Characterization of hyperalkaline fluids produced by low-temperature serpentinization of mantle peridotites in the Oman and Ligurian ophiolites


  • Valérie Chavagnac,

    1. GET UMR5563 (CNRS/UPS/IRD/CNES), Université de Toulouse, Observatoire Midi-Pyrénées, Géosciences Environnement Toulouse, Toulouse, France
    Search for more papers by this author
  • Christophe Monnin,

    Corresponding author
    1. GET UMR5563 (CNRS/UPS/IRD/CNES), Université de Toulouse, Observatoire Midi-Pyrénées, Géosciences Environnement Toulouse, Toulouse, France
    Search for more papers by this author
  • Georges Ceuleneer,

    1. GET UMR5563 (CNRS/UPS/IRD/CNES), Université de Toulouse, Observatoire Midi-Pyrénées, Géosciences Environnement Toulouse, Toulouse, France
    Search for more papers by this author
  • Cédric Boulart,

    1. GET UMR5563 (CNRS/UPS/IRD/CNES), Université de Toulouse, Observatoire Midi-Pyrénées, Géosciences Environnement Toulouse, Toulouse, France
    2. Now at Leibniz Institute for Baltic Sea Research, Marine Chemistry Department, Warneünde, Seestrasse 15, DE-18119, Seestrasse, Rostock, Germany
    Search for more papers by this author
  • Guilhem Hoareau

    1. GET UMR5563 (CNRS/UPS/IRD/CNES), Université de Toulouse, Observatoire Midi-Pyrénées, Géosciences Environnement Toulouse, Toulouse, France
    2. Now at Université de Pau et des Pays de l'Adour, Laboratoire des Fluides Complexes et leurs Reservoirs, UMR CNRS TOTAL 5150, I.P.R.A. Avenue de l'Université, BP 1155, FR-64013 PAU, Cedex, France
    Search for more papers by this author


[1] A regional survey of alkaline springs in Oman and Ligurian ophiolites shows that the alkaline water compositions significantly vary from one ophiolite to the other and within the same ophiolite. The first-order correlation between the Na (and K) and Cl concentrations points to fluid compositions only partly due to evaporation. The scatter around the evaporation line implies that Na and Cl may not be conservative during the alteration of the ultramafic rocks. Mg is almost entirely depleted at pH > 10.5 as a result of serpentine formation within the ultramafic body and of brucite (and minor hydrotalcite) precipitation at the springs. Ca accumulates in the high-pH fluids and is consumed by Ca-carbonate formation at the springs, by mixing with river waters or by the CO2 supply from the atmosphere. Thermodynamic calculations show that brucite saturation is reached at pH values around 10.5 which triggers major changes in the water composition. The waters evolve from a quartz-saturated low-pH continental environment to a brucite-dominated high-pH serpentinizing system at low temperature. The highest water salinities are found in springs located along the basal thrust plane of the ophiolite. The highest Al concentrations are found in some springs located on the crustal side of the mantle/crust boundary. This poses the question of the hydrologic pathways and of the role of the mineralogical composition of the altered formations.

1. Introduction

[3] Serpentinization is the alteration process that transforms olivine and pyroxene-rich rocks (essentially peridotites, pyroxenites, but also troctolitic gabbros and picritic basalts) into serpentinites. Different sequences of chemical reactions have been proposed to describe the serpentinization process [e.g., Bach et al., 2006; Frost and Beard, 2007; Kelemen and Matter, 2008; Klein et al., 2009; Palandri and Reed, 2004; Putnis and Fernandez-Diaz, 2010; Sleep et al., 2004] according to the temperature ranges considered [Evans, 2010]. This complex process covers several types of simple reactions.

[4] The first one is silicate dissolution in which H+(aq) or H3O+(aq) ions play a central role [Pokrovsky and Schott, 2000]. In the case of forsterite and enstatite (the Mg end members of the olivine and of the ortho pyroxene solid solutions, respectively), global dissolution reactions can be written as follows:

display math(1)
display math(2)

[5] The left to right single arrows indicate that these silicates that are the main constituents of mantle peridotites or that crystallize from mantle-derived melts during the formation of the oceanic crust can be dissolved but cannot precipitate from an aqueous solution in the low temperature range considered in this study. At ambient pressure and temperature, these reactions also express the main mechanism of low-temperature chemical weathering of ultramafic rocks exposed on continental surfaces. They are similar to those prevailing during the alteration of the continental crust through the dissolution of Ca-bearing silicates. Such reactions consume H+(aq) which results in a pH increase from rainwater-type values to the more alkaline pH values of river waters.

[6] At adequate T and P conditions, reactions (1) and (2) are followed by serpentine crystallization:

display math(3)

[7] In this case equilibrium between serpentine and the aqueous fluid can be achieved and is indicated by the double arrow in reaction (3).

[8] The addition of reactions (1) and (3) leads to a chemical reaction expressing the alteration of olivine producing serpentine as a secondary mineral:

display math(4)

[9] Reaction (4) shows that water itself is a reactant in the process of serpentine production. Such reactions are sometimes termed silicate hydration. This term can be judged inappropriate from the point of view of reaction mechanism in the sense that hydration would mean that the structure of the water molecule is preserved in the process. In fact silicate alteration consumes H+(aq) (with the consequence of OH(aq) production by water dissociation). Reaction (4) also shows that aqueous silica is consumed. Silica activity is, accordingly, a key parameter in the control of the serpentinization process [Evans, 2008; Frost and Beard, 2007; Katayama et al., 2010].

[10] Water reacts in a second manner as it oxidizes metals contained in the Mg-silicates. For example ferrous iron is oxidized to ferric iron:

display math(5)

[11] Reaction (5) is responsible for the simultaneous production of hydrogen and of the aqueous hydroxyl ion due to the reduction of water during the alteration of ultramafic rocks. It is written here with aqueous species as reactants and products but it is most of the time written in terms of oxide components of minerals [see e.g., Seyfried et al., 2007; Palandri and Reed, 2004]. This OH(aq) production supplements the increase in pH due to the consumption of H+(aq) due to silicate hydrolysis (reactions (1) and(2)) which cannot be the sole cause of the very high-pH values (as proposed by Pfeifer [1977]).

[12] Natural H2 generation during serpentinization is given a lot of interest in regard to energy resources and has been documented at various locations at the seafloor [Charlou et al., 2002; Keir et al., 2008; Konn et al., 2009; Nakamura et al., 2009] and in ophiolites [Abrajano et al., 1988; Katayama et al., 2010; Neal and Stanger, 1983]. Such a redox reaction also produces a very reductive environment that can lead to the formation of metallic alloys such as awaruite Ni3Fe or wairauite CoFe [Klein and Bach, 2009; McCollom and Bach, 2009].

[13] Besides the main reactions (4) and(5), a series of reactions produces secondary minerals, by products of the serpentinization process. Reactions (1) and (5) produce aqueous magnesium and aqueous hydroxyl that may form brucite (solid magnesium hydroxide) the dissolution reaction of which is:

display math(6)

[14] When formed during serpentinization of oceanic lithosphere at hydrothermal conditions brucite contains important quantities of Fe [e.g., Bach et al., 2006; McCollom and Bach, 2009; Sleep et al., 2004] while it is very pure when it forms in low-temperature subsurface environments [Neal and Stanger, 1984a]. Portlandite may form naturally at low temperature in such environments, but is much less common [Neal and Stanger, 1984a]. Reaction (5) leads to the formation of magnetite Fe3O4, which, as brucite, is commonly found in serpentinites. Its distribution in serpentinites and its impact on the natural production of hydrogen is the subject of current investigation [Bach et al., 2006; Evans, 2008; Klein and Bach, 2009; Klein et al., 2009].

