from the breakdown of olivine, pyroxene, and feldspar [93]. The transition between trioctahedral smectites and chlorite was constrained to occur at 200°C in the present model due to the difficulties in reproducing intermediate metastable mixed-layer phases. In natural environments, the transformation from trioctahedral smectites to chlorite is more gradual and generally involves a continuous mixed-layer chlorite/smectite series or a discontinuous sequence consisting of smectite ± corrensite ± chlorite [49, 94, 95]. It is therefore expected that the major discontinuity observed in the model at 200°C is artificial and could not be corroborated by geochemical observations of natural systems, where this transition occurs over a temperature range of at least 50°C [49]. This artifact could be removed in future modeling efforts by the implementation of the thermodynamic data of these minerals and their intermediate phases.

The presence of kaolinite in the simulations is consistent with the occurrence of this mineral in several geothermal systems from Iceland, Japan, and New Zealand [48, 82, 96] as well as in submarine hydrothermal environments dominated by mafic rocks [97, 98]. Kaolinite occurs in altered mafic to felsic volcanic rocks and characterizes cation-depleted fluid/rock equilibria at relatively acidic conditions [43, 48, 99]. The appearance of talc in the intermediate to high temperature range of the simulations does not have any analogs in active geothermal systems, where this mineral is only rarely observed. However, talc is present in submarine hydrothermal systems within active chimneys and massive sulfide mounds, where it was observed as secondary mineral precipitates [50, 100, 101]. Other low temperature minerals formed in the model include the carbonates calcite and ankerite. The precipitation of calcite is in agreement with the occurrence of this mineral in mafic rocks as infill of original open spaces such as vesicles and cements in volcaniclastic facies and as euhedral crystals replacing olivine and plagioclase [88, 102]. The low temperature occurrence of the Fe carbonate ankerite has been reported in young basalts from the Galápagos Spreading Center, where it is thought to be metastable [103], and from laboratory experiments [31]. Celadonite is widely documented as a product of low temperature alteration of basalt [9] and is also predicted to form in the model at temperatures between 2 and 170°C, where it coprecipitates with dolomite. Dolomite has, however, not been observed in seafloor hydrothermal systems at low temperature [104], where its formation is likely inhibited by kinetic restrictions [105].

The simulated high temperature mineral assemblage characterized by epidote, albite, prehnite, and clinozoisite is consistent with the alteration mineralogy observed in natural systems. The presence of albite above the smectite-chlorite transition is in agreement with the occurrence of feldspar as an alteration phase in geothermal systems at temperatures exceeding 200°C, where it forms by dissolution and replacement of primary plagioclase [64, 106109]. Hydrothermal epidote and prehnite coexist in geothermal systems and represent common alteration products of volcanic rocks at temperatures above 230–260°C [43, 48, 110]. Epidote and prehnite are typical of low temperature metamorphism, and the growth of these minerals sets the boundary between late stage diagenesis and metamorphism where volcanic glass is replaced by sheet silicates in subaerial hydrothermal systems [48, 111]. Clinozoisite occurs as a minor phase accompanying epidote within geothermal systems in zones of magmatic-hydrothermal alteration and contact metamorphism [43, 48].

Anhydrite is documented in most active geothermal systems where it precipitates over a broad range of temperatures [48] and is also a common component of chimney structures in submarine hydrothermal systems. Its pronounced retrograde solubility leads to dissolution in seawater if the high temperature fluid flow is interrupted or terminated [112]. Anhydrite was predicted to form in the model, but became unstable at temperatures below ~150°C. Pyrrhotite and pyrite have both been described from active black smoker chimneys and high temperature massive sulfides recovered from mid-ocean ridge vent fields [44, 113, 114]. The equilibrium mineral assemblage that is precipitated during mixing between the hydrothermal fluid and seawater depends to a large extent on the chemical buffering of hydrothermal fluids by fluid/rock interaction processes [13]. Fluids that are buffered to lower Eh values produce a pyrite-pyrrhotite-magnetite assemblage while fluids at higher Eh values precipitate only pyrite; pyrrhotite is also commonly replaced by pyrite at low temperature [44, 115, 116].

