The oceanic crust represents the most extensive igneous formation of the Earth crust even though the oldest basement of the present day ocean is only 165 Ma old. Since the time of its emplacement by accretion along the 60 000 km of active oceanic ridges which circumscribe the Earth, until its recycling into the mantle by subduction, the oceanic crust is affected by various alteration processes resulting from its interactions with seawater at seafloor conditions (i.e., ‘submarine weathering’), and seawater-derived fluids in hydrothermal conditions. The successive seawater/oceanic crust interactions are thought to play an essential role in participating to the world ocean budget and regulating the chemical composition of seawater. With 18×10²¹ kg of seawater, the world ocean water represents 94.7% of the total water reserve of the Earth. It is the largest reservoir of fluids involved in submarine water/rock interactions particularly as compared with the mid-oceanic ridge magmas, which only contain less than 0.4% of H₂O⁺. Hence magmatic volatiles cannot be considered as important contributors to submarine alteration processes of oceanic crustal rocks.

Occurrences of metamorphic rocks derived from the oceanic crust and upper mantle are widely distributed through the various geodynamic settings of the world seafloor. Samples of metabasalts, metagabbros or metaserpentinites have been recovered at numerous locations, e.g., by dredging from fault scarps of oceanic rift valleys or transform faults, and by drilling into the oceanic crust at ridge axes and flanks. Now we know that these metabasites result from various hydrothermal interactions between seawater-derived fluids and the rocks forming the oceanic lithosphere. Oceanic metabasites were found [40,124,151,207] long before submarine hydrothermal activity was directly observed at oceanic ridges in the late 1970s [45,64,182]. Serpentinites, ‘spilites’, amphibolites, ‘diabases’ and other metabasites were already known to be essential constituents of the ophiolite ‘trinity’, since the first definition of the term by Steinmann [187,188]. Ophiolite complexes are now recognized to represent sections of ancient oceanic lithosphere emplaced upon continental crust by dominantly thrusting mechanisms during plate collision ([43,148,149] and references therein). However, this article will exclusively deal with metabasites originated in the present day ocean floor, and ophiolites will only occasionally be quoted as comparison terms.

Thanks to the collaborative efforts of the various ongoing international programs such as ODP (Ocean Drilling Program), InterRidge, DORSALES, etc., we know that hydrothermal circulation through the oceanic crust is a widespread phenomenon related to the generation of new oceanic lithosphere by accretion at spreading centres and above hot spots. It is responsible for an important heat loss from the Earth's interior [186], and the resulting interactions between seawater and oceanic rocks which represent one of the major mechanisms regulating the chemistry of the ocean [1,58]. In addition, hot springs discharged on the seafloor supply energy and food to colonies of chemiothrophic organisms, whose symbiota are used as model for the origin of life in the Earth primitive ocean (e.g., [56,75]), and for the search of life in the solar system (e.g., [14,15]). It is now clear that anaerobic microbes exist within submarine hydrothermal systems supported by degassing and alteration reactions (e.g., [52,191]).

However, to date, it is still difficult to reconstruct the geometry of the fluid recharge and discharge pathways of hydrothermal circulation through the oceanic crust, to assess the relative importance of ridge axis vs. ridge flank hydrothermal circulations and the relative contribution of focused vs. diffuse fluid discharges, and to describe the evolution of a given submarine hydrothermal system with time. Finally, the geochemical fluxes and budgets resulting from the various alteration processes are still uncertain for most of the major and trace chemical elements.

It will be shown in this article that all of the oceanic metabasites are more or less intensely altered rocks that result from hydrothermal water/rock interactions between the oceanic crust and seawater-derived solutions. However an entire range of hydrothermal alteration intensities is encountered among the samples collected form the seafloor. On the one hand hydrothermally altered metabasalts and metagabbros can be the products of localized but intense and pervasive water/rock interactions in crustal locales such as stockworks proximal to the focused discharges of seawater-derived hydrothermal fluids. Such rocks are similar to hydrothermally altered host-rocks of land-based sulphide ore deposits. Their primary phases are almost totally replaced by secondary minerals, their igneous texture might be obliterated, and their chemical composition is strongly metasomatic. On the other hand, the so-called ‘ocean-floor metamorphism’ generally corresponds to non-pervasive and not so intense water/rocks interactions in widespread regions away from the focused hydrothermal upwelling zone. In such rocks the replacement of primary phases is generally only partial, igneous relict minerals and structures are quite abundant, and the chemical composition is not very different from that of their protoliths. The comparison between studies of the TAG (Trans-Atlantic Geotraverse) active hydrothermal mound of the Mid-Atlantic Ridge, and DSDP drilled Hole 504B into the 5.9-Myr-old flank of the Costa Rica Ridge (Equatorial East Pacific) leads to a better understanding of the wide range of alteration mechanisms of the oceanic crust, and of the spatial and temporal distributions of the various types of metabasalts associated with submarine hydrothermal activity. These two case histories are completed by similar examples drawn from the literature.

A series of special issues dedicated to various aspects of ocean floor hydrothermalism have been published up to the present time. They are actual mines of information about alteration systems and processes, their effects and products. They cannot all be listed here but it is highly recommended to consult the following monographs: Nato Conference Series, Marine Sciences [157]; The Canadian Mineralogist [14]; Economic Geology [156]; US Geological Survey Bulletin [138]; The Geological Society [147]; Geophysical Monographs [91]; The Royal Society [41].