[15] High-pH fluids produced by serpentinization have been sampled at divergent and convergent geological settings. Among the six ultrabasic-hosted hydrothermal systems located along the Mid-Atlantic ridge, the lost city hydrothermal field (LCHF) [Kelley et al., 2001] has become the reference example of a low-temperature serpentinizing environment. It is located near the top of a corner high massif, at a depth of 800 m ∼10 Km off the mid-Atlantic ridge and ∼10 km north from the Atlantis Fracture Zone, at 30° N. It produces alkaline fluids in the 9–11 pH range with seawater-type salinity. Yet only a single composition of the fluids produced at this site (with a pH between 9 and 9.8) has been reported so far [Kelley et al., 2001]. When these alkaline waters discharge into seawater at the ocean floor, they form hydrothermal chimneys mainly composed of calcium carbonate minerals [Jones et al., 2010; Lein et al., 2007a, 2007b; Ludwig et al., 2006] in contrast to high-temperature acidic hydrothermal vents where sulfide and sulfate are the main minerals to precipitate [Halbach et al., 2003; Marques et al., 2006] in a convergent geological setting, the dehydration of the subducting plate in the Marianna forearc produces some of the highest pH fluids reported on Earth (i.e., 12.5) [Mottl et al., 2004; Wheat et al., 2008; Mottl, 2009, Highest pH? Available at gn141oct09/highestph.htm] ascending through the sediment pile of mud volcanoes and discharging at the seafloor. Such extreme pH values have also been measured in fluids collected in boreholes drilled in kimberlite formations of Northern Ontario [Sader et al., 2007]. On land hyperalkaline springs can be found in sections of the oceanic lithosphere (ophiolites) exposed by obduction and erosion in Oman [Neal and Stanger, 1984a, 1984b], in the Philippines [Abrajano et al., 1988], in New Caledonia [Launay and Fontes, 1985], in Northern Italy [Boschetti and Toscani, 2008; Cipolli et al., 2004], in Portugal [Marques et al., 2008], in British Columbia [Power et al., 2007] and in Cyprus [Neal and Shand, 2002] among other locations. Apart from these well documented locations, such hyperalkaline fluids are presently not very common on the Earth's surface. At the seafloor the number and occurrence of Lost City-type hydrothermal systems are still to be evaluated. They may have been more numerous in the primitive Earth where serpentinization may have provided the adequate conditions for the emergence of life [Muntener, 2010; Russell, 2007; Russell et al., 2010; Shibuya et al., 2010; Sleep et al., 2012]. Such hyperalkaline fluids are also currently given attention in light of the on-going focus on the geological storage of atmospheric carbon dioxide in mafic-ultramafic rocks [Cipolli et al., 2004; Matter and Kelemen, 2009]. Indeed, Kelemen and Matter [2008] proposed that the carbonation of the sole Oman ophiolite could be a low-cost, safe, and efficient process to capture atmospheric CO2 in proportions meaningful in comparison to its atmospheric content.

[16] Whatever the geological environment in which they are located the characterization of such high-pH fluids is hampered by mixing with what can be called local waters: ocean bottom seawater for the Lost City fluids, pore waters modified from seawater by diagenetic processes for the Mariana mud volcanoes, formation waters flowing through glacial sediments for the North Ontario kimberlite intrusion, mixing with continental runoff for ophiolitic hyperalkaline springs. As such the characterization of the pristine fluids produced by serpentinization first requires the evaluation of their contamination by local waters. The examples above show that the fluids produced in various serpentinizing environments do share common characteristics very different from those produced by the hydrothermal circulation in basalts. The conditions encountered in these various environments are nevertheless variable and require a detailed investigation in order to constrain the relationships between hyperalkaline fluids formation and the mechanisms of serpentinization occurring in the subsurface.

[17] The chemical erosion of aerial ultramafic rocks (ophiolites, orogenic peridotite massifs) by rainwater (meteoric water) produces continental waters of the Mg-HCO3 type, as shown by extensive hydrological studies in Oman [Stanger, 1985; Weyhenmeyer, 2000] and in the Ligurian province of Northern Italy [Cipolli et al., 2004; Marini and Ottonello, 2002]. When these waters infiltrate into the ultramafic formation and react at depth with (Mg, Fe) silicates, they undergo dramatic chemical changes due to the serpentinization reactions as mentioned above. This circulation leads to the discharge at the surface of high-pH waters of the Ca-OH type, which have been almost entirely stripped of their dissolved inorganic carbon (DIC) [Sader et al., 2007]. At their contact with the atmosphere, these high-pH waters are able to capture atmospheric CO2 and precipitate carbonate minerals.

[18] These surface phenomena may be viewed as undue processes blurring the “primary” chemical composition of the fluids directly linked to the serpentinization process, in spite of their intrinsic scientific interest. Their role must be evaluated in order to retrieve the compositions of the end-member fluids coming from the subsurface zone where serpentinization takes place. We investigate the surface phenomena (mixing with continental waters, evaporation, uptake of atmospheric CO2, etc.) in order to understand which elements are trapped in or exported out of ultramafic environments by serpentinizing fluids, following the early proposition that “serpentinization occurs at nearly constant composition, except for loss of CaO” [Barnes and O'Neil, 1969]. Additional detailed results on the mineralogical assemblages of the precipitates forming in these springs and on the gases emitted therein [Monnin et al., 2011] will be given elsewhere. The present study supplements those of Neal and Stanger [1984a, 1984b] in Oman and those of Bruni et al. [2002], Cipolli et al. [2004], and Marini and Ottonello [2002] in the Ligurian Alps.

2. Geological Environments and Sampling Locations

2.1. Generalities

[19] The previous works targeting alkaline springs in Oman and in Liguria are parts of larger studies aiming at the evaluation of water resources [Marini and Ottonello, 2002; Stanger, 1985]. Such efforts have produced copious databases on the chemical composition of continental surface waters with the inventory of alkaline springs over the ophiolite belts as an outcome. For example, in his very extensive study [Stanger, 1985] reported the composition of 486 water samples collected in Oman, along springs, wadis (streams), falajs (irrigation channels) and wells, out of which 78 had a pH above 10 (a criterion that we will here retain to designate these waters as “hyperalkaline”). In the ophiolite of the Voltri group in Liguria, Marini and Ottonello [2002] report the chemical composition of 589 water samples collected at springs and streams in the catchments, out of which only 25 (collected at 19 different sites) have pH values above 10. In the present work, we have used the water chemical compositions given by Stanger [1985] and Marini and Ottonello [2002] of samples having a pH above 9, along with our own data (Table 1).

Table 1. Coordinates of the Hyperalkaline Springs Collected in the Omani and Liguria Ophiolites
SiteNameType of DischargeLatitude(N)Longitude (E)Altitude (m)Geological Context
Ophiolite of the Sultanate of Oman
1NidabOpenair water pumping station23°12'74”58°08'22”460Bedded gabbros
2Nidab (a few hundred meters from site 1)Openair water pumping station23°12'57"58°-8'25"473Gabbros
3Yellowstone du pauvreHyperalkaline fluid discharge22°49'10"57°48'48"479In serpentinite, just above the contact with Hawasinah
5.6Little grand cañonHyperalkaline fluid discharge22°50'49"58°03'22"533In serpentinite, just above the contact with Hawasinah
8Izki les 2 puitsOpenair well22°55'40"57°46'20"530In serpentinite, just above the contact with Hawasinah
10Le partage du midiHyperalkaline fluid discharge22°52'26"57°31'12"472In serpentinite, just above the contact with Hawasinah
11Lac bleu de BahlaHyperalkaline fluid discharge22°59'28"57°17'34"562Serpentinite
12Ain Al-WaddahHyperalkaline fluid discharge22°59'28"57°17'34"562Contact peridotite/gabbro, MOHO
13Gaz pas chaudHyperalkaline fluid discharge23°24'20"56°51'46"745Transition peridotite/gabbro, MOHO
14Les enfants d'AlqarHyperalkaline fluid discharge23°58'52"56°24'36"716Peridotite
15Piscine bleu d'AlqarHyperalkaline fluid discharge23°58'09"56°25'17"652Peridotite
16Le pique-nique de L'ascèteHyperalkaline fluid discharge23°58'13"56°25'31"657Peridotite
17Le pique-nique de L'ascèteHyperalkaline fluid discharge23°58'12"56°25'35"/Peridotite
18CuiCuiHyperalkaline fluid discharge23°57'31"56°26'35"550Peridotite
19Le petit guerrier des ruinesHyperalkaline fluid discharge23°43'10"57°01'52"469Bedded gabbros
20Le salaire de la peurHyperalkaline fluid discharge23°37'39"57°06'55"322Bedded gabbro (several 100m above peridotite)
21RustaqWell in town23°23'36"57°24'41"373In serpentinite, just above the contact with Hawasinah
22La poule au potFalaj23°34'19"58°01'38"94In serpentinite, just above the contact with Hawasinah
234×4 practiceHyperalkaline fluid discharge23°33'49"58°06'44"140Bedded gabbros
24Gabbro litéHyperalkaline fluid discharge23°33'30"58°06'25"80Bedded gabbros
25L'ane blancWater pumping station23°28'12"58°19'26"/In serpentinite, just above the contact with Hawasinah
26La grande LigurieHyperalkaline fluid discharge23°37'12"57°06'48"337In peridotite, 100m above the MOHO
27La ligne bleueHyperalkaline fluid discharge24°42'05"56°16'15"309Within the crustal section, wherlite/gabbro
27bisLa ligne bleueHyperalkaline fluid discharge24°42'00"56°16'11"318Within the crustal section, wherlite/gabbro
28Mamy NovaHyperalkaline fluid discharge24°31'45"56°17'07"395Within the crustal section, bedded gabbro/ultrabasic cumulates
29Two shoesHyperalkaline fluid discharge23°57'43"56°26'26"625In serpentinites, gabbro 100m above
30Rencontre de la colombeHyperalkaline fluid discharge22°54'22"58°25'35"660Peridotite
31Les lauriers rosesHyperalkaline fluid discharge22°53'44"58°23'41"689Peridotite near MOHO
32graviereHyperalkaline fluid discharge23°19'21"58°13'42"367Mixing peridotite-gabbro, very altered
33Irma (Yellowstone du pauvre)Hyperalkaline fluid discharge22°48'43"57°50'16"485In serpentinite, just above the contact with Hawasinah
33bisIrmaHyperalkaline fluid discharge22°48'52"57°50'12"485In serpentinite, just above the contact with Hawasinah
34l'abreuvoir de la belle chamelleHyperalkaline fluid discharge22°37'36"58°42'40"386Peridotite
Voltri Massif of the Liguria OphioliteX UTM (WGS 84)Y UTM (WGS 84)  
C11Fiorino villageHyperalkaline fluid discharge4757684924028302Beigua unit (serpentinised metagabbros)
L43AcquasantaHyperalkaline fluid discharge4816224922794160Beigua unit (serpentinised metagabbros)
LER2Porte ArmaHyperalkaline fluid discharge4696614917003189Beigua unit (serpentinised metagabbros)
NG1Rio LeoneHyperalkaline fluid discharge4727674918911135Beigua unit (serpentinised metagabbros)
LER21Rio LeoneHyperalkaline fluid discharge4728194919057207Beigua unit (serpentinised metagabbros)
BR1Rio BranegaHyperalkaline fluid discharge482416492137677Beigua unit (serpentinised metagabbros)
GOR35Gorzente (lago Lavagnina)Hyperalkaline fluid discharge4830524938156392Erro-Tobio unit (serpentinised lherzolites)
GOR34GorzenteHyperalkaline fluid discharge4827914938136363Erro-Tobio unit (serpentinised lherzolites)
ERR20Maddalena (Don Orione)Hyperalkaline fluid discharge4601584928508373Beigua unit (serpentinised metagabbros)