4.3. Comparison of the Simulated Fluid Chemistry to Icelandic Waters

As described above, the modeled alteration mineralogy is consistent with the various mineral assemblages occurring as a result of basalt alteration in natural hydrothermal systems. The applicability of the model can also be tested through a comparison of the modeled fluid chemistry including the activities of Mg2+, Ca2+, and Na+ with the measured concentrations of these elements in natural geothermal waters of Iceland (Figure 4) from the studies of Stefánsson et al. [63] and Stefánsson and Arnórsson [108]. A comparison between the simulated fluid chemistry resulting from seawater-basalt interaction at mid-ocean ridges and natural waters from the Icelandic geothermal systems is justified by the similar host basaltic rock composition and the nearly identical temperature range of the two hydrothermal environments. It has to be noted, however, that the starting fluid compositions are different, and that the geothermal systems in Iceland are mainly characterized by low chlorine concentrations, with only few exceptions that include variable amounts of mixing with seawater [85, 117].

The overall fluid/rock interaction presents analogies to a chemical titration process [62, 118], where the basalt acts as a base by consuming protons and releasing cations upon dissolution, while the fluid acts as an acid due to the presence of components containing ionizable hydrogen in seawater [5]. The cation to proton (H+) activity ratio of natural waters is a good indicator for this fluid/rock interaction process and yields information on the reaction progress associated to basalt alteration [31, 71, 119]. As a consequence, the cation to proton ratio will increase with increasing reaction progress or varying F/R ratios at fixed temperature. Figure 4 shows these comparisons, where the variations of the cation to proton activity ratios for various simulated F/R ratios reflect the combined effects of primary mineral dissolution and secondary mineral precipitation-dissolution reactions, as well as the release of protons from the ionization of the acids present in the water followed by their neutralization by the rock. Notable variations can be observed in these simulations for aCa2+/(aH+)2 and aNa+/aH+ ratios as function of F/R varying from 1 to 100. The ratio aMg2+/(aH+)2, however, maintains constant values across the entire range of simulated F/R ratios. This can be explained by the fact that Mg2+ is buffered by the stability of the sheet silicates (i.e., smectites and chlorite), which have a considerable control on pH and hence aH+. The simulated cation to proton activity ratios also show a systematic decrease with temperature. This overall trend can be explained by an increase in the ionization of pure water and by the increase in acidity of heated seawater alone, as shown for pH in Figure 3.

The excellent agreement between the simulations and the natural Icelandic waters up to temperatures of 300°C (Figure 4) provides confidence in the capacity of the model to estimate the processes affecting the fluid chemistry of natural open systems. This reinforces previous findings, which indicate that fluid/rock interaction may approach equilibrium conditions in geothermal systems, where the fluid chemistry is controlled by the equilibrium of the fluids with primary and secondary minerals [63, 85, 119]. Furthermore, the major element chemistry in these systems can be predicted by thermodynamic modeling down to a temperature of ~50°C, where kinetic factors begin to have a dominant control on the alteration mineralogy and water chemistry [63, 85, 120].

4.4. The Role of Fluid/Rock Ratios for Basalt Alteration at Mid-Ocean Ridges

The F/R ratio effective during hydrothermal alteration at mid-ocean ridges has an important control on both the mineralogy of the altered basalt (Figure 2) and the chemistry of the vent fluids (Figures 3 and 4), confirming previous results by Mottl and Seyfried [11] and Reed [12]. Two end-member scenarios can be distinguished based on the present simulations, corresponding to a rock-dominated and seawater-dominated seafloor hydrothermal system, respectively.