The first oceanic metabasalts were collected from fault scarps associated with the crest of slow spreading oceanic ridges such as the Mid-Atlantic Ridge [124–126] or the Carlsberg Ridge, in the Indian Ocean [38,40,207]. They were called ‘greenstones’ or ‘spilites’, and initially interpreted as the result of  “burial-regional metamorphism rather than of localized hydrothermal alteration, or of autometamorphism” [125]. However, only two years later, Melson et al. [126] proposed that the same ‘greenstones’ from the Mid-Atlantic Ridge were the product of a “post-cooling (of the lavas) hydrothermal metamorphism”. Similarly, the ‘spilites’ form the Indian Ocean [38,40] were considered as “metamorphosed and hydrothermally altered rocks […] formed by the action of hot fluids on already crystalline rocks”. Oceanic metabasalts were then explained as the result of  “burial at a depth of about one kilometer or less, coupled with a rise in temperature to about 200 °C (which) would bring the rocks into the lower greenschist facies of metamorphism” [40]. Concluding about the ‘spilites’ from the Carlsberg Ridge, Cann [38] finally wrote: “these spilites were formed by low-grade metamorphism, and were neither the products of reaction of hot basalt with sea water nor that of the crystallization of a special spilitic magma […] (but they were) buried underneath later lava to be heated sufficiently for an assemblage corresponding to the lower greenschist facies to be formed”. Miyashiro et al. [134] and Miyashiro [135] coined the term ‘ocean-floor metamorphism’, which they regarded “as a kind of burial metamorphism in high heat-flow regions of the oceanic ridge crests”. The authors ascribed the metamorphic reactions to either the presence of igneous intrusions or the flow of hot fluid. In the latter case, the agent of ‘ocean-floor metamorphism’ would have been hydrothermal. In the most recent revised edition of Winkler's Petrogenesis of metamorphic rocks, Bucher and Frey included Miyashiro's ‘ocean-floor metamorphism’ in the ‘regional-extent’ type of metamorphism along with ‘orogenic’ (synonymous of ‘regional’) and ‘burial metamorphisms’ [36 (pp. 5–6)]. Nevertheless, the authors concluded the paragraph dedicated to this subject with the following statement: “This process [of ocean-floor metamorphism] leads to chemical exchanges between rocks and sea water; in this respect, ocean-floor metamorphism resembles hydrothermal metamorphism.” Indeed, ‘ocean-floor metamorphism’ is a case of hydrothermal alteration.

As a matter of fact, it had already been concluded, as early as 1973, that metagabbros [29], rodingites [78], and chalcopyrite mineralizations disseminated in greenschist metabasalts [30] dredged from the Mid-Atlantic Ridge resulted from a ‘contact hydrothermal metamorphism’ due to interactions between seawater and the oceanic crust beneath the axis of oceanic ridges, in the vicinity of axial magma chambers acting as heat sources. Similarly, Humphris and Thompson [88,89] proposed that the metabasalts from the median valley of the Mid-Atlantic Ridge at 4°S and 22°S latitude, which exhibit a secondary mineral assemblage similar of the parageneses of greenschist facies of regional metamorphism, resulted from hydrothermal alteration.

However, the equivalent to the hydrothermally altered host rocks surrounding the large land-based volcanogenic massive sulphide (VHS) ore deposits such as, for instance the propylitized, sericitized, chloritized, or argillized lavas, have been rarely collected from the present-day oceanic crust. Such rocks are generally pervasively altered with their primary igneous mineralogy completely replaced and their chemical composition totally changed. Quon and Ehlers [151] described one quartz+epidote rock sample with relict basaltic texture from the North Mid-Atlantic Ridge that could be called an epidosite, but the sample was never studied in detail, and has now been lost. Epidosites are rocks made up of epidote and quartz, with minor chlorite and pyrite. They are depleted in transition metals and generally lack the parental rock igneous texture. The first and only present-day ocean epidosites were dredged from the Tonga fore-arc [13]. The authors concluded that these epidosites metasomatically replaced basaltic and plagiogranite protoliths, and formed under similar conditions to epidosites found in ophiolite complexes. The latter are known but not widespread in various ophiolite complexes (e.g., [145,153,161]) where they occur as veins, mainly in the lower part of the sheeted dike complexes. According to Banerjee et al. [13], the tectonic setting would play a role in the formation of epidosites since both Tongan and ophiolite epidosites were collected from locales related to subduction zones, and none from mid-oceanic ridges. In addition to epidosites rare occurrences of chloritized or ‘sericitized’ metabasalts have been collected from active and inactive submarine hydrothermal fields in the present-day ocean. They will be described and their origin will be discussed in the last section of this paper. Intensely altered oceanic metabasites such as epidosites, and sericitized or chloritized metabaslts are present in the modern oceanic crust but they are very restricted in abundance and extent. Such rocks probably do not form a major part of the oceanic crust.

Both the widespread metabasalts and metagabbros ascribed to ‘ocean-floor metamorphism’, and the rare epidosites, chloritized or ‘sericitised’ basalts, and rodingitized gabbros result from hydrothermal interactions between the oceanic lithosphere and seawater-derived fluids. They differ in the degree of alteration and, hence, in the nature of the end products. The former are non-pervasively altered and abundant relict minerals subsist, whereas the latter are pervasively and intensively replaced. The former result from distal alteration processes away from the up-flow zone of hydrothermal fluids, whereas the latter result from proximal processes, in the vicinity of focused discharges of hydrothermal fluids.

