Chlorine cycling during subduction of altered oceanic crust

ELSEVIER

Earth and Planetary Science Letters 161 (1998) 33–44

Chlorine cycling during subduction of altered oceanic crust Pascal Philippot a,Ł , Pierre Agrinier b , Marco Scambelluri c b

a CNRS-URA736, Laboratoire de Pe ´ trologie, T26-E3, Universite´ Paris 6 and 7, 4 place Jussieu, 75005 Paris, France Laboratoire de Ge´ochimie des Isotopes Stables, IPGP, T54-E1, Universite´ Paris 7, 4 place Jussieu, 75005 Paris, France c Dipartimento di Scienza della Terra, Corso Europa 26, 16000 Genova, Italy

Received 26 September 1997; revised version received 11 June 1998; accepted 12 June 1998

Abstract Eclogitic rocks that have experienced devolatilization, with little or no interaction with external fluid sources, can be viewed as analogues of crustal material which may be transferred back into the mantle during subduction. Thus they can be used to evaluate the extent of the recycling of volatile elements, such as chlorine. We report new oxygen isotope ratios of omphacite, and fluid inclusion data determined from eclogitic metagabbros, of the Rocciavre´ massif (Italian Alps). The data are compared with those obtained for the Monviso, Cyclades and the Franciscan Complex high-pressure rocks. In all localities, relics of early dehydration fluids are preserved as primary fluid inclusions in the cores of omphacite megacrysts (Rocciavre´, Monviso and Franciscan Complex) or garnet (Cyclades). Salinity estimates of the inclusion fluids range from 32 to 45 wt% NaCl in Rocciavre´, 17 to 21 wt% NaCl in Monviso, and are similar to seawater in other areas. Omphacite and bulk-rock δ18 O values of Rocciavre´ (5.1–6.8‰) and Monviso (3.0–5.3‰) metagabbros are markedly lower than those of Cyclades (6.8–14.3‰) and the Franciscan (6.7–13.1‰) metabasites. The fluid salinity–δ18 O systematics of eclogitic rocks is similar to that documented along a typical section of the altered oceanic crust and unmetamorphosed ophiolites. This suggests that high-pressure metamorphism, and associated processes, did not modify significantly the variability in chlorine concentrations and oxygen isotope ratios, inherited from a stage of sea-floor hydrothermal alteration under low(basaltic layer) and high-temperature (gabbroic layer) conditions, respectively. Extrapolating the estimated H2 O and Cl contents of eclogitic rocks to a representative section of the subducted oceanic crust indicates that a minimum of 100 to 200 ppm Cl could be recycled into the mantle during subduction. This yields a Cl=H2 O ratio of 3.6 to 7:5 ð 10 2 for the subducted oceanic crust, which is similar to E-MORB. On the basis of available Cl isotopic data, we infer that a large proportion (70%) of the Cl stored in the altered crust should be recycled to the mantle to generate an isotopic composition of the subducted crust equivalent to the source of unaltered mid-ocean ridge basalt (δ37 Cl D 4.7‰).  1998 Elsevier Science B.V. All rights reserved. Keywords: chlorine; geochemical cycle; subduction; eclogite

1. Introduction Understanding the exchange of volatiles between the mantle and the exosphere, including their acŁ Corresponding

author. Fax: 33 1 44 27 39 11; E-mail: [email protected]

cumulation in surface reservoirs via outgassing of the Earth’s interior, the degree of mantle depletion and the extent of recycling back into the mantle, represents a central issue of terrestrial geodynamics. Two processes of great significance to the global H2 O and Cl distribution are the alteration of the oceanic crust by seawater, and the devolatilization

0012-821X/98/$19.00  1998 Elsevier Science B.V. All rights reserved. PII S 0 0 1 2 - 8 2 1 X ( 9 8 ) 0 0 1 3 4 - 4

34

P. Philippot et al. / Earth and Planetary Science Letters 161 (1998) 33–44

Fig. 1. Schematic section of the oceanic crust at spreading center showing the thermal cycle, fluid salinity and a synthetic oxygen stable isotopic profile through the entire section. The thermal and fluid salinity profiles are from [4], the δ18 O profile is from [2,5–7]. See text for discussion.

of the subducted oceanic slab. Owing to the fairly large amount of samples available for analysis, the alteration pattern and hydrothermal characteristics of the oceanic crust can be estimated reasonably well ([1–4]; see below and Fig. 1). In contrast, estimates of fluxes out of the subducted slab are very uncertain and there are a range of conflicting models [8–11]. A common approach for evaluating the fate of H2 O in subduction zones is derived from experimental petrology; the behaviour of H2 O is constrained by consideration of the stability field of the common hydrous minerals of the oceanic crust (e.g. amphibole, chlorite, talc etc.). Thompson [12] has recently reviewed the situation of water in the mantle. According to Ahrens [13], there may be two times more water stored in the mantle as there is in the oceans. Another approach consists in examining the processes of fluid-rock interactions in subducted material that have been returned to the Earth’s surface using a combination of stable isotope, fluid inclusion and mineral assemblage analyses [11,14–28]. There appears now to be some agreement that dramatic loss of water and associated large-scale fluid transport, is confined to relatively shallow levels of subduction zones (50 km), but that fluid circulation is rather limited at greater depths (e.g., [11,27]). With regards to Cl, samples of early compaction and devolatilization fluids can be collected directly in accretionary prisms and oceanic troughs [29]. For

more deeply buried rocks, much difficulty arises from the paucity of data on the Cl content of subducted materials (e.g., [26]). As a consequence, attempts to model the cycle of chlorine have generally considered that the budget of chlorine in the exosphere was due to mantle degassing alone or, in other words, that all the chlorine incorporated in the altered oceanic crust was lost during early subduction [30–32]. Recently however, on the basis of chlorine isotope analysis of oceanic material, Magenheim et al. [33] showed that the current evolution of the Earth’s Cl distribution likely involved a combination of degassing of the mantle and the recycling of altered crust through subduction. The quantity of exospheric Cl that can be transferred back to the mantle remains largely unknown, however. The aim of this paper is to show that eclogitic rocks, that have experienced devolatilization, with little or no interaction with external fluid sources, can be viewed as relevant analogues for crustal material that may be transferred back into the mantle. Thus they can be used to evaluate the magnitude of Cl recycling into the mantle during subduction. The approach we use combines oxygen isotope analysis of high-pressure minerals with fluid inclusion analysis. The observed correlation between 18 O=16 O ratios, which are good monitors of fluid movement in metamorphic rocks, and chlorine concentrations of the fluid inclusions, which are likely to be buffered by