2.2. Oman Ophiolite

[20] We have conducted two main field trips in Oman (December 2008 and January 2011) during which we sampled waters, mineral precipitates and gas emissions at 34 different locations (Table 1), most of them reported in a database kindly provided by the Ministry of Water Resources of Oman. A few springs have been also sampled again in January 2010. The geological map of the Omani ophiolite and the location of the sampled hyperalkaline springs are shown in Figure 1. Our study, as well as that of Stanger [1985] that was based on a significantly different data base, covers the entire Oman ophiolite (Figure 1), from the interior of the mantle section and of the lower crustal section, to its contact with the metamorphic sole of the ophiolite (metabasalt and metasediments), with the underlying non metamorphosed sedimentary formations (Hawasina nappes) and with the autochthonous limestones. A description of the different geological units of the Oman mountains can be found in Abily and Ceuleneer [2013], Béchennec et al. [1989], Combe et al. [2006], Glennie et al. [1973], Lippard et al. [1986], and Python and Ceuleneer [2003].

Figure 1.

Simplified geological map of the Oman ophiolite (Sultanate of Oman). Sampling locations are reported.

[21] It appears that the alkaline springs are quite numerous and can be found almost all along the Oman mountain range. At many locations the discharge of alkaline waters forms water pools of an intense blue color while snow-like precipitates that covers the bottom of the pools and wadis (Figure 2). Away from the discharge but still within the wadi bed, the color of the precipitates turns progressively to a yellow-brown color, forming a dense deposit of carbonate (travertine).

Figure 2.

Photo of the hyperalkaline springs at site 27 “La Ligne bleue” showing the snow-like precipitate forming in the area where the wadi (river) waters and the hyperalkaline waters mix. Note the calcium carbonate precipitate at the water surface due to atmospheric CO2 uptake.

[22] Hyperalkaline springs occur preferentially, but not only, along two major structural discontinuities: the mantle/crust transition zone (so called “Moho”) within the ophiolite and the basal thrust plane of the ophiolite. Some springs are located away from these discontinuities, either within the mantle peridotites of within the lower crustal section (Figure 1). Overall, hyperalkaline springs are found either at the edge of carbonate terraces or within the wadi beds. In the first case, because of their situation above the wadi beds, these waters are preserved from mixing with runoff wadi waters. They are often channelled to pools feeding an irrigation system (falaj) that has been in use for centuries.

[23] Gas emission is clearly evidenced by bubbling within the wadi bed as well as by pockmark structures within the snow-like precipitates. Gas emission is commonly observed at the springs but probably also occurs directly from the barre bedrock surface, which can only be detected with a gas detector.

2.3. Voltri Group of the Liguria Ophiolite

[24] Figure 3 displays the geological map of the Voltri Group and the location of the hyperalkaline springs. The Voltri Group represents the remnants of the Liguro-Piedmont oceanic domain, which was subducted and exhumed during the Cretaceous-Tertiary Alpine convergence [Vanossi et al., 1984]). The Voltri Group is composed of three mains Units: (1) the Beigua Unit (metagabbros and metabasalts essentially), (2) The Voltri-Rossiglione Unit (metasediments and metavolcanites), and (3) the Erro-Tobbio Unit (partly serpentinized lherzolites). A field study conducted in Liguria in June 2010 allowed us to sample 10 springs (Table 2) out of the 19 studied by Marini and Ottonello [2002] and Cipolli et al. [2004]. Most of the springs (if not all) are in the river beds just above the water level. Two of them (GOR34 and GOR35) are located within the Erro-Tobbio Unit whereas all the others are within the upper section of the metamorphosed oceanic crustal section (the Beigua Unit). At each site, the flow is often harnessed with some tubing which allows the collection of the alkaline fluid not diluted by the river waters. The flow rate of the Ligurian spring is so limited that seepage better describes their mode of occurrence. The alkaline waters flow into perennial mountain rivers and streams where they are right away diluted. The mixing zone is so narrow that it cannot be sampled. Among the 589 water compositions listed by Marini and Ottonello [2002] a sole datum in the 9–10 pH range (LER20 with a pH of 9.45) is reported, which again stresses the fact that in Liguria, the alkaline waters are readily diluted in runoff waters. The alkaline water discharges can be detected by a brownish precipitate that covers the bedrock. There are no blue pool and snow-like white precipitates as those found in Oman. The formation of carbonate concretions is restricted to the close vicinity of the springs. There are no extensive carbonate terraces like those found in Oman. At some places conglomerates can be observed, but also never to a large extent [Marini and Ottonello, 2002]. In addition, the vegetation coverage in Liguria is dense compared to the barren landscapes of Oman. Finally, a white bacterial mat was observed only at the discharge area outside the tube at C11.

Figure 3.

Simplified geological map of the Liguria ophiolite displaying the sampling locations visited during this work.

Table 2. Chemical Composition of the Liguria Hyperalkaline Springs
Site NumberSite NameT (°C)pHDIC (ppm)F (µmol/l)Cl (mmol/l)SO4 (µmol/l)NO3 (µmol/l)Mg (mmol/l)Sr (µmol/l)Ca (mmol/l)K (mmol/l)Na (mmol/l)SiO2(aq) (µmol/l)
C11Fiorino village14.
C11Fiorino village14.
C11-riverRio Dellecave16.
LER2Ponte Arma17.
NG1Rio Leone19.
NG1Rio Leone19.
LER21Rio Leone17.511.<0,000113.60.720.020.434.9
LER21Rio Leone17.511.<0,000212.80.710.020.435.0
BR1Rio Branega17.311.
BR1Rio Branega17.311.
GOR35Gorzente (lago Lavagnina)21.411.
GOR35Gorzente (lago Lavagnina)21.411.
GOR35-RiverGorzente (lago Lavagnina)n.m.n.m.
GOR35Gorzente (lago Lavagnina)20.0511.
GOR35Gorzente (lago Lavagnina)24.711.21.451.50.311.20.30.00313.40.700.080.4113.8
ERR20Maddalena (Don Orione)23.711.
ERR20Maddalena (Don Orione)23.711.

[25] Compared to Oman, we observed and sampled gas emission at only three locations: either within the river bed (site L43) close to the spring discharge, or directly in the spring within the river bed (GOR34 and GOR35).

3. Sampling Methods and Analysis

3.1. Sampling Methods

[26] In Liguria, the fluid was most of the time sampled directly at the tubing-channeled source (Figure 4). In contrast, the water sample collection has been more difficult in the Oman ophiolite as hyperalkaline springs discharge into the wadi waters. Water samples were thus collected where the pH and temperature were the highest. In most cases, we avoided sampling in stagnant water pools to avoid biases linked to evaporation and anthropic contamination.

Figure 4.

Photo of spring BR1 in Liguria showing the channelled discharge of the hyperalkaline water above the river bed. The formation of carbonate precipitate is limited to vicinity of the springs.

[27] Before sampling, a number of chemical parameters were measured directly in the hyperalkaline springs using a set of electrodes (temperature-corrected pH, salinity, conductivity, total-dissolved solids). Water samples were collected using a 60 ml “sterile” syringe and filtered through a Millipore filter (Millex-HA, mixed-esters of cellulose membrane), which removes microorganisms, particles, and precipitates larger than 0.45 µm. The hyperalkaline waters were stored in a LDPE plastic bottle for further chemical analyses in the laboratory.

3.2. Analytical Techniques

[28] All the chemical analyses on water, mineral and gas were carried out at the laboratory “geosciences environnement toulouse” using state of the art analytical facilities.