Rock-buffered conditions (F/R of 1 to 50) are characterized by relatively high values of aCa2+/(aH+)2 and aNa+/aH+. These high cation to proton activity ratios can be directly related to the primary mineralogy of mid-ocean ridge basalts, which includes plagioclase and diopside. At these conditions, the fluid/rock interaction chemistry is dominated by the composition of the fresh rock and the effect of seawater is largely catalytic [85, 86]. High Ca2+ and Na+ activities at rock-buffered conditions favor the formation of alteration assemblages including calcite and zeolites (scolecite, natrolite, thomsonite, and analcime) at low temperature, while epidote, prehnite, albite, wairakite, and clinozoisite are predicted to form at high temperature (Figure 2). The high variations in total dissolved Ca, Na, and K at low F/R ratios (Figure 3) can be related to extensive fluid/mineral interaction, characterized by numerous secondary mineral phases buffering the fluid composition. The concentrations of total dissolved Ca at low F/R ratios in the model are higher than those at high F/R ratios for most of the temperature range covered, as supported by similar results from experimental work of Hajash and Chandler [121]. In contrast, the total dissolved Mg concentration was negligible at rock-buffered conditions, with the lowest values < 10−1 mmol/kg obtained at F/R between 1 and 20. These low Mg concentrations are explained by the near complete uptake of seawater-derived Mg2+ and OH to form smectites and chlorite. The excess H+ is consumed by hydrolysis reactions of silicate minerals [11]. Fluid/rock interaction at rock-buffered conditions is characterized by alkaline conditions, in agreement with simulated pH values ranging from 7 to 11. Redox reactions are also taking place between the reduced basaltic rocks and oxidized seawater, causing a sharp drop of the redox potential at rock-buffered conditions with Eh values ranging between −0.2 and −0.6 V. These reduced conditions correspond to the maximum stability of pyrite and the precipitation of pyrrhotite, resulting from the buffering of seawater sulfate by sulfide minerals.

Seawater-buffered conditions occur at high F/R ratios (F/R of 50 to 100) and are characterized by relatively high values of aMg2+/(aH+)2 and a low pH. At these conditions, seawater components gain increasingly in importance relative to rock components and are very effective at buffering the system to lower pH values (Figure 3). This can be related to the production of excess H+ over OH associated to the precipitation of Mg2+ and OH from the solution into a magnesium hydroxide sulfate hydrate upon heating of seawater [5]. The same mechanism occurs during the alteration of basalts with the formation of Mg-bearing hydrated minerals, where Mg2+ and OH are taken up by smectites and chlorite [11]. While the excess of H+ produced by seawater at rock-buffered conditions is consumed by the alteration minerals, the production of H+ at fluid-buffered conditions exceeds that of H+ consumption and the pH is lowered. The low pH values present at seawater-dominated conditions also have a direct influence on the alteration mineralogy. High H+ activities favor the dissolution of rock components and the precipitation of Al-Si-rich minerals such as kaolinite at low temperature. The total dissolved Mg concentrations at seawater-buffered conditions attain values several orders of magnitude higher than in the rock-buffered environment (Figure 3); an observation also made in experimental studies of Hajash and Chandler [121]. The large volumes of reacted seawater involved in the system preserved near-maximal concentrations of Mg while large quantities of smectites and chlorite were precipitated (Figures 2 and 3), indicating that the amount of Mg available from the fluid is magnitudes higher to what can be buffered by the rock and precipitated into the Mg-bearing alteration assemblage. The precipitation of significant quantities of talc at relatively high F/R values of 50 and 70 can further be explained by the high concentrations of Mg in the fluid at these conditions. The progressive decrease of aCa2+/(aH+)2 towards increasingly seawater-buffered conditions results from the precipitation of Ca with seawater sulfate into anhydrite. Anhydrite in mid-ocean ridge hydrothermal systems is characterized by isotopic 34S values identical to those measured in seawater sulfate [112, 122], validating the importance of seawater as a source of available sulfur to form this mineral. The circulation of more oxidized seawater in the reduced basaltic rocks at high F/R ratios generates a relatively more oxidized alteration assemblage than at rock-buffered conditions, as illustrated by a redox potential ranging between −0.25 and 0 V (Figure 3). These more oxidized conditions are in agreement with the maximum stability of hematite.