P. Philippot et al. / Earth and Planetary Science Letters 161 (1998) 33–44

the fluid (not the host rock), is invaluable in assessing the dynamics of chlorine during devolatilization processes. We first present new oxygen isotope and fluid inclusion data from eclogitic metagabbros of the Rocciavre´ massif, a meta-ophiolitic complex equivalent to the Monviso massif and Zermatt–Saas-Fee area in the western Alps. We then compare the pattern of δ18 O values and Cl distribution of the altered oceanic crust with that of high-pressure rocks from a variety of geological settings. Finally, we evaluate the significance of saline fluids in deeply subducted oceanic material in terms of the mantle Cl inventory.

2. Oceanic protoliths of Alpine eclogites The oceanic Piemonte units of the Western Alps consist of three major metamorphic nappes emplaced one above the other. All units consist of oceanic metasediments with large masses of more or less coherent ophiolitic units. The structurally lowest unit (Zermatt–Saas Zone, Viu` area, Rocciavre´, Monviso and Voltri Group), is an eclogite to high-grade blueschist facies unit. This ophiolite suites formed during opening of the Piedmont–Ligurian ocean basin during the Jurassic and were then brought to great depths (50–100 km) during closure of the basin by Cretaceous–Eocene subduction. In the Monviso and Rocciavre´ Massifs, eclogitic assemblages are particularly well preserved in Fe-rich metagabbros. These are generally interlayered with less differentiated metagabbros and form cumulate sequences a few hundred metres thick (Lago Superiore unit in Monviso [34] and Punta del Lago unit in Rocciavre´ [35]). In the Lago Superiore unit, Fe-metagabbros form a major eclogite-facies ductile shear zone [22]. The shear zone consists of small domains of lowstrain rocks that preserve precursor igneous textures surrounded by large volumes of mylonitic rocks; the mylonites are in turn cut by different generations of eclogitic veins [22,36]. In the Punta del Lago unit, eclogitic metagabbros largely escaped plastic deformation and consist of massive, coarse-grained elongated bodies bounded by relatively narrow shear zones [35]. Locally, massive metagabbros are cut by randomly oriented eclogitic veins a few centimetres in width. In both units, the rocks contain the same

35

omphacite, garnet, rutile and apatite mineral assemblage. In addition, quartz is present in the Rocciavre´ rocks. In Rocciavre´, incomplete recrystallization of precursor magmatic assemblages led to the local preservation of augite in the core of omphacite clasts, andesine, partly pseudomorphed plagioclase and ilmenite relics in rutile grains [37]. In a few Rocciavre´ rocks, Pognante [35] described augite partly pseudomorphed by brown amphibole, which he interpreted to be of hydrothermal origin. Such relics are absent in the Monviso Fe-metagabbros probably because of intense ductile deformation that favored complete mineral re-equilibration under high-pressure conditions. In both units, P –T conditions of high-pressure crystallization have been estimated at P > 1:1 GPa for a temperature of 450–550ºC [34,35].

3. Rocciavre´ samples 3.1. Textures The Rocciavre´ samples collected for fluid inclusion and oxygen isotopic analyses are listed in Table 1. Stable isotope analyses were performed on omphacite clasts (omphacite 1 of the terminology of Nadeau et al. [14]) pseudomorphing precursor augite in low-strain rocks and on omphacite fibres (omphacite 4 of [14]) extracted from one eclogitic vein. Fluid inclusions were analyzed in omphacite clasts from three particularly well-preserved lowstrain rocks (i.e., showing no evidence of late-stage retrograde alteration overprint). The inclusions are 5–30 µm tubular inclusions that are randomly distributed throughout the crystals and preferentially oriented parallel to [001]. Following Nadeau et al. [14], these inclusions can be interpreted either as exsolution inclusions, that precipitated from the host during omphacite replacement of augite, or as dehydration fluids that derived from the breakdown of amphibole and=or chlorite. The latter minerals partly or entirely replace magmatic pyroxene in the hydrothermally altered gabbroic protolith of the Monviso and Rocciavre´ rocks (Chenaillet massif, see [14] for a discussion). At room temperature, all fluid inclusions contain three or more phases (liquid, vapour and a halite cube š additional undetermined phases).

36

Location

Sample

Rocciavre´ RC 21a RC 21b RC 55 RC 56 RC 57 RC 58 RC 70 Monviso

See [14]

Rock type

Host mineral δ18 OSMOW Tm .ice/ range, (no. of analyses, 1σ) ºC (no. of analyses)

Tm .halite/ range, Th .L/ range, NaCl Solid phases ºC (no. of analyses) ºC (wt% eq.)

vein low-strain rock low-strain rock low-strain rock low-strain rock low-strain rock low-strain rock

omphacite omphacite omphacite omphacite omphacite omphacite omphacite

6.01 (1) 6.09 (2, 0.06) 6.75 (2, 0.18) 5.90 (2, 0.03) 5.76 (3, 0.03) 5.87 (1) 5.21 (2, 0.10)

212–380 (14)

270–329

33–45

halite, undetermined phases

250–321 (6) 201–297 (11)

275–285 275–379

35–40 32–38

halite, undetermined phases halite, undetermined phases

omphacite a

3.0–5.3 (13, <0.2)

190–246

17–21

calcite, pyrite

12.8 to

18.3 (103)

Tm .ice/ : temperature of final melting; Tm .halite/ : temperature of salt dissolution; Th .L/ : temperature of homogenisation to liquid. a δ18 O values from omphacite in low-strain rocks, mylonites and veins; fluid inclusion results from omphacite in low-strain rocks.