[29] All hyperalkaline fluid samples were clear, void of visible particles, and not acidified during storage. Alkalinity has been measured in the laboratory by a Gran titration using a Schott Titroline alpha plus instrument a few months after collection. We noticed a small drop in the water pH values (by at most 0.5 pH units) between the time of collection and that of alkalinity titration. DIC was measured using a Shimadzu TOC-VCSN instrument. Major and trace element concentrations (Si, Ca, Na, K, Mg, Sr) were measured on all collected samples by inductively coupled plasma optical emission spectrometry (Horiba Jobin Yvon Ultima 2). The instrument was calibrated using synthetic standards and achieved a precision of 2% or better. The full set of standards was run before and after each group of analyses to check the performance of the instrument. In addition, the drift was checked by running one standard as a sample before, during and after each group of analyses. All the concentrations reported in Tables 1 and 2 are, therefore, drift and blank corrected. Anion concentrations (Cl, SO42−, F) were measured by ion chromatography (Dionex ICS 2000) which was calibrated using synthetic standards with a precision of 2% or better. The chemical compositions of hyperalkaline springs are reported in Table 1 for Oman and in Table 2 for the Voltri Group.

3.3. Thermodynamic Calculations of Water Rock Interactions in Hyperalkaline Systems

[30] The interplay between temperature, water composition and mineral formation can be addressed through the calculation of the mineral saturation indices. Following the earlier work of Pfeifer [1977] for a variety of alkaline waters, such calculations are reported by Bruni et al. [2002] and Marini and Ottonello [2002] for the Ligurian waters, by Neal and Shand [2002] for Cyprus, by Marques et al. [2008] for the waters of Cabeço de Vide (central Portugal), and very recently by Paukert et al. [2012] for Oman. An alternative approach to the thermodynamic state of the water-rock system consists in simulating the alteration of a given set of minerals to calculate the evolution of the system as a function of temperature, water rock ratio, the nature of the starting mineral assemblage, the initial composition of the percolating fluid, etc. The results are then compared to the composition of waters sampled in the field using a variety of phase diagrams [Marques et al., 2008; Cipolli et al., 2004; Boschetti and Toscani, 2008; Sader et al., 2007]. Such simulations of the serpentinization process at hydrothermal temperatures are also reported by McCollom and Bach [2009] and Palandri and Reed [2004] who predicted the evolution of the mineralogical assemblages and of the composition of the fluids. These calculations reveal major trends such as the role of iron and the mobility of various elements in the serpentinization process. Other constraints on the evolution of serpentinizing systems come from the experimental alteration of a spinel lherzolite by an artificial seawater at 200°C and 500 bars [Seyfried et al., 2007]. In the present work, we calculated the mineral saturation indices with the PHREEQC code [Parkhurst and Appelo, 1999] using the composition data reported in Tables 2 and 3. Because alkalinity is dominated by the aqueous hydroxide ion at high pH (see below), we have used the DIC as an input parameter for the carbonate system.

Table 3. Chemical Composition of the Oman Hyperalkaline Springs
Site NumberSite NameTemperature (°C)pHConductivity (ms/cm)Alkalinity (mol/l)TDS (mg/l)TOC (ppm)DIC (ppm)F (µmol/l)Cl (mmol/l)SO4 (mmol/l)NO3 (mmol/l)Al (µmol/l)SiO2(aq) (mmol/l)Mg (mmol/l)Sr (μmol/l)Ca (mmol/l)K (mmol/l)Na (mmol/l)
3Yellowstone du pauvre20.810.12.702.7E-031332n.m.n.m.2.121.0520.3490.019n.m.0.0030.125n.m.0.0810.93522.086
3Yellowstone du pauvre24.811.92.215.0E-03630n.m.0.446.49.8440.0150.000n.m.0.0060.001n.m.0.4500.36410.740
5Little Grand Cañon28.311.81.73n.m.453n.m.2.541.16.0910.0050.020n.m.0.0000.001n.m.2.0810.2156.261
6Little Grand Cañon32.211.71.694.5E-03435n.m.0.1619.75.7080.0010.017n.m.0.000b.d.l.n.m.2.1370.2245.939
8Izki les 2 puits23.47.80.812.7E-03436n.m.n.m.2.25.1360.0710.017n.m.0.1781.324n.m.0.5950.2237.647
8Izki les 2 puits23.111.20.921.7E-03340n.m.1.363.55.4640.0280.008n.m.0.0070.009n.m.0.2940.1855.373
10Le partage du midi22.411.72.255.1E-03707n.m.1.032.910.3010.0030.021n.m.0.0470.001n.m.1.2050.39311.965
11Lac bleu de Bahla26.310.80.789.0E-04353n.m.
11Lac bleu de Bahla31.211.0n.m.n.m.n.m.n.m.n.m.1.23.3520.2090.139n.m.0.0920.779n.m.0.1210.2164.543
11Lac bleu de Bahla34.912.0n.m.n.m.n.m.n.m.n.m.0.66.6720.0010.000n.m.0.000b.d.l.n.m.2.1290.3989.418
12Ain Al-Waddah33.311.71.924.8E-03521n.m.
13Gaz pas chaud28.411.71.985.9E-03450n.m.
13Gaz pas chaud28.
13Gaz pas chaud28.
15Piscine bleu d'Alqar31.
16Le pique-nique de L'ascète34.011.51.403.5E-03369n.m.0.412.05.2210.0030.012n.m.0.0080.001n.m.1.7730.1584.587
17Le pique-nique de L'ascète20.911.61.734.4E-03476n.m.
19Le petit guerrier des ruines36.711.31.012.6E-03273n.m.
20Le salaire de la peur23.99.10.68n.m.272n.m.n.m.1.73.0130.3700.021n.m.0.0001.627n.m.0.3650.1223.037
20Le salaire de la peur28.511.20.702.0E-034501.721.050.93.7240.1480.025b.d.l.0.0200.8562.20.0000.2025.263
20Le salaire de la peur28.511.20.701.9E-034501.500.930.83.7680.1380.024b.d.l.0.0180.7832.30.0000.2035.371
20Le salaire de la peur29.911.21.351.5E-036701.
20Le salaire de la peur29.911.21.351.6E-036701.480.050.84.3560.0810.012b.d.l.0.0370.0072.70.2430.2196.205
20Le salaire de la peur22.39.30.572.8E-033101.212.430.62.2360.2340.025b.d.l.0.1121.6402.70.0710.1092.921
22La poule au pot35.811.31.872.7E-03688n.m.1.021.411.0270.0450.009n.m.0.0000.031n.m.1.1760.29810.094
234×4 practice26.310.4n.m.n.m.1590n.m.n.m.1.721.8332.1490.072n.m.0.0000.142n.m.2.3850.39421.348
234×4 practice26.011.42.84n.m.1287n.m.n.m.2.718.2621.4870.027n.m.0.0000.004n.m.2.3190.35316.889
234×4 practice21.58.72.22n.m.1278n.m.n.m.4.911.5583.1320.372n.m.0.0006.972n.m.0.8170.21214.505
234×4 practice22.39.8n.m.n.m.1408n.m.n.m.3.116.9672.7060.231n.m.0.0002.057n.m.1.6530.40917.410
24Gabbro lité21.011.4n.m.n.m.1200n.m.1.2827.718.3130.9580.009n.m.0.0000.008n.m.1.1530.23917.469
25L'ane blanc65.57.91.363.3E-03773n.m.n.m.37.85.3272.4330.026n.m.0.6661.437n.m.3.3740.2726.446
26La grande Ligurie31.311.31.002.8E-036601.
26La grande Ligurie31.311.31.002.8E-036601.
27La ligne bleue21.99.50.582.0E-033001.620.080.52.1540.2340.0313.80.1651.4032.70.1340.0622.512
27La ligne bleue21.99.50.582.0E-033000.821.760.62.1310.2320.0303.80.1691.4032.70.1120.0612.487
27La ligne bleue31.211.10.592.0E-032801.
27La ligne bleue31.211.10.592.0E-032801.
27bisLa ligne bleue32.411.21.302.0E-035901.310.030.63.1550.0290.00922.60.059b.d.l.5.00.9250.0923.729
27bisLa ligne bleue32.411.21.302.0E-035901.
28Mamy Nova38.611.0n.m.2.5E-03n.m.
28Mamy Nova38.611.0n.m.2.5E-03n.m.1.380.120.43.2850.0050.00719.40.033b.d.l.4.91.1780.1003.748
28Mamy Nova29.510.20.481.4E-032501.410.880.72.3340.1680.0324.60.1181.1182.40.0470.0762.583
28Mamy Nova29.510.20.481.4E-032501.350.970.42.3290.1700.0324.90.1221.1332.40.0300.0792.581
29Two shoes23.811.61.483.6E-037801.370.050.44.5110.0400.008b.d.l.0.0030.0053.61.0080.1466.581
29Two shoes23.811.61.483.6E-037801.330.050.43.0090.0410.001b.d.l.0.0030.0043.61.0580.1486.691
30Rencontre de la colombe27.511.71.796.2E-038801.
30Rencontre de la colombe27.511.71.796.2E-038801.510.040.22.8690.0020.007b.d.l.0.019b.d.l.6.71.9600.1446.682
31les lauriers roses26.411.61.533.6E-038101.300.050.42.8360.0030.0072.60.006b.d.l.3.41.6280.1205.074
31les lauriers roses26.411.61.533.7E-038101.
31les lauriers roses23.49.80.572.7E-033102.
31les lauriers roses23.49.80.572.7E-033101.981.970.61.7860.1540.007b.d.l.0.2671.7702.20.0150.0562.635
31les lauriers roses28.511.0n.m.n.m.n.m.n.m.n.m.2.91.6830.2180.068b.d.l.0.2541.936n.m.0.1580.0042.061
33Irma (Yellowstone du pauvre)24.610.52.543.6E-039702.681.990.511.1540.0440.010b.d.l.0.0920.4472.2b.d.l.0.49818.311
33Irma (Yellowstone du pauvre)24.610.52.543.6E-039702.631.990.611.6320.0460.010b.d.l.0.0930.4522.3b.d.l.0.50618.579
33Irma (Yellowstone du pauvre)
33Irma (Yellowstone du pauvre)
34l'abreuvoir de la belle chamelle27.712.0n.m.n.m.n.m.n.m.n.m.4.810.4260.0100.001n.m.0.017b.d.l.n.m.2.2580.84216.002
34l'abreuvoir de la belle chamelle22.510.5n.m.n.m.n.m.n.m.n.m.1.511.2520.0200.002n.m.0.200b.d.l.n.m.0.8465.38999.612