4.5. Fluid Chemistry of Mid-Ocean Ridge Hydrothermal Vents

A compilation of 123 fluid compositions collected from submarine hydrothermal vents in various mid-ocean ridge settings (Cayman Spreading Center, Central Indian Ridge, East Pacific Rise, and Mid-Atlantic Ridge) and analogous submarine basalt-dominated environments (East Scotia Ridge, New Hebrides, and Western Manus Basin) is provided in Table 2. Natural vent fluid compositions are compared to the simulated fluid compositions for Mg, Ca, Na, and K in Figure 5. Vent concentrations in Mg present an excellent agreement with the model at all temperatures, while vent concentrations in Ca present a poor agreement at low temperature and an excellent agreement at higher temperature, except above ~300°C. Both the vent concentrations in Na and K present an excellent agreement at all temperatures, except above ~300°C. Since natural hydrothermal fluids have been sampled from active vents on the seafloor, their compositions may not only reflect the result of fluid/rock interaction processes in the reaction zone alone, but of a combination of these processes with processes related to their subsequent rise along the volcanic structures towards the seafloor environment. Kelley et al. [123] noted that submarine hydrothermal fluids showed relative enrichments and depletions of NaCl with respect to seawater and suggested that these variations may represent phase separation processes of hydrothermal fluids enriched in NaCl. The fluid compositions selected in our vent data compilation also present a wide range of concentrations in Cl, from 31 to 929 mmol/kg (Table 2), in comparison to the ambient seawater value of 546.7 mmol/kg. A classification of the fluids by their chlorinity shows the presence of an excellent correlation between total dissolved Cl and Ca-Na-K-(Mg) concentrations in the fluids. This suggests that secondary processes of phase separation have affected the cation composition of those hydrothermal fluids, an observation reinforced by the more pronounced dispersion of the cation concentrations above 300°C, where these processes are more likely to occur [131].

A number of natural fluids show an apparent enrichment/depletion in Mg-Ca-Na-K associated to an enrichment/depletion in chlorinity (Cl > 567 mmol/kg and <527 mmol/kg, respectively). The model of this study therefore presents some limitations that hamper any further interpretation for these specific fluids subject to fluid phase separation processes. However, a limited number of natural fluids have conserved their original seawater chlorinity (ClSw, 2°C = 546.7 mmol/kg ± 20 mmol/kg) at temperatures up to ~370°C (Figure 5), suggesting that they have not experienced phase separation processes during their ascent. At temperatures > 250°C, the cation concentrations of these fluids deviate systematically from the original seawater concentration. Chlorine was documented to maintain a conservative behavior in hydrothermal fluids up to temperatures of ~400°C, at which point the precipitation and dissolution of chloride-bearing minerals begin [132]. We therefore interpret that the deviations observed for dissolved Mg-Ca-Na-K concentrations at temperatures > 250°C and at conserved fluid chlorinity result from fluid/rock interaction processes that can be predicted using the model. The concentrations of these four elements show systematic deviations towards high temperatures, where Mg and Na are depleted while Ca and K are enriched in the fluid (dashed red arrows; Figure 5). In all cases, the cation concentrations tend towards the rock-buffered conditions predicted by the model. This evolution of the fluid composition towards rock-buffered conditions at high temperature is in agreement with the more limited circulation of seawater (i.e., low porosity and permeability) near the cracking front of the subaxial magma chamber, as documented by petrographic studies [9].