P. Philippot et al. / Earth and Planetary Science Letters 161 (1998) 33–44

Table 1 Summary of oxygen isotope and fluid inclusion data from Rocciavre´ [this work] and Monviso [14] massifs

P. Philippot et al. / Earth and Planetary Science Letters 161 (1998) 33–44

3.2. Analytical methods Samples were analyzed using preparation techniques and experimental conditions described in Nadeau et al. [14]. Oxygen in omphacite was extracted by bromine pentafluoride reaction at 600ºC. The standards NBS-28 and Circe 93 (a MORB glass) yielded values of 9:45 š 0:07‰ (1σ, n D 5) and 5:65 š 0:08‰ (1σ, n D 5), respectively. Oxygen isotopic ratios were measured on a delta E Finningan Mat(TR) spectrometer. Isotopic ratios are reported in the conventional δ-notation, in permil, relative to Standard Mid-Ocean Water (SMOW). Fluid inclusions in doubly polished sections were analyzed by microthermometry using a Fluid Inc. freezing– heating stage. Temperatures of final melting of ice (Tm .ice/ ) and of halite (Tm .halite/ / and temperatures of homogenization to the liquid phase (Th .L/ ) were reproducible to š0.2ºC and š1.0ºC, respectively. 3.3. Results Omphacite isotopic ratios obtained on omphacite clasts and vein omphacite and fluid inclusion data from omphacite clasts are listed in Table 1. The δ18 O values of omphacite clasts from low-strain metagabbros range from 5.2 to 6.7‰. Vein omphacite has a δ18 O value of 6.0‰. The first melting temperatures .Tmi / in the inclusions range between 20.4 and 25.4ºC; temperatures below 20.8ºC (the NaCl– H2 O eutectic) indicate the presence of additional components in the fluid, such as KCl ( 22.9ºC D eutectic temperature in the system KCl–NaCl–H2 O) and=or divalent charged cations such as Ca, Fe or Mg. On heating, halite crystals were generally the first phase to dissolve, between 201 and 380ºC. Interpreting these values using the NaCl–H2 O binary [38] yields salinities of 32 to 45 wt% equivalent NaCl. The liquid and vapour phases homogenize to liquid (Th .L/ / between 270 and 379ºC. In some inclusions, heating up to 500ºC led to the partial leakage of the fluid inclusions before complete dissolution of NaCl crystals.

37

4. Hydrothermal alteration imprint of oceanic material and their HP counterparts Fig. 1 shows an idealized section of the oceanic crust at a spreading center (modified after Nehlig [4] and references therein). Also shown is the thermal cycle, fluid salinity and a generalized oxygen isotopic profile through the entire section. Isotopic and fluid inclusion data of altered oceanic and ophiolitic rocks suggest that there is a major decoupling between the upper (volcanics C sheeted dike complex; SDC) and the lower (gabbroic layer or transition zone) hydrothermal systems, probably due to a drastic change in rock permeability. The upper system, or low-temperature alteration zone, is characterized by maximum temperatures of about 400ºC and low-salinity fluids (10 wt% NaCl). This system can be further subdivided into an upper extrusive layer that has reacted with large volumes of seawater at temperatures of <150ºC, and a lower intrusive (SDC) layer that has experienced somewhat more restricted fluid flow (e.g. [1,7]). The upper extrusive layer displays high δ18 OSMOW values between 5.2 and 14‰ (fresh basaltic magmas have δ18 O values of 5:7 š 0:2‰; e.g. [39]) whereas the intrusive layer shows a range of δ18 OSMOW values between about 3 and 10‰ [2,7,6]. The lower hydrothermal system, or high-temperature alteration zone, contains highly saline brines (typically from 40 to 60 wt% NaCl) sometimes in association with a low-density vapour phase, and is characterized by low δ18 OSMOW values (3.0 to 5.7‰, [2,5,7]). The changes in δ18 O values between the different hydrothermal systems reflect the temperature dependence of mineral–water δ18 O fractionations and the relative degree of rock alteration with depth. The saline brines may have exsolved directly from late-stage differentiated melts [4] or have formed during supercritical phase separation as the seawater penetrates the high-temperature alteration zone [3]. Variations in fluid salinity with depths is also reflected in the average Cl concentration of hydrous minerals. Data from most common hydrothermal amphiboles indicate that high-temperature alteration favors the formation of hornblendes containing a few tenths of a percent Cl [40,41], but that lower