Similarities between the mineral assemblages of oceanic metabasites and the parageneses of land-based metamorphosed rocks resulting from burial or regional metamorphisms has lead Eskola's metamorphic facies concept to be applied to seafloor and ophiolites metabasites. Ophiolites were commonly presented as vertically zoned sections of oceanic crust with a progressively increasing degree of metamorphism with burial depth in the complex, ranging from zeolite, to greenschist, and to amphibolite facies of metamorphism (e.g., [54,66,67,190]). It “led to the view of metamorphic facies boundaries laying more or less parallel to the original ocean floor […] The degree of alteration is more or less uniform and extensive” [39]. The term “brownstone facies” was even introduced by Cann [39] for oceanic basalts altered at low temperature in oxidizing conditions during ‘submarine weathering’, i.e., halmyrolysis, a non-metamorphic process, in the uppermost portion of the seafloor (for a review of low temperature alteration of oceanic basalts, see [76]). However, already in 1979, Stern and Elthon [189] presented a more refined model whereby “disequilibria retrograde assemblages” were noticed as being the common rule. It was also observed that basalts displaying the mineralogical assemblages characteristic of the prehnite–pumpellyite facies were rarely found in the seafloor, but were more frequent in ophiolites. Prehnite alone is common in oceanic metagabbros (e.g., [78,129]), but pumpellyite has only been observed three times in ocean-floor metabasalts [93,120,128]. Pumpellyite was associated with prehnite in two of these three oceanic occurrences.

In this paper, the expression ‘greenschist-like (or another) facies mineral assemblage’ will be used to designate a non-equilibrium mineral assemblage similar to the diagnostic mineral equilibrium (or near-equilibrium) assemblage, characteristic of the various facies of regional metamorphism.

The main arguments supporting that ocean-floor metabasites are the product of hydrothermal interactions between crustal rocks and seawater-derived hydrothermal fluids are the following.

• (1) In crustal regions with average heat flow, the shallow burial depth below the seafloor surface at which the oceanic metamorphic rocks were collected prevents high enough temperatures to be reached, which would correspond to the observed metamorphic mineral assemblages during burial-regional metamorphism. On the other hand, much higher heat flows are locally measured along the ridge crests or ridge flanks as compared to those normally measured in deep basins away from spreading centres.
• (2) Non-equilibrium mineral assemblages are generally observed in oceanic metabasites. In opposition to Cann's experience, who claimed that “it is possible to conclude from the close spatial juxtaposition and wide development of secondary phases that they represent some sort of equilibrium assemblage” [39], the presence and often, the abundance, of primary basaltic relict minerals along with assemblages of secondary minerals belonging to various parageneses characteristic of different metamorphic facies, indicate that non equilibrium conditions prevailed during replacement (see also [189]).
• (3) In addition, it is observed that the metamorphic assemblages are generally restricted to open space fillings such as cracks, vesicles and vugs, and or alteration halos replacing rock adjacent to veins. Multiple openings and in-fillings of the veins are also commonly observed.
• (4) Finally, it will be shown later that the chemical compositions of hydrothermal solutions collected from present-day ocean floor hot springs, clearly indicate hydrothermal interactions between crustal rocks and seawater-derived fluids.

These observations indicate that the sea floor water/rock interactions were related to recurrent but short-lived circulations of exogenous hydrothermal fluids along open fractures, and that they did not produce equilibrium metamorphic mineral precipitations in the voids nor complete recrystallizations of the host rocks.

One can conclude that the oceanic metabasites are not the result of the classical burial-regional metamorphism, which leads to the replacement of protolites by equilibrium mineral assemblages. Hence Eskola's concept of metamorphic facies cannot be directly applied to the oceanic crustal context. In addition, it can be concluded that seafloor metabasites are not derived from primary ‘spilitic’ magmas since the metabasalts result from post-eruptive seawater-MORB reactions in hydrothermal conditions. This was the ultimate argument that put an end to the theory of the primary origin of spilites (e.g., [9] and references therein).

Miyashiro et al. [135] and Miyashiro [134] noticed that the vast majority of the oceanic metabasites are non-schistose and the degree of cataclasis not extreme except for a small number of ‘gneissic metagabbros’ from the Mid-Atlantic Ridge at 30°N (already mentioned by Quon and Ehlers [151] and Thayer [195]) which were explained as the products of “contact metamorphism in the lower crust or the upper mantle”. It was shown later (Miyashiro, in litt. and unpublished) that these ‘gneissic metagabbros’ were ice-rafted rocks of Precambrian age probably derived form the Labrador continental shield. However, actual ‘gneissic amphibolites’ exist in the oceanic crust [81] (see below).

On the other hand, strongly deformed metabasites and metaserpentinites were collected from transform faults that intersect and offset the slow spreading mid-oceanic ridges. For instance, gneissic amphibolites from the Vema Fracture Zone (Mid-Atlantic Ridge at 11°N) were demonstrated to result from the complete recristallization of Layer-3 gabbros at temperatures comprised between 400 and 650 °C, and with low water/rock ratios [81]. They are only about 10 Myr old and result from a dynamic, contact metamorphism of gabbros in the lower oceanic crust (Layer 3) that completely re-equilibrated in the conditions of the amphibolite facies along an active transcurrent fault [81]. Similarly, mylonites, ultramylonites and phyllonites were recovered, along with dominant, undeformed or cataclastic metabasites, from the Romanche Fracture Zone that offset by 950 km the Mid-Atlantic Ridge in its equatorial portion. The deformations took place under the conditions of the greenschist to amphibolite facies [77,79]. However, strongly deformed and recrystallized oceanic metagabbros do not appear to be restricted to transform faults where transcurrent tectonics generates crustal shear zones. Metagabbros collected from the other geotectonic settings of the seafloor display cataclastic textures.