The systematic compositional evolution of submarine hydrothermal fluids towards low F/R ratios at high temperature can be synthesized in a four-cation diagram (Figure 6) using ionic pairs typified by opposite behaviors in the fluid (i.e., Mg-Ca and K-Na), as previously proposed by Giggenbach [133]. The diagram shows the predicted changes of seawater compositions associated to fluid/rock equilibria with basaltic rocks from seawater-buffered conditions towards the deepest and rock-buffered parts of the reaction zone. Instead of pointing at the composition corresponding to dissolved mid-ocean ridge basalts, the rock-buffered fluid end-member compositions are characterized by Mg-Na depletions and Ca-K enrichment characteristic of the overall fluid/rock interaction process predicted by the model (Figure 6). At the highest observed vent temperatures of ~400°C, fluid/rock interaction processes occurring at an F/R ≈ 1 are interpreted to constitute the end-member composition of submarine hydrothermal fluids in mid-ocean ridge systems.

5. Conclusions

Hydrothermal processes at mid-ocean ridges are characterized by the equilibration of seawater with mafic rocks at temperatures up to 400°C, leading to high temperature fluids of distinct compositions and the conversion of the fresh rocks into secondary mineral assemblages.

The existence of similar cation to proton activity ratios between the natural Icelandic waters and the simulations indicates that the model is able to closely reproduce the chemistry of fluid/rock interaction processes occurring in natural open systems. Although the Icelandic geothermal environment and submarine hydrothermal environment differ in fluid composition and rock permeability, the similar ionic exchange patterns occurring during fluid/rock interaction can be used to delineate the fundamental controls on the chemistry of natural fluids at mid-ocean ridges. Along with temperature, the F/R ratio prevailing during seawater-basalt interaction has a fundamental control on the resulting fluid chemistry. Rock-buffered conditions occur towards low F/R ratios and are characterized by relatively high aCa2+/(aH+)2 and aNa+/aH+ values. A complex secondary mineral assemblage forms at these conditions, with numerous Ca- and Na-bearing phases, and the relatively alkaline nature of the solution is balanced by the nearly entire precipitation of Mg. Hydrothermal alteration taking place under rock-buffered conditions is dominated by fresh basalt constituents and the effect of seawater is largely catalytic [85, 86]. Seawater-buffered conditions occur towards high F/R ratios and generate relatively high aMg2+/(aH+)2 and low pH values. At these conditions, hydrothermal alteration is more influenced by seawater components than by the rock components. The mineral assemblage formed in seawater-dominated conditions consists of simple Al-Si- and Mg-bearing phases and a relatively acidic solution characterized by low Ca concentrations while Mg approaches the initial value of ambient seawater.

The excellent agreement obtained between the model and the compositions of a compilation of natural vent fluids from mid-ocean ridges suggests that fluid/rock interaction processes in the oceanic crust control both the chemistry of the fluids and the composition of the subseafloor alteration minerals. The vent chemistry is controlled by fluid/rock interaction processes during seawater circulation through the oceanic crust as well as by the subsequent processes of phase separation, as indicated by a wide range of fluid chlorinity. Only a small number of natural vent fluids have preserved their original fluid/rock interaction compositions without undergoing subsequent fluid phase separation. The study of these selected data reveals that at a near-maximum temperature of ~400°C, the fluid composition is controlled by rock-buffered compositions marked by systematic Mg-Na depletions and Ca-K enrichments. This compositional evolution of the fluid towards rock-buffered conditions at high temperature confirms the more limited circulation of seawater near the cracking front of the subaxial magma chamber previously indicated by petrographic studies [9]. At the highest observed vent temperatures of ~400°C, fluid/rock interaction processes occurring at an F/R ≈ 1 are interpreted to constitute the end-member composition of submarine hydrothermal fluids in mid-ocean ridge systems. This study provides a verifiable model for the compositional evolution of submarine hydrothermal fluids at mid-ocean ridge and analogous basalt-dominated environments and constitutes a fundamental framework for future research on the controls on mass transfer associated with submarine hydrothermal systems.

Data Availability

The data used to support the findings of this study are included within the article. Thermodynamic data used for the minerals and aqueous species are archived as part of the MINES 17 database available at http://tdb.mines.edu.

Conflicts of Interest

The authors declare that there are no conflicts of interest regarding the publication of this paper.