38

P. Philippot et al. / Earth and Planetary Science Letters 161 (1998) 33–44

temperature promote formation of low-Cl actinolite [40,42]. However, as shown by Vanko [40], high-Cl amphiboles can also form at relatively low-temperature under low water=rock conditions. Accordingly, although the overall Cl concentration in oceanic hydrothermal systems is likely to increase with depth, the extent of fluid–rock exchange accompanying hydration reactions can have a strong control on the local Cl distribution in oceanic rocks. 4.1. Oxygen isotope signature Fig. 2 shows a compilation of oxygen isotope data from the literature for which fluid inclusions data are available (see Fig. 3). The metabasic rocks of Sifnos and Franciscan Complex consist of metavolcanic rocks (blueschists, eclogites, amphibolites) either interbedded with thin layers of metasediments as is the case in Sifnos, or occurring as exotic blocks within matrices of metasedimentary and meta-ultramafic rocks (Franciscan Complex). Although the history of these latter rocks may have been complex, it is likely that they originally formed as extrusive material where the interaction with seawater was maximum (see Margaritz and Taylor [19] for a discussion of alteration processes of Franciscan rocks). In contrast, as shown above, the eclogitic metagabbros from Monviso and Rocciavre´ are part of a coherent meta-ophiolitic sequence. The presence of a cumulate structure and of local diabase dikes clearly demonstrates that the rocks were part of the transition zone (gabbroic layer). The two main mechanisms that can affect oxygen isotopic ratios of rocks undergoing high-pressure metamorphism are fluid infiltration and devolatilization. The effects of fluid flow on δ18 O has been addressed in some detail by Taylor and Coleman [17] and Margaritz and Taylor [19] to explain the low δ18 O values of eclogitic rocks compared to their blueschist counterpart in the Franciscan Complex. According to [17] and [19], the original seawater involved in the hydrothermal alteration of the Franciscan rocks, relics of which are found in the San Luis Obispo ophiolites, became progressively enriched in 18 O during blueschist and eclogite facies metamorphism, evolving to values of C8‰ at the highest grades. This in turn resulted in a slight decrease of the bulk rock δ18 O values (see Fig. 2).

In the case of the Sifnos, Monviso and Rocciavre´ rocks, recognition of local-scale isotopic heterogeneities between different rock types [15] or microstructural domains (vein, mylonite, low-strain rock) within a single lithology [14] indicates that extensive fluid–rock exchange has not occurred during high-pressure metamorphism despite ongoing devolatilization. The high δ18 O compositions of the metabasic rocks from Sifnos and the low δ18 O values of the Monviso metagabbros have been attributed to a stage of hydrothermal alteration at low and high temperatures, respectively [15,14]. Eclogitic metagabbros from Rocciavre´ have δ18 O values of 5.1 to 6.7‰ which are similar to unaltered oceanfloor igneous rocks. Values as low as 5.1‰ and as high as 6.7‰ could be inherited from a stage of high- and low-temperature alteration, respectively. Eclogitic samples preserving a fresh basalt isotopic signature but also presenting evidence for low- and high-temperature alteration have been reported in the Zermatt Zone in the Alps [25]. Fractionation of oxygen isotopes during devolatilization is known to be small at temperatures of 400–600ºC and in all cases should be less than 1‰ [50]. Accordingly, the striking similarity between the isotopic composition of the hydrothermally altered oceanic and ophiolitic material and their high-pressure counterparts shown in Fig. 2 is likely to reflect to some extent the state of the rocks prior to subduction zone metamorphism. 4.2. Fluid salinities Fluid inclusion salinities expressed in weight percent equivalent NaCl (wt% NaCl) are plotted as a function of the δ18 O compositions of their host rocks in Fig. 3. In Monviso samples, δ18 O values correspond to analyses performed on the various types of omphacite in low-strain rocks, mylonites and veins. Taken as a whole, δ18 O values range from 3.0 to 5.3‰. Fluid inclusion results correspond to analysis performed on omphacite clasts in low-strain domains. On textural grounds, this fluid population is identical to the one described in Rocciavre´ samples. In contrast to Rocciavre´, however, no NaCl crystals have been identified in these inclusions. A salinity range of 17–21 wt% NaCl was estimated using ice melting temperatures (see [14] for further details). In the Cyclades high-pressure rocks, the isotopic

P. Philippot et al. / Earth and Planetary Science Letters 161 (1998) 33–44

39

Fig. 2. Oxygen isotope compositions of high-pressure metamafic rocks compared with their unmetamorphosed or hydrothermally altered equivalent. The data are from modern oceanic environment [2,5–7], Semail [43], Troodos [44] and San Luis Obispo [19] ophiolites, and Sifnos [15], Franciscan Complex [17–19], Monviso [14] and Rocciavre´ [this work] high-pressure terranes. The different symbols refer to different rock types. These include differentiated gabbros and plagiogranites (triangles), sheeted dike complex (vertical bars) and basalts (circles). These symbols correspond also to whole rock analysis whereas reversed triangles refer to analyses performed on pyroxene separates (augite in altered gabbros and basalts and omphacite in high-pressure metagabbros and metabasalts). Note the good correspondence between the bulk-rock versus clinopyroxene-separate δ18 O values. Oceanic and ophiolitic rocks are shown as empty symbols. High-pressure rocks (blueschists and eclogites) that have equilibrated at temperatures of 400 to 600ºC and pressures of 0.9 to 1.5 GPa are shown in black (Sifnos, Monviso, Rocciavre´ and part of the Franciscan Complex rocks; Type 4 metabasalts of Ward Creek, Marin County, Tiburon and Valley Ford eclogites). In the Franciscan Complex, some metabasalts (Type 1, 2 and 3 of Ward Creek and those of Laytonville) have equilibrated at lower PT conditions (200-325ºC and 0.6-0.7 GPa); these are shown with superposed colours (black C grey).