Mével and Cannat [131] studied gabbros samples collected from slow spreading ridges by dredging or by submersible (e.g., the MARK area in the Mid Atlantic Ridge South of the Kane Fracture Zone) or through low-recovery drilling (e.g., DSDP hole 334 [75], and DSDP hole 556 [130]) which do not provide continuous, in situ sections of the lower oceanic crust. For such gabbro samples, no correlation between rock deformation textures and outcrop tectonic features can be established. The authors also studied gabbro samples from ODP hole 735B (in the southwestern Indian Ocean), which supplied them with a more continuous set of samples [55]. Mével and Cannat [131] observed that the metagabbros from ODP hole 735B are remarkably similar to the dredged gabbros with respect to their secondary mineralogy, and texture. Even though their study samples were collected in the vicinity of transform fractures zones, Mével and Cannat [131] claimed that “the geometry of deformational structures in these samples (inferred from submersible observations or from the drill cores) is not compatible with strike-slip deformation in the transform fault. Therefore, we infer that these deformational structures resulted from tectonic processes occurring at the ridge axis.” The metagabbros “are often plastically deformed […] from slightly deformed with abundant magmatic porphyroclasts in a recrystallized matrix, through gneissic, to truly mylonitic.” Mével and Cannat [131] presented a three-stage evolutionary model of the deformation and alteration of the gabbros forming the deeper crustal portion of slow-spreading ridges, i.e., during periods of low magma supplies (see Section 6.3). According to these authors, oceanic stretching of the gabbros starts right after their emplacement at or close to the spreading axis where synkinematic ductile deformation generates shear planes at temperatures ranging from 800 to 900 °C, i.e., in anhydrous conditions similar to those of the granulite facies. When the crust spreads away from the ridge axis, oceanic gabbros are successively deformed and altered as the rock temperature decreases and continuing shearing associated with synkinematic ductile cracks allow seawater derived fluids to penetrate and react with the rocks at temperature ranging from 750 to 500 °C, i.e., similar to the amphibolite facies conditions [185,201]. Later and farther away form the ridge axis, seawater penetrates and reacts with the gabbros, at temperatures lower than 500 °C, along brittle deformation cracks of Lister's cracking front (see Section 6.2). Mével and Cannat [131] proposed that the amount of ductile shearing and related high-temperature alteration in oceanic Layer 3 might well be a function of the spreading rate.

Various types of metabasites have been commonly recovered from taluses along the lower walls of the axial rift valley of slow spreading ridges, or from the numerous transform faults that intersect the latter. The oceanic metabasites include metabasalts with mineral parageneses resembling those of the greenschist facies, and metagabbros, rodingites and metaserpentinites with mineral assemblages similar or nearly identical to the diagnostic greenschist to amphibolite facies. In fast-spreading ridges, on the other hand, metamorphic rocks are only known from the deepest transform faults (e.g., the Garret Fracture Zone [19]), or peculiar tectonic settings such as the Hess Deep [68,103,121], and drill sites related to hydrothermal fields ([18] and ODP Sci. Contr. Legs 139 and 168). Hence the recovery of oceanic metabasites was thought to be generally restricted to strongly faulted tectonic settings of oceanic ridges until metabasalts with mineral assemblages like those of greenschist to zeolite facies were recovered by drilling DSDP Hole 504B into the 5.9-Myr-old flank of the Costa Rica Ridge, in the Equatorial East Pacific (see Section 7). Metamorphic rocks collected by drilling into ridge flanks or by dredging along fracture zones have probably been laterally transported by seafloor spreading away from the accretion zone at ridge axes. Later large vertical tectonic movements characteristic of slow spreading ridges, such as the Mid-Atlantic Ridge and the Carlsberg Ridge in the Indian Ocean, would bring up the metamorphic rocks to exposures where they can be collected. As a consequence, many authors thought that oceanic metabasalts were restricted to normal or transcurrent fault scarps associated with slow spreading ridge tectonics. On the other hand, the relatively smooth topography of fast spreading ridges prevents the deep-seated metamorphic rocks from outcropping. Presumably, metamorphosed crustal rocks are present beneath all of the ridges but the contrasted tectonically induced topographies of the two types of ridge easily explain the dissimilar abundances of the metamorphic rock occurrences. However other parameters such as the crustal permeability related to the primary morphology of the lavas (e.g., pillow vs. lobate vs. massive vs. sheet flows [107]), or to posteruptive ridge tectonics, probably play an important role in controlling the extent of hydrothermal fluid circulation, hence of the associated alteration. Therefore it is suggested that ocean-floor metamorphism has taken and is still taking place today at some depth beneath the seafloor, particularly where new crust is created by seafloor spreading and the temperature gradient is greater than the average gradient measured elsewhere. As a consequence it is thought that metamorphic rocks represent an important and probably a major component of the oceanic crust.

Before ocean-floor hydrothermal vents were directly observed, hydrothermal activity related to oceanic ridge ‘volcanoes’ had already been suggested as responsible for the precipitation of Fe, Mn-rich metalliferous sediments found on the flanks of the East Pacific Rise [28,31,180]. Lister [111] had argued for the presence of hydrothermal circulation in the oceanic crust on the basis of heat flow measurements transects across active ridges. For instance, across the Explorer Ridge spreading zone, in the NE Pacific, heat flow minima of 1.3–1.9 HFU are localized at the axis of the ridge valley, whereas the adjacent ridge flanks present maxima of 6.6–8.4 HFU, which progressively decrease to the normal lower values farther away form the spreading centre. Lister [112–114,116] attempted to demonstrate that hydrothermal circulation of fluids with temperatures in the range of the transition from greenschist to amphibolite facies, i.e., from 200 to 400 °C, within the oceanic crust is theoretically possible from the mechanics of propagating thermal shrinkage cracks in cooling igneous bodies near magma chambers at the ridge axis. Lister already recognized that crustal permeability is the key parameter controlling oceanic ridge hydrothermal circulation (see Section 6.2 below).