[31] The choice of the data base is of course of primary importance. We have checked that the differences in the calculated saturation indices using different databases provided with PHREEQC are not generally significant in light of the scatter in the data. While this is true for most minerals, there are noticeable discrepancies between values of the brucite solubility product reported in the literature. According to authors [Harvie et al., 1984; Königsberger et al., 1999; Lambert and Clever, 1992; Palmer and Wesolowski, 1997; Wagman et al., 1982; Xiong, 2008], the brucite solubility product (the pK of reaction (6)) may vary over almost 1 order of magnitude (from 10.88 to 11.63). Xiong [2008] has carried out a thorough experimental study from which he derived new values of the thermodynamic properties of brucite that he critically compared to literature values and that we retained in our study. Brucite solubility decreases with temperature, so that its formation should be favoured in Oman. It decreases by a factor of about 2 when the temperature increases from 25 to 50°C [Lambert and Clever, 1992]. These latter authors note that the solubility of brucite depends on its physical state, an amorphous form called “labile” or “active” being slightly more soluble (by about 20%) than the “stable” or “inactive” form.

[32] Below, we report the calculated saturation indices of brucite, calcite and quartz in the sections where we discuss the concentration of the various elements.

4. Composition of Hyperalkaline Fluids of Oman and Liguria

4.1. pH

[33] We have seen that, in Liguria, it is possible to collect the hyperalkaline waters directly at the outlet and avoid dilution by river waters. Consequently most of the samples that we collected have pH values above 9.5 (site C11) and up to 11.7 (site GOR 34; Table 2).

[34] In Oman, the waters display pH values ranging from those of the wadi waters (pH around 7 and 8) to those of the hyperalkaline springs (the maximum pH value that we measured is 12.17 at site 17 “Pique-nique de l'ascète”). In Stanger's extensive database, only 7% of the studied water samples, collected mostly in falajs and wells, have pH values between 9 and 10. When the springs are located directly in the wadi beds (Figure 2), the alkaline waters can easily mix with lower pH waters, leading to very rapid pH changes. This is observed, for example, at site 26 (“Grande Ligurie”), where we have measured a pH drop of about 2 pH units in less than one meter. Another cause of pH decrease is the atmospheric CO2 uptake. This process is rather sluggish which explains why high-pH values (above 10) can be measured several tens of meters away from the source. This produces a very large heterogeneity in the water samples than can be collected at a single site emphasizing the role of mixing of fluids produced by serpentinization at depth with waters coming from the continental runoff. Also, although not a conservative tracer, pH provides a first-order evaluation of the extent of such a dilution.

4.2. Temperature

[35] Temperature data are reported in Figure 5 along with those of Cipolli et al. [2004] for Liguria and of Stanger [1985] for Oman.

Figure 5.

Temperature of the spring waters versus their pH (for pH values above 9). Open circles: Liguria (this work). Black dots: Oman (this work). Triangles: Oman [Stanger, 1985]. Diamonds: Liguria [Cipolli et al., 2004].

[36] First of all, Figure 5 shows that the temperatures of the springs in Oman and in Liguria have not changed drastically since the measurements of Stanger [1985] in 1980 and of Marini and Ottonello [2002] from 1995 to 2000. At pH values above 11 in the case of Oman, both temperature data set completely overlap each other, while at pH values comprised between 9 and 10.5, our measurements are at the lower end of the data range provided by Stanger [1985]. This difference can be related to the fact that we performed our sampling during the coolest season (December and January) while the data base of Stanger includes temperature measurements at different times of the year. A marked stratification of the waters in the deeper pools can be observed during the warmer season. Surface waters in the pools (around 35 −40°C close to the air temperature) can be 10–20°C warmer than the bottom waters. So differences in temperature can be the result of the heating of the surface waters and not of the variability of the temperature of the subsurface high-pH waters.

[37] Following Neal and Stanger [1984b], the T range of the spring waters in Oman (between 20 and 40°C) cannot be the sole reflection of the temperature of normal groundwaters that are close to 33°C [Weyhenmeyer, 2000]. These authors have observed that many spring waters have temperatures higher than those of nearby stream (wadi) waters and have attributed this temperature offset to the heating of the waters during their circulation in the geothermal gradient. For a temperature difference of 15°C between the spring and wadi waters and a tentative estimate of the geothermal gradient of 20°C/km, this corresponds to a penetration down to a depth that does not exceed one km. This rough estimate neglects conductive cooling of geothermal fluids during their ascent toward the surface, but according to Neal and Stanger [1984b] the waters do not penetrate deeper than the base of the ultramafic formation the thickness of which is estimated to reach a maximum of about 5 km in the central part of the ophiolitic massifs on the basis of gravimetric data [e.g., Lippard et al., 1986]. Also note that springs in Oman flow all year long despite a very dry climate where it rains only a few times a year (average of about 150 mm/yr), which is accounted for by the very low-hydraulic conductivity and high-storage coefficient of rocks from the ophiolite [Dewandel et al., 2005]. Major aquifers hosting fossil groundwaters are known in Oman and in the Arabic peninsula [Clark and Fontes, 1990] and may feed the springs all year round. Weyhenmeyer [2000] provides paleo-temperature estimates based on measurements of Ne, Ar, Kr, and Xe dissolved in 14C-dated groundwater (well deeper than 300 m). This study indicates that the average ground temperature during the Late Pleistocene (15,000–24,000 years before present) was 6.5 ± 0.6°C lower than that of today.

[38] An increase in temperature at the springs can also be due to solar heating [Neal and Stanger, 1984b] which can lead to a change in the water composition because of evaporation (see below).

[39] In the case of the Liguria springs, apart from one site, all samples with pH values above 11 display temperatures at the upper end but still overlapping the data range provided by Cipolli et al. [2004]. Alkaline waters discharge at temperatures ranging from 10 to 20°C in Liguria and are then cooler than in Oman. Our measurements were made in June and the results are similar to those reported by Cipolli et al. [2004] who sampled the same springs in January, March, September and October. The average daily air temperatures in the Genoa region are between 4°C and 8°C in February and between 19°C and 26°C in July. Accordingly, the local climatic conditions have no significant influence on the temperatures of the alkaline springs in Liguria. Also these low-temperature values preclude heating in a geothermal gradient, unless they have cooled down before discharge, which can be likely in regard to the very low-flow rates of these springs.

[40] Other hyperalkaline springs over the world hosted on ultramafic substratum display temperatures similar to our data set, between 10°C and 30°C for New Caledonia and Cyprus [Barnes et al., 1978; Neal and Shand, 2002], at 12–13°C at the Taro-Ceno valleys (Northern Appenine, Italy), [Boschetti and Toscani, 2008] and at 10 −30°C in the Cabeço de Vide region (Central-Portugal) [Marques et al., 2008]. Nevertheless, temperatures measured at the alkaline springs on land are lower than those at the LCHF (between 40 and 90°C) [Kelley et al., 2001].

4.3. Composition of the Alkaline Spring Waters

[41] At first, a water type is usually characterized by the concentrations of its major elements. In Oman and Liguria, waters at “more neutral” pH values (i.e., below 9 or so) flowing over ultramafic substratum, like those found in the wadis in Oman or in the streams in Liguria, are of the Mg-HCO3 type. This feature drastically differs from common continental waters that are of the Ca-HCO3 type, as they flow over rocks that are in most cases Mg-poorer than ophiolitic rocks. Hyperalkaline waters (that we have defined as those with a pH above 10) are of the Ca-OH type. As such, we suspect, as it is the case during the alteration of the oceanic crust by seawater, that the Mg decrease and the simultaneous Ca increase in the hyperalkaline fluid are related to the fluid circulation at low temperature within the ophiolite [Alt, 1995]. Alkaline waters sometimes belong to other water types as discussed extensively by Neal and [Shand, 2002]. Taking the example of the Cyprus ophiolite, the concentrations of Na, Cl and SO4 are large enough that some of the waters can be defined as Na-Mg-HCO3-Cl [Neal and Shand, 2002]. This could also be applied to the saltiest of the high-pH Omani waters that could be characterized as belonging to the Na-Ca-Cl-OH type. Such classification is only qualitative.