40

P. Philippot et al. / Earth and Planetary Science Letters 161 (1998) 33–44

Fig. 3. Inclusion fluid salinities against δ18 O(whole rock) or δ18 O(omphacite) in oceanic and high-pressure environments (see text). Fluid inclusion data are from DSDP [45,3,46,47] and Troodos [48] and Semail [49] ophiolites. In high-pressure environment, several processes have been proposed to explain the presence of high-salinity fluids. These include: (1) Immiscibility between non-polar fluid species (CO2 , N2 ) and aqueous brines [23], (2) eclogitization of anhydrous protoliths driven by hydration reactions [21], (3) Cl partitioning between a high-density melt phase and a residual CO2 -rich, low-density vapour phase [21,26], and (4) coeval hydration-decarbonation or hydration-dehydration reactions which deplete the fluid in water but not salt [28]. Such mechanisms for enhancing fluid salinity are not available in the Monviso and Rocciavre` eclogitic metagabbros.

data of [15] from Sifnos are combined with the fluid inclusion results of [16] from Syros Island. Fluid salinity values shown in Fig. 3 (4 wt% NaCl on average) correspond to the fluid inclusion population preserved in garnet cores in blueschist facies rocks; this population characterizes an early sample of the fluid phase attending high-pressure metamorphism ([16], p. 166). Recognition that the compositions of the fluids varies on a centimetre-scale between adjacent layers has led [16] to suggest that the blueschist rocks from Syros evolved in a closed system with little ability to interact with external fluid sources. For the Fransciscan rocks, we used the fluid inclusion results of [20] obtained on eclogitic blocks at Jenner, Tiburon and Mt. Hamilton in combination with the oxygen isotope data of [17–19]. Significantly, the fluid inclusion population described by [20] occurs as “non-planar clusters of elongated

tubes oriented parallel to the c-axes of host omphacite grains”. Texturally, these inclusions are identical to the ones described in Monviso and Rocciavre´ omphacite clasts, thus validating the Alpine-Franciscan comparison. Calculated salinities are lower than 5.2 wt% NaCl (average 3.1 wt% NaCl). In addition to the early inclusion fluids found in Franciscan blocks, [20] described secondary lowsalinity aqueous fluid inclusions trapped after highpressure metamorphism while the rocks were exhumed to the surface. Both primary and secondary inclusions have similar compositions [20]. Recognition that all samples contain a fairly homogeneous population of primary and secondary aqueous fluid inclusion has led [20] to propose that the rocks have interacted during and after high-pressure metamorphism with large volumes of externally derived fluids.

P. Philippot et al. / Earth and Planetary Science Letters 161 (1998) 33–44

5. Discussion 5.1. Fluid salinity–Ž18 O systematics A striking feature emerging from Fig. 3 is the similarity in the salinity–δ18 O systematics recorded between oceanic rocks and their high-pressure equivalent. This observation suggests that high-pressure metamorphism, either involving devolatilization alone as in the case of Monviso, Rocciavre` and Syros=Sifnos rocks, or devolatilization and extensive fluid infiltration as for the Franciscan setting, did not modify significantly the variability in chlorine concentrations documented along a typical section of the altered oceanic crust (Fig. 1). Giaramita and Sorensen [20] proposed that the salinity discrepancies recorded between Franciscan and Monviso inclusion fluids could reflect different fluid-rock regimes attending high-pressure metamorphism. In this model, mafic eclogites present in a sedimentary sequence along the Benioff zone have greater potential to interact on their burial and exhumation paths with large volumes of infiltrating fluids derived from material being devolatilized at greater depths. In contrast, coherent Alpine ophiolitic masses would represent material that has been devolatilized from within the slab with limited ability to interact with external fluid sources. This scenario is in agreement with the model developed by [11], which show that, even in the cold slab environment, up to 90% of the original water present in the altered oceanic crust is released prior to 40 š 10 km depths, to potentially travel along the top of the slab to higher levels. It cannot account, however, for the results obtained on Cyclades rocks, which display low-salinity brines and high δ18 O values typical of surficial extrusive material but preserve local-scale chemical and isotopic heterogeneities indicating lithological buffering of fluid activity during high-pressure metamorphism [15,16]. This together with the recognition that the Cyclades rocks occur as relatively undisrupted sequences preserving an original interlayering of metabasites and metasediments argue for a quite distinct tectonic setting than Franciscan terrane during subduction. As shown by [19], however, not all rocks in the Franciscan Complex were involved in the postulated fluid-controlled isotopic homogenization process; profiles across se-

41

quences of interlayered metabasites and metacherts show steps at contact, which indicate a lack of isotopic equilibrium between layers. Accordingly, it is likely that some of the well-preserved high-pressure rocks from Franciscan-like complexes represent material that has experienced devolatilization on the prograde path but limited interaction with external fluid sources on their exhumation history. 5.2. Net transfer of chlorine to the mantle It is important, when evaluating the Cl inventory of the deep mantle, to realize that the chlorine content of the rocks under consideration must be considered a net amount, without assumptions on what is expulsed in forearc regions and arc magmas. Indeed, Monviso and Rocciavre´ eclogites contain omphacite and garnet with no hydrous phases and thus have achieved complete devolatilization. The only fluid phase preserved in Alpine rocks occurs in fluid inclusions, which are thought to be derived from the breakdown of hydrous minerals such as amphibole and chlorite. The H2 O contents of lowstrain rocks from Monviso, due to the presence of fluid inclusions, is of the order of 2000–3000 ppm [14]. Such a small amount of H2 O (ca. 10% of pre-subduction water), is equivalent to that stored in nominally anhydrous mantle minerals (e.g. [51]) and hence could easily remain in the rocks down into the deep mantle. With regards to the Cyclades and Franciscan-type of high-pressure rocks, uncertainties arise from the presence of hydrous minerals which are likely to breakdown with ongoing metamorphism to produce H2 O, which in turn can dilute the Cl content of the inclusion fluids. However, primary fluid inclusions occurring in nominally anhydrous minerals like omphacite and garnet, some of which being texturally similar to the primary inclusions of Alpine eclogites (see above), are likely to escape further interaction with external fluids during progressive metamorphism. Accordingly, it is suggested that Alpine eclogites and possibly Cyclades and some of the well-preserved Franciscan rocks can be viewed as relevant analogues of the type of material to be transferred back into mantle. Using a typical water content of eclogites of 2000–3000 ppm and an average fluid salinity of 30 wt% NaCl for the gabbroic layer (2=3 of the oceanic