Submarine hydrothermal activity was directly observed at oceanic ridges only in the late 1970s [45,64,182], but hydrothermal water/rock interactions had already been suggested as the mechanism forming the metabasites and disseminated sulfide mineralizations recovered from oceanic ridges (see paragraph 1.1 and references in [30]). Early models had been proposed (e.g., [27,44]), according to which seawater percolating through ridge basalts along thermal contraction cracks reacted with the still hot rocks and mobilized by dissolution as chloride complexes of elements such as Mn, Fe, Co and the rare earths, etc. Corliss [44] concluded that “these solutions may be the metal-bearing hydrothermal exhalations” (responsible for) “the pelagic (i.e., metalliferous) sediments”, and that his “model is similar to one often proposed for the origin of hydrothermal ore solutions on the continents.” Whereas only Mn, Fe oxihydroxides precipitated onto the seafloor upon discharge of the hydrothermal solutions, transition metal sulphides were thought to form at depth within the crust, in altered host rocks containing secondary mineral assemblages similar to the greenschist facies parageneses [30].

The temperature of the hydrothermal solutions were said to decrease by conductive cooling and mixing with cold seawater along the discharge pathway towards the surface. However, the penetration depth into the crust, the geometry of hydrothermal circulation cells, the actual composition and temperature of the hydrothermal solutions, and length of their lifetime were totally unknown. The reason for the relatively late discovery of hydrothermal hot springs, ‘black smokers’ and sulphide precipitates is related to the strategies and tools used to find them. The first direct observations and samplings of hydrothermal activity and its products were carried out using submersibles on the Galapagos Spreading Centre and the East Pacific Rise, at sites selected on the basis of prior detailed heat flow and bathymetric surveys. The studies of active hydrothermal fields have brought a ten-fold increase in our understanding of submarine hydrothermal systems and stimulated the study in the laboratory of hydrothermal water–rock interactions between seawater and MORB in an attempt to duplicate natural oceanic hydrothermal systems. These experiments were carried out at temperatures ranging from 25° to 425 °C, and pressures up to 500 bar [19,21,25,97,127,141–144,159,170–173,175,177,178]. These experiments demonstrated that the black smoker fluids result from MORB-seawater interactions in the conditions of the greenschist facies (i.e., temperatures ranging from 250 to 400 °C, and at pressures corresponding depths of 1.5–2 km beneath to the oceanic ridges). They duplicated the secondary mineral assemblages identified in the natural samples of hydrothermally altered rocks collected form the seafloor. They also emphasized the role of the water–rock ratio during the interactions (e.g., [143]). Finally elemental fluxes and budgets were inferred from the laboratory experiments, e.g., the trapping of 50% of the Mg⁺⁺ river input to the ocean [57] was explained by the crystallization, in the end products, of secondary Mg-silicates such as actinolite and chlorite already observed in the natural samples. Similarly, the Si, alkalis, and transition metal fluxes measured in the black smoker fluids were verified in the laboratory experiments. A very important side effect of these studies, besides several PhD theses (e.g., [70,139,169]), was the design of new laboratory equipments to carry out the hydrothermal experiments (e.g., [176]). Moreover, the thermodynamic behaviour of seawater itself had to be studied [22,25,26] in order to understand the various cases of hydrothermal systems observed on the oceanic ridges: e.g., the roles of adiabatic expansion [20] and phase separation on metal transport [23] in seafloor hydrothermal system, or supercritical two-phase separations [26,49–51] vs. near-critical conditions [23,24,174].

6.1 The causes of oceanic hydrothermal circulation, the motor of the flows of matter and thermal energy, the thermal engine

6.2 Oceanic ridge hydrothermal systems

6.3 Hydrothermal circulation models with or without Layer 3?

6.4 Elemental fluxes: what comes out, what stays behind?

DSDP/ODP Hole 504B is presently, and will probably remain for a long time, the only reference section in the modern ocean crust and, in the next section, its alteration history will be compared to that of the basement under the TAG active hydrothermal mound. Let us summarize our observations in DSDP Hole 504B [7].

The record penetration depth of 2111 m into the oceanic basement was reached in DSDP Hole 504B as the result of drilling efforts during DSDP legs 69, 70, and 83, and ODP legs 111, 137 and 148. The oceanic crust of this site was generated 5.9 My ago by the Costa Rica Ridge, an intermediate spreading ridge with a 3.25 cm/y half spreading rate, located in the East Equatorial Pacific (see Fig. 3a). Heat flow measured on the seafloor at the site is about 200 mW/m² corresponding with the temperature of 59 °C measured at the sediment/basaltic basement interface [10] under a 274.5 m thick cover of pelagic sediment. As a consequence, abnormally precocious cherts and limestones are present in the lowermost portion of the relatively young sediment cover. Under the sediments, the basaltic basement is formed by, from top to bottom (see Fig. 3b): 571.5 m of pillow lavas with few massive flows and intraflow basaltic breccias, grading through a 209 m thick lithologically transitional zone (in which effusive basalts are progressively replaced downward by doleritic dykes), and ending with a 1056 m thick sheeted dyke complex (100% dykes). No gabbro was recovered from Hole 504B. On the other hand, various seismic experiments, down hole loggings, and physical property measurements indicate that the transition between crustal Layers 2 and 3 seismic velocities is located at about 1500 m bsf±200 m, i.e., near the middle of the sheeted dyke complex [53]. Hence, it was demonstrated for the first time, that the Layer 2–Layer 3 boundary in the oceanic crust does not correspond to a lithological transition from basalts and dolerites to gabbros, but to a change in the physical properties of the crust (e.g, seismic wave velocities) probably related to secondary mineral formation due to hydrothermal alteration. In hole 504B the boundary between Layers 2 and 3 is therefore an alteration front.