[42] We first report the water compositions in plots of the element concentrations versus pH on one hand and versus the chloride content on the other hand. The range of variation of the concentrations is so large that it is necessary to use a logarithmic scale in most cases. We also use Cl-normalized concentrations to remove the concentration changes due to evaporation (refer to Seyfried and Shanks [2004]) for a discussion on the use of chemical tracers to investigate phase transitions in high-temperature hydrothermal systems, that is relevant to the present case). Also note that Mg-normalized concentrations are often used to characterize hydrothermal processes (see for example [Sader et al., 2007] who studied the serpentinization of kimberlites). In our case using these Mg-normalized concentrations did not help reducing the data.

4.3.1. Variation of the Spring Chemical Compositions with Time

[43] The data reported by Neal and Stanger [1984b] have been collected over a period of several years around 1980. These authors have also measured the composition of a single spring, the Karku alkaline spring, at10 different dates over a period of 5 years. Their results showed that the chemical composition of waters has remained stable over time, even a few days after a heavy rain which is expected to dilute the waters toward more neutral pH values. Moreover, our recent data are in agreement with those of Neal and Stanger [1984b]. This stability of the alkaline spring composition is also evidenced by the agreement between our data for Liguria and those of Cipolli et al. [2004], who have sampled the Ligurian springs at different periods of the year and did not mention any noticeable variations in their composition. This has also been stressed by Früh-Green et al. [2009] and Schwarzenbach [2011].

4.3.2. Sodium and Chloride

[44] The Omani waters are saltier that the Ligurian ones. Their Cl content varies from very low values (dilute waters similar to those of Liguria, i.e., around 0.5 mmol/L) to values as high as 72 mmol/L (Figure 6). In the entire data set, only three water samples have a Cl content higher than 30 mmol/l. These three samples (432, 474, and 478 in Stanger [1985]) have pH values around 9.5 and were collected in hand dug wells. They are obviously outliers in the entire data set and have not been considered any further in this study. The vast majority of the Oman waters has a total dissolved solid (TDS) content of at most about 2 g/L, but remains around 0.8 g/L at pH value above 11 and decreasing down to 0.2–0.3 g/L at pH value around 9.

Figure 6.

The sodium concentration versus that of chloride in the Omani and Ligurian springs. Open circles: Liguria (this work). Black dots: Oman (this work). Triangles: Oman [Stanger, 1985]. Diamonds: Liguria [Cipolli et al., 2004]. The solid line with a slope of 1 indicates the evaporation trend.

[45] The Cl concentration of the Ligurian waters is as low as 0.16 mmol/L and does not exceed 0.65 mmol/L, with a TDS load of 50 mg/L with a highest value of 800 mg/L [Marini and Ottonello, 2002].

[46] Away from the peculiar conditions of halite precipitation, Cl is a conservative element that can be used to trace evaporation and mixing of waters of different salinities. When the element concentrations are expressed in a molar (molality or molarity) scale, their variation due to evaporation as a function of the Cl content defines a trend with a slope of 1. This is the case for the Na concentration of the Omani hyperalkaline waters but not for the Ligurian hyperalkaline springs (Figure 6) which show a large variation of the Na concentration for a narrow range of Cl concentrations. The data scatter around the evaporation line is nevertheless quite significant. The Na concentration can be twice higher or lower than the Cl content. The values for Oman are less scattered than those for Liguria but still around the evaporation line and not strictly on top of it as the sole evaporation would demand. The departure of the data from the evaporation trend excludes evaporation as the sole cause of the variation in the Cl and Na contents of the waters. This offset can be due to either a Na (respectively Cl) release to or consumption from the waters. Taking the Taro-Ceno valley waters (Northern Appenine, Italy) as an example, Boschetti and Toscani [2008] point to the Na and Cl release to the aqueous fluid due to leaching of fluid inclusions during serpentinization. Scambelluri et al. [1997] also report the presence of high-salinity fluid inclusions in the Italian alpine serpentinites. However, the departure from the evaporation line is mainly due to increasing Na concentration rather than Cl ones, suggesting that Na-bearing minerals such as plagioclase feldspar may have played a significant role [Boschetti and Toscani, 2008]. Scambelluri et al. [1997] argue for the presence of alkali-bearing phyllosilicates with 0.2–0.55 wt% Na2O for the Italian alpine ophiolites. The Na content of the Oman mantle peridotites (whole rocks) is very low (around 0.01 wt % Na2O) [Monnier et al., 2006], but reaches 1–2 wt% in the gabbros [Benoit et al., 1996]. In their study at high temperatures, Seyfried et al. [2007] point to the eventual contribution of clinopyroxene dissolution to the sodium budget. Oman mantle peridotites are essentially harzburgites with a very low-cpx content (<1% on average); moreover, the Na content of Oman mantle clinopyroxene is quite low (≤0.1 wt% Na2O). Gabbroic cumulates are, on the contrary, cpx-rich (up to 50%) but the Na content of these crustal cumulates is also quite low (≤0.6 wt% Na2O), the Na in gabbros being essentially concentrated in plagioclase) [Monnier et al., 2006; Python and Ceuleneer, 2003]. Accordingly, this contribution is likely minor in Oman.

[47] Besides, if the relationship between the Na and Cl concentrations was solely due to evaporation, this would mean that all the high-pH fluids sampled at the springs in Oman originate from a single fluid the composition of which would be modified only by mineral precipitation and contact with the atmosphere during discharge, which is very unlikely.

4.3.3. Calcium

[48] The Ca concentrations reported versus pH do not show any clear trend. The data for Oman and for Liguria lie on the same trend for the high-pH values but the Ca content reaches values higher in Oman than in Liguria for the same pH (Figure 7).

Figure 7.

Major element concentrations versus pH for the Omani and Ligurian springs. Open circles: Liguria (this work). Black dots: Oman (this work). Triangles: Oman [Stanger, 1985]. Diamonds: Liguria [Cipolli et al., 2004].

[49] When reported versus the Cl content the Ca concentrations are very scattered and lower than the values predicted by the evaporation line (Figure 8). But the Cl-normalized Ca concentrations versus pH (Figure 9) show that the Ligurian waters define a trend where the Ca/Cl ratio increases for pH values above 11, while that for the Oman waters is constant. Calcium carbonate precipitation either as calcite or aragonite is very common in the springs in Liguria as well as in Oman, as revealed by the mineralogical observations and the travertines found at the springs [Chavagnac et al., 2013].

Figure 8.

Major element concentration versus Cl in the Omani and Ligurian springs. Open circles: Liguria (this work). Black dots: Oman (this work). Triangles: Oman [Stanger, 1985]. Diamonds: Liguria [Cipolli et al., 2004]. The downward arrow in the Mg plot indicates the direction of increasing pH.

Figure 9.

Cl-normalized major element concentration versus pH in the Omani and Ligurian springs.Open circles: Liguria (this work). Black dots: Oman (this work). Triangles: Oman [Stanger, 1985]. Diamonds: Liguria [Cipolli et al., 2004].

[50] In their water-rock reaction simulations of alteration in ultramafic systems by fresh water and by seawater, Palandri and Reed [2004] showed that the reacted waters becomes hyperalkaline and strongly reducing with increasing dissolved Ca concentration together with decreasing water-rock ratios. These authors also discuss which minerals likely participate in the control of the dissolved Ca concentration and conclude that calcium may be exported out of the system by the serpentinizing fluid, which meets the conclusion of Barnes and O'Neil [1969].

4.3.4. Alkalinity and the DIC Content

[51] The data set for alkalinity and for DIC is not as extensive as that for other elements: what is available are a few of our own measurements for Oman waters and the data of Marini and Ottonello [2002] for Liguria. Alkalinity is about constant with values scattered between 1 and 4 mmol/L up to pH 11 where it sharply increases up to 15 mmol/L (Figure 7). Alkalinity can be defined in several ways (see the discussion in Zeebe and Wolf-Gladrow [2001] for the case of seawater) but titration alkalinity is unambiguously defined as the sum of all the bases that can be titrated by a strong acid. In river waters titration alkalinity is dominated by the carbonate ions, but in high-pH waters the concentration of the hydroxide ion becomes significant. Even if silicic acid is dissociated to form H3SiO4(aq) at high pH, it is a negligible contribution to alkalinity due to the very low-silica content of the waters (Figure 7). Stanger [1985] reports the direct measurement of the carbonate and bicarbonate concentrations as well as that of hydroxide (but almost never at the same time for the same water sample), apparently from a single acidic titration. In absence of any element of comparison, we did not take these data into account, which were also ignored by Stanger himself.