42

P. Philippot et al. / Earth and Planetary Science Letters 161 (1998) 33–44

crust) and of 4 wt% NaCl for the upper oceanic crust (1=3 of the oceanic crust), result in a net Cl content of the eclogitized crust of 151 to 226 ppm. Taking 20 wt% NaCl as an average fluid salinity for the gabbroic layer leads to an overall Cl content of 110 to 167 ppm. The estimated contents of subducted Cl and H2 O yield a Cl=H2 O ratio of 3.6–7.5 ð 10 2 , which is closer to the E-MORB ratio (3–10 ð 10 2 ; [32]) than to the N-MORB ratio (0.7–3.0 ð 10 2 ) and seawater ratio 2 ð 10 2 ). 5.3. Chlorine behaviour during devolatilization An important implication of the considerations above concerns the behaviour of chlorine during prograde metamorphism and ongoing devolatilization. Recent work on chlorine has shown that its isotopic composition may be modified by hydration and dehydration processes [33,52]. According to Ransom et al. [52], devolatilization reactions attending prograde metamorphism strongly fractionates chlorine isotopes, 35 Cl being more easily released to the fluid and the residual solid being enriched in 37 Cl. Considering that the Cl content of the primary fluid inclusions present in the different high-pressure rocks investigated represent a residual excess chlorine to be recycled into the mantle, simple conservation law written for chlorine isotopes implies that mantle chlorine is likely to be enriched in 37 Cl. Magenheim et al. [33] determined that the source of unaltered Mid-Ocean Ridge Basalt (MORB) glass has a stable chlorine isotope ratio of C4.7‰ relative to seawater. Thus, given the many uncertainties concerning Cl isotope fractionation, the amount of Cl loss required to generate a Cl isotopic composition of the subducted crust of C4.7‰ can be inferred from the chlorine mass balance equation Cloceanic crust D Clreleased fluid C Cleclogite and from the chlorine isotope mass balance equation δ37 Cloceanic crust D .δ37 Clreleased fluid ð Clreleased fluid C δ37 Cleclogite ð Cleclogite /=Cloceanic crust using an intermediate value of hydrothermal amphiboles of 2.1‰ [33] for δ37 Cloceanic crust , a value

of subduction-zone pore waters of 4.5‰ [52] for δ37 Clreleased fluid , and assuming that the chlorine content of eclogites (110–226 ppm D Cleclogite ) represents x% of the original chlorine of the altered oceanic crust (Cloceanic crust ). Results of the calculations show that a minimum of 70% of the Cl stored in the altered oceanic crust (Cloceanic crust D 157–322 ppm) is to be subducted to generate a δ37 Cleclogite of C4.7‰. Assuming that the volume proportion of fluid inclusions in altered oceanic rocks is not significantly different from that of eclogitic rocks (i.e., that both types of rocks contain the same amounts of fluid-inclusion Cl (110– 226 ppm, see Fig. 3)) implies that about 47 to 96 ppm Cl of the altered oceanic crust is stored in hydrous minerals such as amphibole. Interestingly, this estimates is essentially the same as that determined by Ito et al. [53] (50 š 25 ppm) on the basis of mineral proportion estimates from drill cores and dredged samples.

6. Summary (1) Typical eclogitic rocks that have achieved complete devolatilization can preserve small amounts of early dehydration fluids in the core of nominally anhydrous minerals such as omphacite. The relative proportion of salt in these fluid inclusions appears to be a function of the type of sea-floor hydrothermal alteration encountered by the rocks prior to subduction. Devolatilization processes during subduction apparently have little effect on the chlorine variability profile documented along a typical section of the altered oceanic crust. (2) We have estimated that a minimum of 100– 200 ppm chlorine could be transported to the deep mantle by the subducted oceanic crust. Eclogitic rocks bearing this chlorine can be regarded as the intermediate link between exogenic and mantle reservoirs and hence present great potential for better constraining the Earth’s distribution of chlorine. On the basis of available isotopic data, we have speculated that the amount of Cl loss to the exosphere during subduction to generate a substantial 37 Cl enrichment of the mantle relative to seawater should not exceed 30% of the Cl stored in the altered oceanic crust.

P. Philippot et al. / Earth and Planetary Science Letters 161 (1998) 33–44

Acknowledgements We thank Catherine Me´vel and David Vanko for discussions, David Vanko and Stuart Boyd for correcting various versions of the manuscript and for improving the prose, two anonymous reviewers for helpfull comments, and the CNRS-INSU for financial support (Contribution No. 110). [MK]