The alteration history of a given section of the oceanic crust between the time of its emplacement to that of its sampling by drilling or dredging, could be reconstructed by deciphering the superimposed depositional sequences of secondary mineral assemblages in altered crustal rocks. The uppermost 310 m of basaltic basement was affected by halmyrolysis, a low temperature (i.e., <30 °C) oxidizing alteration process by the down-going seawater of a possible diffuse recharge limb [80]. During this alteration, the estimated water/rock ratios were large, i.e., several 10s. Seawater percolated down through the highly permeable lavas of the upper oceanic crust by following open fissures, microfractures, and interpillow spaces. The secondary minerals characteristic of the ‘submarine weathering’ of surficial oceanic basalts are celadonite-glauconite, saponite and Fe-oxihydroxides (the latter two minerals often combine to form a mineraloid called ‘iddingsite’) which replaces most of the primary olivine, whereas basaltic plagioclase and augite are usually unaltered. Primary titanomagnetite was already partly replaced by maghemite during an earlier stage of alteration, occurring probably while and/or briefly after the lavas erupted on the seafloor (see discussion in [106,209–211]). As a consequence of the oxidative alteration, the basalt cores display red halos, a few mm to a few cm in thickness, concentric with exposed surfaces such as open fissures. The latter were filled with the following secondary mineral sequence forming veins: saponite along the walls, then phillipsite, and calcite or aragonite at centre. Minor pyrite and marcasite specks are sometimes aligned in the host rock, just beyond the oxidative front of the red halos. Fresh basaltic glass was found in pillow lava rims down to the lithological transition to the dykes where effusive rocks disappear.

A intermediate alteration zone is found in the lowermost lava section (i.e., about 250 m thick) lying between the overlying zone of halmyrolyzed basalts, and the underlying hyrothermally altered dolerites (see next paragraph). The main secondary minerals in the lavas of the intermediate alteration zone are, sequentially: saponite, pyrite and calcite filling up open fissures and vesicles, and forming the cement of basaltic breccias. Saponite is also found in the basalts replacing olivine and glass. Such an assemblage reflects relatively low temperature (i.e., <60 °C), anoxic conditions during alteration at low water/rock ratios.

The lowermost kilometre of the doleritic dyke complex contains the higher temperature secondary minerals characteristic of reducing hydrothermal alteration at variable but generally low water/rock ratios (i.e., < or = 1) by upwelling hydrothermal fluids [5,82]. The latter pervaded the rock permeability which is much smaller in the dykes than in the surficial basalts altered at lower temperature. Hydrothermal fluids also have to follow open fractures if one wants to explain the high fluid flows measured in black smokers which adiabatically discharge the solutions on the seafloor at rates of several metres per second. The secondary minerals include albite partly replacing primary plagioclase, chlorite totally replacing olivine, actinolite and chlorite partly to totally replacing augite, and titanite totally replacing titanomagnetite. The secondary mineral assemblage corresponds to the conditions of the greenschist facies of regional metamorphism, and oxygen isotope analyses [7] indicate alteration temperatures ranging from 220 to 380 °C. However, primary mineral relicts (more particularly plagioclase and augite) are common, showing that equilibrium was not reached during alteration even though replacement is sometimes almost complete in the lowermost dyke section.

The downhole variation in the secondary phyllosilicate mineralogy (see Fig. 3b) appears to be related to both alteration temperature and altering fluid chemistry. The halmyrolyzed lavas of the upper low temperature alteration zone contain celadonite and saponite, but in the transitional alteration zone is found saponite which progressively becomes, with increasing depth, a saponite-chlorite irregular mixed layer mineral. Finally, in the dyke section, pure chlorite is dominant but minor irregular mixed layer saponite-chlorite is often observed. The regular saponite-chlorite mixed layer corrensite is rare and was observed in only 8 samples from lower half of Hole 504B (Alt, unpublished; [179]).

From the study of Hole 504B, one can conclude that, at any given place and time, two circulation cells with opposite directions generated three superimposed alteration zones in the oceanic crust, i.e., from basement-sediment interface to the deepest known portion of dolerite dyke complex:

• (1) The surficial layer, a few 100s of m thick, of upper crust basalt halmyrolyzed by down-welling cold seawater;
• (2) The deepest section of dolerites hydrothermally altered by upwelling hot hydrothermal solutions;
• (3) An intermediate basalt section 250 m tick where anoxic alteration probably resulting from the mixing of hydrothermal fluids ascending from the ‘reactor zone’, with downwelling seawater derived fluids after they reacted at low temperature with the surficial basalts.

As the crust spreads away form the ridge axis, it cools down and new permeability is created, hence progressively lower temperature secondary mineral assemblages successively precipitate, mainly in veins filling up open cracks [5,7]. The secondary mineral sequences include: saponite, phillipsite and calcite in the halmyrolyzed upper section of the crust; saponite, pyrite and calcite in the transitional alteration zone; and, in the doleritic section, chlorite, actinolite and locally quartz, sulfides or minor epidote, during a first high temperature hydrothermal alteration stage, followed by zeolites, prehnite and calcite during a progressively decreasing temperature alteration stage. The latter is probably still active at the present time, as both the cooling of the crust in a conductive regime, and today's temperature of 193 °C at the bottom of Hole 504B, result from a reheating due to the sealing of the basaltic basement by the sedimentary cover.