[52] When plotted versus pH (Figure 7), our data for Oman and that of Cipolli et al. [2004] for Liguria show that the alkalinity of the high-pH waters is the same, but the Cl-normalized values show two distinct trends: a slight increase of the Alkalinity/Cl ratio for Oman and a very large increase (and a larger scatter) for the Ligurian waters at high pH (Figure 9).

[53] The Cl-normalized dissolved inorganic carbon (DIC/Cl) for the Oman waters is reported in Figure 11 as a function of their Cl-normalized alkalinity. Two trends are apparent. The first one groups the wadi waters (Mg-HCO3 type, pH lower than 9). Their alkalinity is dominated by the carbonate ions (high-DIC values). The second trend is for the high-pH waters (pH above 10.5) for which alkalinity is dominated by the hydroxide ion (low-DIC content). Such trends have already been evidenced by Sader et al. [2007] in their study of the serpentinization of kimberlites. This shows that CO2-rich meteoric water percolating through ultramafic formations are stripped from their DIC and that peridotites are indeed an efficient sink for carbon [Bruni et al., 2002; Cipolli et al., 2004; Kelemen and Matter, 2008]. As a consequence, these high-pH waters have a very low-equilibrium pCO2 and are prone to absorbing atmospheric carbon dioxide at their emergence [Bruni et al., 2002]. This is also confirmed by measurements of the composition of gases emitted at the hyperalkaline springs sites which show that they do not contain any CO2 [Boulart et al., 2013].

[54] The calcite saturation state (Figure 10) shows that it becomes saturated or even surpersaturated at pH values above 10.5.

Figure 10.

The calcite saturation state (Log (Q/Ksp)) versus pH of the Omani and Ligurian waters (dots: data for Oman (this work); diamonds: data for Liguria [Cipolli et al., 2004]). Values above 0 at pH above 10.5 indicate supersaturation with respect to calcite.

Figure 11.

Chloride normalized Alkalinity (Alk/Cl) versus DIC/Cl content for the Omani waters (this work).

4.3.5. Potassium

[55] There is no apparent correlation between the potassium concentration and pH (Figure 7), but there are two trends of the K concentration with the Cl content (Figure 9). The first one is for the Oman waters where K linearly increases with the Cl content. The other trend is that of the Ligurian waters for which K varies between 0.01 and 0.13 mmol/L at almost constant Cl value. As no K-bearing mineral has been found to precipitate in the alkaline springs either in Oman or in Liguria, it can be assumed that K is conservative when it is leached from the ultramafic substratum. Accordingly, the only cause of the increase in the K concentrations of the waters with that of chloride should be evaporation. This is supported by the constant value of the K/Cl ratio (Figure 9), but not by the scatter in the data, like in the case of Na. Although the K concentrations at high pH are the same for Oman and for Liguria (Figure 7), the K/Cl ratio differs, similarly to the behaviour of the Ca/Cl ratio (Figure 9).

[56] Recent geological formations essentially made of clay minerals are well documented in the Oman ophiolite. Bahla is the town of potters who find clays in abundance in quaternary deposits located downstream of an alkaline spring located in the suburbs. Also, a wadi located in the south-eastern part of the Oman mountains and running through the mantle peridotites has been named Wadi Tayin, “tayin” meaning clay in Arabic. Alkaline springs are quite numerous along this wadi. On the other hand, the Battinah plain, the coastal plain east of the Omani mountains, is covered with clay-rich quaternary deposits. It may be hypothesized that fine K-bearing clay particles forming either at the spring discharge or during the alteration process itself are washed away eastward from the ophiolite toward the sea during the flood events. They may sometimes accumulate in hydrological traps such as the Wadi Tayin and the Bahla area inside the mountain range. The question of the formation of clays in ultramafic bodies has never been addressed so far.

4.3.6. Magnesium

[57] The Mg concentrations of the waters sharply decrease with increasing pH over 6 orders of magnitude and drop from the 1–4 mmol/L range to very low values at pH values above about 10 (Figure 7). All Mg values for waters at pH above 11 are below 0.2 mmol/L and most of them are below 0.003 mmol/L, up to the point where magnesium in these waters becomes a trace element.

[58] Three groups of data can be distinguished when the Mg concentration is plotted versus Cl (Figure 8). The first group encompasses the Ligurian waters and show that for a Cl concentration between about 0.5 and 1 mmol/L, the Mg concentration varies over 5 orders of magnitude, from the lowest value of 0.0001 mmol/L [Cipolli et al., 2004] (this work) for the spring L43 at pH 11.52 to the highest value of 0.63 mmol/L measured at the spring ORB101 at pH = 10.59.

[59] The second one is the same as the first one but for Oman waters with a Cl content between 5 and 10 mmol/L. These two data subsets cover the high-pH waters. Waters (mostly from Oman) with pH values lower than 10 make a third group with Mg values roughly constant around 1 mmol/L and with a Cl content varying from about 0.8–10 mmol/L.

[60] Normalizing the Mg content to that of Cl does not reduce the scatter in the data when plotted versus pH. Mg/Cl values are more scattered for Cl concentrations smaller than 5 mmol/L than those at higher Cl values. Considering the range of variation of Mg, the influence of evaporation on the Mg concentration must be very small for these waters.

[61] There are several minerals that can control the Mg concentrations of the high-pH waters. The question is to know whether this control occurs at depth during serpentinization, or if it is due to crystallization at the water discharge. The thermodynamic calculations first show that the alkaline waters are largely supersaturated with respect to a variety of Mg-bearing silicates (talc, serpentine, chrysotile, sepiolite, diopside, etc). Serpentine exerts the main control on the Mg content of the fluids, according to reactions (3) and (4). At the spring discharges in Oman, our mineralogical determinations [Chavagnac et al., 2013] reveal that brucite is found in the white precipitates covering the wadi beds. We did not find any brucite in Liguria, although thermodynamic models [Bruni et al., 2002; Marini and Ottonello, 2002; this work] show that it reaches saturation in the Ligurian high-pH waters. Brucite and serpentine are very insoluble minerals that can keep the Mg contents of the waters at very low levels. Brucite formation (reaction (6)) is almost always reported as an initial step in the serpentinization process [e.g., Foucault and Raoult, 2001; Evans, 2010; Kelemen and Matter, 2008; McCollom and Bach, 2009] although it is rarely observed in mineral assemblages in serpentinites. Simulations of peridotite alteration [McCollom and Bach, 2009; Palandri and Reed, 2004; Sader et al., 2007] show that indeed small quantities of brucite form in narrow ranges of temperature and water/rock ratios. In the aerial discharge of hyperalkaline waters brucite formation appears to be quite common. Its formation is favoured by mixing: river (wadi) waters providing magnesium while the high-pH fluids provide the hydroxide ion. Foustoukos et al. [2008] proposed that the dissolution of diopside and brucite and the formation of chrysotile may control the serpentinizing fluid pH at temperatures below 300°C. Indeed the dissolution of an hydroxide like brucite is able to limit the H+(aq) concentration, but invoking brucite dissolution involves an already altered peridotite.

[62] The brucite saturation indices are reported in Figure 12. Saturation is reached at a pH value of about 10. At high pH the values are scattered around the equilibrium line, in agreement with the results of Marini and Ottonello [2002] (also reported by Bruni et al. [2002]) for Liguria and of Neal and Shand [2002] for Cyprus. At other locations, Marques et al. [2008] report brucite undersaturation for the three high-pH springs of central Portugal while Neal and Stanger [1984b] find supersaturation for the Omani hyperalkaline springs. Although one could expect that the saturation index values reported in these studies would point to equilibrium between brucite in the high-pH fluids, the picture that emerges from these papers is that of a large scatter at a single location and also between various places. This scatter is due in first place to the scatter in the Mg concentration values (Figure 7).

Figure 12.

The brucite saturation state (Log (Q/Ksp)) versus pH in Omani and Ligurian waters (data in Table 2, this work) showing that brucite can precipitate from hyperalkaline springs at pH values above 10.5.

[63] The fact that brucite saturation is reached seems to be the major cause of the change in the solution composition that occurs at pH equal to 10 and above. We have found a place in Oman that we called “Grande Ligurie” where the mixing of high-pH waters (pH = 11) with wadi waters (pH = 9) creates a very strong pH gradient over a very short distance (a drop of 2 pH units in less than a meter). A white precipitate composed of Ca-carbonates and brucite [Chavagnac et al., 2013] appears at a pH value of 10.4.

[64] Lastly, Anraku et al. [2010] report the formation of hydrotalcite, Mg6Al2(CO3)(OH)16•4(H2O), a mineral of the layered double hydroxides (LDH) family [Bravo-Suarez et al., 2004; Goh et al., 2008] in the mixing zone between the wadi waters and the high-pH fluids. Although occurring in trace quantities [Anraku et al., 2010], hydrotalcite may indeed control the Mg concentration of the waters, but also on their carbonate content. It also points to the role that aluminum may play in the process. The presence of hydrotalcite in our samples and its relation with brucite formation are described in an extended mineralogical study of the spring precipitates at the Omani and Ligurian springs [Chavagnac et al., 2013].