References [1] J.C. Alt, J. Honnorez, C. Laverne, R. Emmermann, Hydrothermal alteration of a 1 km section through the upper oceanic crust, Deep Sea Drilling Project hole 504B: mineralogy, chemistry and evolution of seawater–basalt interactions, J. Geophys. Res. 91 (1986) 10309–10355. [2] K. Muehlenbachs, Alteration of the oceanic crust and the 18 O history of seawater, in: J.W. Valley, H.P. Taylor Jr., J.R. O’Neil (Eds.), Stable Isotopes in High Temperature Geological Processes, Rev. in Mineralogy 16. Mineral. Society of America, Washington, DC, 1986, pp. 425–444. [3] D.A. Vanko, Temperature, pressure, and composition of hydrothermal fluids, with their bearing and magnitude of tectonic uplift at mid-ocean ridges, inferred from fluid inclusions in oceanic layer 3 rocks, J. Geophys. Res 93 (1988) 4595–4611. [4] P. Nehlig, Interactions between magma chambers and hydrothermal systems: oceanic and ophiolitic constraints, J. Geophys. Res. 98 (1993) 19621–19633. [5] E. Marion, Inte´ractions crouˆte oce´anique profonde — eau de mer dans les gabbros de la zone M.A.R.K., Ph.D. Thesis, Universite´ Paris 6, Paris, 1992, 159 pp. [6] P. Agrinier, C. Laverne, P. Tartarotti, Stable isotope ratios (oxygen, hydrogen) and petrology of hydrothermally altered dolerites at the bottom of the sheeted dike complex of Hole 504B, in: J. Erzinger, K. Becker, H.J.B. Dick, L.B. Stokking (Eds.), Proc. ODP, Sci. Results 137=140, College Station, TX, 1995, pp. 99–106. [7] P. Agrinier, R. He´kinian, D. Bideau, M. Javoy, O and H stable isotope compositions of oceanic crust and upper mantle rocks in the Hess Deep near the Galapagos Triple Junction, Earth Planet. Sci. Lett. 136 (1995) 183–196. [8] Y. Tatsumi, Migration of fluid phases and genesis of basalt magmas in subduction zones, J. Geophys. Res. 94 (1989) 4697–4707. [9] J.H. Davies, D.J. Stevenson, Physical model of source region of subduction zone volcanics, J. Geophys. Res. 97 (1992) 2037–2070. [10] S.M. Peacock, Large-scale hydration of the lithosphere above subducting slabs, Chem. Geol. 108 (1993) 49–59. [11] P. Philippot, Fluid–melt–rock interaction in mafic eclogites and coesite-bearing metasediments: Constraints on volatile recycling during subduction, Chem. Geol. 108 (1993) 93– 112.

43

[12] A.B. Thompson, Water in the Earth’s upper mantle, Nature 358 (1992) 295–302. [13] T.J. Ahrens, Water storage in the mantle, Nature 342 (1989) 122–123. [14] S. Nadeau, P. Philippot, F. Pineau, Fluid inclusion and mineral isotopic compositions (H–C–O) in eclogitic rocks as tracers of local fluid migration during high pressure metamorphism, Earth Planet. Sci. Lett. 114 (1993) 431– 448. [15] A. Matthews, M. Schliestedt, Evolution of the blueschist and grenschist facies rocks of sifnos, Cyclades Greece: a stable isotope study of subduction-related metamorphism, Contrib. Mineral. Petrol. 88 (1984) 150–163. [16] H. Barr, Preliminary fluid inclusion studies in a high-grade blueschist terrain, Syros, Greece, Mineral. Mag. 54 (1990) 159–168. [17] H.P.J. Taylor, R.G. Coleman, O18 =O16 ratios of coexisting minerals in glaucophane-bearing metamorphic rocks, Geol. Soc. Am. Bull. 79 (1968) 1727–1756. [18] D.E. Vogel, G.D. Garlick, Oxygen-isotope ratios in metamorphic eclogites, Contrib. Mineral. Petrol. 28 (1970) 183– 191. [19] M. Margaritz, H.P.J. Taylor, Oxygen, hydrogen and carbon isotope studies of the Franciscan Formation, Coast Ranges, California, Geochim. Cosmochim. Acta 40 (1976) 215– 234. [20] M.J. Giaramita, S.S. Sorensen, Primary fluids in low-temperature eclogites: evidence from two subduction complexes (Dominican Republic, and California, USA), Contrib. Mineral. Petrol. 117 (1994) 279–292. [21] T. Andersen, H. Austrheim, E.A.J. Burke, Fluid inclusions in granulites and eclogites rom the Bergen Arcs, Caledonides of W. Norway, Mineral. Mag. 54 (1990) 145–158. [22] P. Philippot, J. Selverstone, Trace-element rich brines in eclogitic veins: implications for fluid composition and transport during subduction, Contrib. Mineral. Petrol. 106 (1991) 417–430. [23] J. Selverstone, G. Franz, S. Thomas, S. Getty, Fluid heterogeneities in 2 GPa eclogites: implications for element recycling during subduction, Contrib. Mineral. Petrol. 112 (1992) 341–357. [24] S.R. Getty, J. Selverstone, Stable isotopic and trace element evidence for restricted fluid migration in 2 GPa eclogites, J. Metamorph. Geol. 12 (1994) 747–760. [25] A.C. Barnicoat, I. Cartwright, Focused fluid flow during subduction: oxygen isotope data from high-pressure ophiolites of the western Alps, Earth Planet. Sci. Lett. 132 (1995) 53–61. [26] P. Philippot, P. Chevallier, C. Chopin, J. Dubessy, Fluid composition and evolution in coesite-bearing rocks (Dora– Maira Massif, Western Alps): implications for element recycling during subduction, Contrib. Mineral. Petrol. 121 (1995) 29–44. [27] G. Bebout, Volatile transfer and recycling at convergent margins: Mass-balance and insights from high-P=T metamorphic rocks, in: G. Bebout, D.W. Scholl, S.H. Kirby, J.P. Platt (Eds.), Subduction Top to Bottom, Geophysical

44

[28]

[29]