The geologic setting of the TAG (Trans-Atlantic Geotraverse) hydrothermal field and active mound was reviewed by Humphris et al. [92], and will be only briefly summarized here. The TAG hydrothermal field is located at 26°N on the Mid-Atlantic Ridge (i.e., a slow spreading ridge with a 1.25 cm/y spreading rate), and covers an area of 25 km² near the base of the east wall of the median valley (see Fig. 4a). Low temperature hydrothermal activity, with Mn-oxihydroxides and nontronite precipitates, was discovered in 1972 [154,168] between 2400 and 3100 m water depth on the east wall of the rift valley. Two inactive, partly oxidized sulfide deposits also occur on the lower east wall NE of the median valley, i.e., the MIR and ALVIN zones in Fig. 1. U/Th dating indicates that the inactive deposits range in age from 5×10³ to 14×10³ years [108]. The active TAG mound, discovered in 1985 [158], is located at 26°08′N and 44°49′W at a water depth of about 3600 m (Fig. 1). It lies 1.5–2.0 km east of the spreading axis, on a crust that is about 100×10³ years old. The mound is circular (see Fig. 4b), roughly 200 m in diameter and 30–50 m in height (see Fig. 2 after Kleinrock et al., [104]), and consists in 4 to 5×10⁶ tons of pyrite-, quartz-, and anhydrite-rich breccias. The mound is composed of two terraces: an upper terrace at ∼3644 m and a lower terrace at ∼3650 m. A complex of several chalcopyrite-rich black smoker chimneys discharging hydrothermal fluids at temperatures of up to 368 °C is located to the NW of the centre of the upper terrace. On the lower terrace, the Kremlin white smoker complex is comprised of several small (1–2 m) anhydrite, pyrite and sphalerite-rich, but chalcopyrite-poor, chimneys that eject 260–300 °C fluids. These fluids are thought to be derived from black smoker-type fluids cooled by mixing with seawater, and conductive cooling within the mound [59,96,132,197]. As a consequence of the mixing (see Section 6.2) pyrite, quartz-opal, and anhydrite precipitate within the mound where large fractions of Cu are lost, whereas most of the Zn precipitates as sphalerite close to the mound-seawater interface [96].

The pyrite+silica+anhydrite breccias are thought to result from the repeated mass wasting of highly unstable black smokers chimneys, and the recurrent hydrofracturing of previous hydrothermal precipitates [90]. Both matrix- and clast-supported breccias are observed. Steep slopes around the upper terrace are the result of mass wasting which has exposed material rich in silica (quartz and amorphous silica)+pyrite+Fe-oxihydroxides, and formed an apron which consists of mixtures of pelagic sediment and hydrothermal detritus. Diffuse flow of low-temperature (up to 50 °C) fluids, i.e., ‘diffusers’, occurs through much of the mound surface and the surrounding apron [133] whereas locally, e.g., on the northwest side of the mound, low heat flow suggests recharge of cold seawater into the mound [16].

Such a large accumulation of hydrothermal precipitates as the TAG mound is the result of the long lasting but intermitting discharge of hydrothermal fluids focused in a restricted locale. This, in turn, requires that upwelling hydrothermal fluids follow the same, or recurrently opened pathways, to the seafloor. The TAG mound is thought to be located at the intersection between axis-parallel faults and transverse or oblique faults [102,104]. In addition, U/Th dating of sulfide samples from the TAG mound suggests that hydrothermal activity has occurred episodically over a period of 40 000 to 50 000 years [108]. Such a long duration of hydrothermal activity at the TAG deposit is explained through recurrent reactivation of the faults through seafloor spreading related tectonic activity. The fracture permeability allowing repeated hydrothermal discharges at the same location over periods of time of tens of thousand years such as TAG, are thought to be maintained by fault propagation and interaction [48]. A near-bottom magnetic survey of the TAG area indicated that a magnetic low lies directly beneath the mound which was interpreted as a cylindrical alteration pipe or stockwork dipping to the south, and marking the shallow hydrothermal up-flow zone [196].

During ODP Leg 158, 17 holes were drilled into the TAG mound down to a maximum sub-bottom depth of 125 m. Petrographic and mineralogical studies [83] of drilled cores clearly show (see Fig. 4b) that alteration of basement under and in the mound resulted in a vertically and laterally zoned stockwork which is approximately 100 m in diameter and extends at least 125 m below the mound surface. Hydrofracturing of the basaltic basement by hydrothermal solutions was (and probably still is) the main mechanism, which brecciated the basaltic basement, creating new pathways for the fluids to rise to the surface. The cores recovered by drilling beneath and within the TAG mound do not appear to have been drilled through massive lavas but through a pile of angular basaltic debris or a talus.

We reconstructed the sequence of alteration processes of crustal rocks beneath and within the mound [83]. The first stage of hydrothermal alteration is the chloritization of the basaltic basement beneath the mound as a result of interaction with Mg-bearing hydrothermal solutions or mixtures of seawater with the ‘zero Mg’ hydrothermal fluids similar to that which is presently discharged by the black smoker complex, near the centre of the mound. Chloritization occurred at temperatures ranging from 250 to 370 °C. In the deepest part of the stockwork underlying the central part of the mound, i.e., only 25 m away from the centre of the upwelling zone of hot fluids, the basalt is almost completely chloritized with minor quartz and pyrite. Cr-spinel is the only unaltered primary mineral. Beneath the mound margins, i.e., less than 70 m away from the hot fluid upwelling zone, the chloritization process is restricted to mm- to cm-thick halos parallel to the exposed surfaces of basalt fragments. The second stage of hydrothermal alteration is the progressive replacement of basalt and chloritized basalt by paragonite+quartz+pyrite. At the end of this stage, all of the primary minerals are replaced. Paragonite forms, via replacement of chlorite in the stockwork host rocks, and as a result of direct precipitation from hydrothermal fluids in interstitial spaces of the quartz+pyrite breccias. As a result of the recurrent silicification and pyritization, basaltic minerals and textures are progressively obliterated until completely replaced by mixtures of ‘dirty’ quartz (i.e., quartz containing submicroscopic to microscopic inclusions of paragonite, anatase, etc.), pyrite and interstitial paragonite. The last stage of hydrothermal mineralization consists of anhydrite precipitating in open spaces in the breccias such as veins and voids.