[65] To sum up several secondary minerals are good candidates to limit the magnesium concentration of the fluids. Serpentine forms within the ultramafic body itself, but brucite and hydrotalcite precipitate at the spring discharge during the mixing of Mg-HCO3 type surface waters and the Ca-OH type high-pH fluids.

4.3.7. Silica

[66] The silica concentrations are mostly below 0.5 mmol/L and do not display any tendency with pH (Figure 7). The scatter in the data is as large as that for Mg. The values for Liguria are slightly higher than those for Oman. When plotted versus Cl the data (Figure 8) can be sorted in the same three groups as for Mg, although not as clearly, i.e., a first group encompassing the Ligurian waters with SiO2(aq) values between 0.1 and 500 µmol/L for a Cl content lower than 1 mmol/L, a second groups (mostly Omani waters) with SiO2(aq) values similar to the first group but for Cl concentrations larger than about 5 mmol/L, and a third group with SiO2(aq) values around 500 µmol/L for a varying Cl content. The silica contents of the high-pH waters are quite similar to those of the lower pH fluids. Shibuya et al. [2010] have calculated the composition of highly alkaline high-temperature hydrothermal fluids in the early Archean ocean and have found that the silica concentration should have been quite high (i.e., about 200 mmol/L at pH 11). They assumed equilibrium with quartz in their calculations, the solubility of which does increase at alkaline conditions. These calculations do not match the experimental results of Seyfried et al. [2007] who observed that the silica content of the serpentinizing fluid remains quite low. Also a low-silica activity is a key feature of serpentinizing environments [Frost and Beard, 2007]. Recently, Streit et al. [2012] have described the association of quartz and serpentine in carbonate veins crosscutting highly serpentinized peridotites in the Oman mountains [Nasir et al., 2007]. Such samples are nevertheless uncommon and the origin of quartz, which is a minor phase in such formations [Nasir et al., 2007] is still uncertain.

[67] Quartz saturation is reached for pH values below 10.5 (Figure 13) above which it becomes largely undersaturated [Paukert et al., 2012]. Other forms of silica (chalcedony, amorphous SiO2) are largely undersaturated in the whole pH range.

Figure 13.

The quartz saturation state Log (Q/Ksp) versus pH in Omani and Ligurian waters (data in Table 2, this work). Note the change from quartz saturation to quartz undersaturation at a pH value of about 10.

4.3.8. Sulfate

[68] When plotted versus pH the sulfate concentrations appear also very scattered and do not exhibit any trend (Figure 7). The values for Liguria are below 0.1 mmol/L while those for Oman are between 0.001 and 5 mmol/L. Again, when plotted versus Cl the data show the same three groups as for Mg and SiO2 with a clear distinction between the Ligurian waters and the Oman ones (Figure 8). There is also a very large decrease of the SO4/Cl ratio with pH.

[69] Sulfate can have several origins: leaching of sulfate minerals remnants of the marine origin of the ophiolites or oxidation of sulfides common in marine hydrothermal systems. In Oman, copper sulfate is common along former syn-accretion fault zones [Abily et al., 2011; Leblanc et al., 1991]. For example, Neal and Shand [2002] have shown that high-pH waters of the Cyprus ophiolite are close to equilibrium with gypsum. Calculations using PHREEQC indicate that gypsum (as well as any other sulfate mineral) is largely undersaturated everywhere.

5. Discussion: The Role of Hydrology

[70] The meteoric alteration of continental surfaces leads to the production of runoff waters of various composition and element concentrations that provide the input to continental hydrothermal systems. This variability hampers the exact definition of an end member composition of the recharge waters as precisely as for example for marine systems like the LCHF where the input is well-characterized bottom seawater. When percolating through the ultramafic formation such waters are modified by the serpentinization process before discharging at the surface when the hydrologic conditions allow it. Then surface phenomena at the discharge sites, such as evaporation, formation of precipitates in the springs and mixing of different waters, further modify the waters composition, leading to the waters actually sampled at the springs, as illustrated in the present work. A central question in the way these waters acquire their composition is that of the circulation pathways. Do the waters percolate only in the ultramafic formation or do they have the opportunity to react with a larger spectrum of mineral assemblages? Most alkaline springs of Oman are located along major structural discontinuities such as the mantle/crust boundary (so called “Moho”) inside the ophiolite itself or along the thrust contact between the ophiolite and the metamorphic sole and underlying sedimentary formations (Figure 1). Do the waters interact, along the basal contact, with the underlying metabasites and meta-sedimentary rocks and, along the Moho, with the overlying gabbroic and ultramafic cumulates?

[71] A plot of the Cl content of the Oman waters versus that of K (Figure 14, Table 3) where we have related the springs to their location with respect to the geological units shows that different groups of waters can be distinguished. The saltiest waters (those with the highest Cl content) are those collected at springs located along the basal thurst plane of the ophiolite or in the gabbros. The latter definitely constitute a different group, distinct from the linear trend depicted by the rest of the data set. Waters get their peculiar hyperalkaline compositions by interaction with the mantle peridotite, but they may become saltier if they leach sedimentary formations when they circulate at the base of the ophiolite. Some of the springs located within the gabbros have Na/K ratios different from the other springs. They also display the highest Al concentrations (Table 3). This correlation between the water chemistry and the geological context (Figure 15) clearly shows that there is a mineralogical control on the composition of the springs, whose extend remains to be quantified.

Figure 14.

The chloride content of the Oman waters versus their potassium concentration (data: this work) related to the geological setting of the springs. Filled dots: base of the ophiolitic nappe; filled triangles: gabbros; open triangles Moho on the side of the gabbros; inverted triangles: Moho on the side of the peridotites; diamonds: peridotites.

Figure 15.

A sketch of the geological settings of the Oman alkaline springs to illustrate the geometry of the northern ophiolite massifs where the main lithological contacts are slightly dipping to the east and where the water divide is located on outcrops of mantle peridotites (not to scale). Meteoric waters percolate throughout the densely fractured peridotites and discharge as alkaline springs along the basal thrust contact or at the Moho or after leaching in addition to peridotites, either metamorphic and sedimentary rocks (at the basal contact) or gabbroic cumulates (at Moho level).

[72] In Liguria, the springs are also located on different geological formations within the Voltri group (metagabbros and metabasalts for the Beigua unit, and serpentinized lherzolites for the Erro-Tobio unit), but our data set is too scarce to reveal any direct control applied by the host bedrock on observed aqueous geochemistry. In her detailed study of the mineralogical, isotopic, and chemical compositions of the basement rocks in Liguria, Schwarzenbach [2011] proposed that the elevated Na and K concentrations in some spring waters were due to the presence of weakly serpentinized peridotites along the alkaline water flow paths.

6. Concluding Remarks

[73] Our regional study in Oman and in Liguria first shows that the composition and the temperature of the high-pH fluids produced during low temperature serpentinization have not changed since the time when their first values have been reported (i.e., over the last 20–30 years) in spite of significant changes in the level of the underground water reservoir due to a dramatic increase in the exploitation of the water resource during the same period. The first-order correlation between the Na and Cl concentrations tends to point to a concentration of the fluids by evaporation, mainly in Oman, but this is not entirely supported by the scatter of the data around the linear trend which indicates that Na and Cl may not be conservative during the alteration of the ultramafic rocks.

[74] This is also observed for K. The K concentration in the high-pH fluids is scattered but shows an increasing linear trend with Cl for the Oman waters. The K/Cl ratio is roughly constant and scattered, with distinct values for Liguria and for Oman. This departure from the evaporation line for Na and K may be due to the leaching of high-salinity fluid inclusions, to the dissolution of alkali-bearing silicates (pyroxenes) in the subsurface and to the formation of clay minerals at the springs that can accumulate in sedimentary traps such as the Wadi Tayin and Bahla areas in Oman, but that are generally washed away from the ultramafic outcrops.

[75] Mg is almost entirely depleted at high pH (pH > 10.5), which can result from serpentine formation within the ultramafic body and from brucite and hydrotalcite precipitation at the springs. There is no obvious control of the sulfate concentration of the waters.

[76] Ca accumulates in the high-pH fluids and is consumed by Ca-carbonate formation at the springs, either by mixing with river waters or by the CO2 supply from the atmosphere. The calculation of the saturation state of the waters with respect to various minerals show that brucite saturation is reached at pH values around 10.5 and triggers major changes in the water composition and in the element concentrations. The waters evolve from a quartz-saturated low-pH continental environment to a brucite-dominated high-pH serpentinizing system at low temperature.

[77] Finally, the salinity and the alkali (Na and K) and Al concentration data point to a link between the composition of the springs and their geological setting (gabbros, periodotites and contact with the metasediments) which addresses the question of the hydrologic pathways followed by the serpentinizing fluids and of the impact of the mineralogical diversity on the water composition.


[78] We are very indebted to the Ministry of Water Resources of Oman for giving us access to their data base on Oman springs. Special thanks to Jean-Paul Breton for his help with public relations in Muscat. We also acknowledge the help of Philippe Besson, Stéphanie Mounic and Carole Causserand for the chemical analyses and of Anne-Marie Cousin for the nice Oman drawing. Financial support was provided by the CNRS-INSU CESSUR Program.