[30] [31] [32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

P. Philippot et al. / Earth and Planetary Science Letters 161 (1998) 33–44 Monograph 96, American Geophysical Union, Washington, DC, 1996, pp. 179–193. M. Scambelluri, G.B. Piccardo, P. Philippot, A. Robbiano, L. Negretti, High salinity fluid inclusions formed from recycled seawater in deeply subducted alpine serpentinite, Earth Planet. Sci. Lett. 148 (1997) 485–499. M. Kastner, H. Elderfield, J.B. Martin, Fluids in convergent margins: what do we know about their composition, origin, role in diagenesis and importance for oceanic chemical fluxes?, Philos. Trans. R. Soc. London 335 (1991) 275– 288. A.T. Anderson, Chlorine, sulfur and water in magmas and oceans, Geol. Soc. Am. Bull. 85 (1974) 1485–1492. J.G. Schilling, C.K. Unni, M.L. Bender, Origin of chlorine and bromine in the oceans, Nature 273 (1978) 631–636. A. Jambon, B. De´ruelle, G. Dreibus, F. Pineau, Chlorine and bromine abundance in MORB: the contrasting behaviour of the Mid-Atlantic Ridge and East Pacific Rise and implications for chlorine geodynamic cycle, Chem. Geol. 126 (1995) 101–117. A.J. Magenheim, A.J. Spivack, P.J. Michael, J.M. Gieskes, Chlorine stable isotope composition of the oceanic crust: implications for the Earth’s distribution of chlorine, Earth Planet. Sci. Lett. 131 (1995) 427–432. B. Lombardo, R. Nervo, R. Compagnoni, B. Messiga, J. Kienast, C. Mevel, L. Fiora, G.B. Piccardo, R. Lanza, Osservazioni preliminari sulle ofiolite metamorfiche del Monviso (Alpi Occidentali), Rend. Soc. It. Miner. Petrol. 34 (1978) 253–305. U. Pognante, Magmatic and metamorphic evolution of two Fe–Ti gabbroic series from the Piemonte ophiolite nappe in the Susa valley area, Italian western Alps, Mem. Sci. Geol. Padova 35 (1981) 21–34. P. Philippot, H.L.M. van Roermund, Deformation processes in eclogitic rocks: evidence for the rheological delamination of the oceanic crust in deeper levels of subduction zones, J. Struct. Geol. 14 (1992) 1059–1077. U. Pognante, J.R. Kienast, Blueschist and eclogite transformations in Fe–Ti gabbros: a case study from the western Alps ophiolites, J. Petrol. 28 (1987) 271–292. R.J. Bodnar, Revised equation and table for determining the freezing point depression of H2 O–NaCl solutions, Geochim. Cosmochim. Acta 57 (1993) 683–684. K. Muehlenbachs, R.N. Clayton, Oxygen isotope studies of fresh and weathered submarine basalts, Can. J. Earth Sci. 9 (1992) 172–184. D.A. Vanko, High-chlorine amphiboles from oceanic rocks: product of highly-saline hydrothermal fluids?, Am. Mineral. 71 (1986) 51–59. D. Stakes, C. Me´vel, M. Cannat and T. Chaput, Metamor-

[42]

[43]

[44]

[45]

[46]

[47]

[48]

[49]

[50]

[51]

[52]

[53]

phic stratigraphy of Hole 735B, in: R.P. Von Herzen, P.T. Robinson et al. (Eds.), Proc. ODP, Sci. Results 118, College Station, TX, 1991, pp. 153–180. C. Laverne, D.A. Vanko, P. Tartarotti and J.C. Alt, Chemistry and geothermometry of secondary minerals from the deep sheeted dike complex, Hole 504B, in: J. Erzinger, K. Becker, H.J.B. Dick, L.B. Stokking (Eds.), Proc. ODP, Sci. Results 137=140, College Station, TX, 1995, pp. 167–190. R.T. Gregory, H.P.J. Taylor, An oxygen isotope profile in a section of Cretaceous oceanic crust, Samail Ophiolite, Oman: evidence for d18 O buffering of the oceans by deep >5 km) seawater-hydrothermal circulation at mid-ocean ridges, J. Geophys. Res. 86 (1981) 2737–2755. T.H.E. Heaton, S.M.F. Sheppard, Hydrogen and oxygen isotope evidence for seawater-hydrothermal alteration and ore deposition, Troodos complex, Cyprus, J. Geol. Soc. London 7 (1977) 42–57. D.S. Kelley, J.R. Delaney, Two-phase separation and fracturing in mid-ocean ridge gabbros at temperatures greater than 700ºC, Earth Planet. Sci. Lett. 83 (1987) 53–66. D. Vanko and D. Stakes, Fluids in oceanic layer 3: evidence from veined rocks, Hole 735B, southwest Indian Ridge, in: R.P. Von Herzen, P.T. Robinson et al. (Eds.), Proc. ODP, Sci. Results 118, College Station, TX, 1991, pp. 181–215. D.S. Kelley, K.M. Gillis, G. Thompson, Fluid evolution in submarine Magma–hydrothermal systems at the MidAtlantic ridge, J. Geophys. Res. 98 (1993) 19579–19596. D.S. Kelley, P.T. Robinson, J.G. Malpas, Processes of brine generation and circulation in the oceanic crust: fluid inclusion evidence from the Troodos ophiolite, Cyprus, J. Geophys. Res. 97 (1992) 9307–9322. P. Nehlig, Salinity of oceanic hydrothermal fluids: a fluid inclusion study, Earth Planet. Sci. Lett. 102 (1991) 310– 325. J.W. Valley, Stable isotope geochemistry of metamorphic rocks, in: J.W. Valley, H.P. Taylor Jr., J.R. O’Neil (Eds.), Stable Isotopes in High Temperature Geological Processes, Rev. in Mineralogy 16, Mineral. Society of America, Washington, DC, 1986, pp. 445–489. S.C. Kohn, Solubility of H2 O in nominally anhydrous mantle minerals using 1 H MAS MNR, Am. Mineral. 81 (1996) 1523–1526. B. Ransom, A.J. Spivack, M. Kastner, Stable Cl isotopes in subduction-zone pore waters: implications for fluid-rock reactions and the cycling of chlorine, Geology 23 (1995) 715–718. E. Ito, D.M. Harris, A.T. Anderson, Alteration of oceanic crust and geologic cycling of chlorine and water, Geochim. Cosmochim. Acta 47 (1983) 1613–1624.