Among the metabasites recovered in the cores drilled into the TAG Mound during ODP Leg 158, can we distinguish between the products of hydrothermal alterations resulting from distal vs. proximal water/rock interactions, respectively? Are these processes related to one and the same hydrothermal system as we can see it today? Before answering these questions, we will briefly describe the major types of metabasites recovered from the TAG Mound.

• – Stage 1 chloritized metabasalts: Chloritized metabasalt fragments were recovered from several drill sites in the active TAG hydrothermal mound, both from deep holes located beneath the central part of the mound (and hence close the focused discharge of hot hydrothermal fluids), and shallow holes located at the rim of the mound (and hence about 100 m away form the central discharge). The chloritized metabasalts recovered in the lower part of the deepest hole under the central part of the active mound have been totally replaced by a chlorite+quartz+pyrite assemblage, and the Cr-spinel is the only relict primary phase. Chlorite is by far the dominant secondary mineral replacing the groundmass clinopyroxene, the olivine and plagioclase microphenocrysts, and only partly the plagioclase microlites. Chlorite also filled up vesicles, miarolitic voids, and thin open fractures. Quartz mainly replaced the plagioclase microlites. The primary intersertal structure of the parental sparsely phyric basalt is preserved. Veins of chlorite+quartz+pyrite (in sequential order of deposition) ranging from <1 mm to >10 mm in thickness crisscross the metabasalts. On the other hand, chloritized green halos, only <1–2 cm in thickness, were observed parallel to the exposed surfaces of basalt fragments in cores from distal drill holes into the margin of the mound, i.e., less than 70 m away from the hot fluid upwelling zone. Quartz is not present in these halos, and pyrite is almost completely oxidized, giving the halos a brown hue, and indicating that chloritization was followed by oxidative halmyrolysis by cold seawater. The rock from the inner part of these basalt fragments does not contain any hydrothermal minerals, but exhibits evidence of halmyrolysis: olivine microphenocrysts are partly replaced by smectite (saponite?) sometimes associated with Fe-oxihydroxides (iddingsite) whereas plagioclase microlites and the augite rich groundmass appear to be unaltered. Chlorite under the SE mound margin is Mg-rich and formed via the reaction of basalt with heated seawater. In one core basaltic glass shards were totally replaced by the most Mg-rich chlorite known from the ocean floor [83]. On the other hand chlorite under the NW margin of the mound and in the deep, central chloritized stockwork is poorer in Mg and formed from black smoker-like hydrothermal fluids mixed with small amounts of seawater [3,83,193,194].

Hence, during the first alteration stage, similar hydrothermal processes affected the basaltic basement at TAG both close to and away from the main upwelling zone of hydrothermal solutions, but they varied in intensity, and in the chemistry of the secondary products, i.e., chlorite, varied according to the composition of the altering fluids and, hence the distance to the up-flow zone of hydrothermal fluids. Under the central part of the hydrothermal mound the chloritization process lead to the total replacement of the basalt, whereas under the mound margins the chloritization process only slightly affected the outer portion of the basalt fragments. The variable chemistry of chlorite reflects the variable degree of mixing between Mg-poor black smoker fluid and Mg-rich seawater. Both the variable intensity of chloritization and variable composition of chlorite seem to be related to the distance from the focused discharge pathway of the hydrothermal solutions.

• – Stage 2 paragonite+quartz+pyrite replacement of basalts: A detailed X-ray diffraction and transmission, and TEM study [84] indicates that the TAG paragonitic mica is a mixture between a pure paragonite and an ordered irregular mixed-layer paragonite/dioctahedral smectite. This mixed-layer phyllosilicate has never been identified before. Paragonitic micas reflect the particularly high Na/K ratio of the present-day hydrothermal fluids discharged by the TAG black smokers at temperatures of up to 370 °C [37,59,65]. The TAG paragonitic mica often contains Cr and Ti but contamination by microscopic to submicroscopic inclusions of anatase (identified by X-ray diffraction) and Cr-oxides could be responsible for the high Cr and Ti contents measured by electron microprobe. The paragonite+quartz+pyrite assemblage, which characterizes the second stage of hydrothermal alteration, is clearly overprinted onto the already chloritized metabasalts resulting from the first alteration stage, and is restricted to the central part of the mound. Therefore the second hydrothermal alteration stage coincides with the main upwelling of undiluted, high temperature hydrothermal fluids generating the black smokers at the top of the mound. Therefore we can conclude that, at TAG, both chloritization and paragonitization are two successive hydrothermal alteration processes. Chloritization affected a much larger volume of basement than paragonitization. Chlorites seem to result from interactions between basalts and hydrothermal fluids which can be mixed with seawater, at temperatures in the lower part of the 250–370 °C range. On the other hand paragonite formed later at the highest temperatures, and from undiluted black smoker fluids. It is restricted to the central zone of hydrothermal fluid upwelling.

According to Saccocia and Gillis [151] secondary mineral assemblages in hydrothermal breccias indicate a relationship between fluid chemistry and rock alteration in the up-flow zones of hydrothermal fluids through the upper oceanic crust. Based on a mineralogical and fluid inclusion study of breccia samples from the MARK area (Mid-Atlantic Ridge at 23°40′N lat.) and the Hess Deep (East Pacific Rise-Galapagos Spreading Center triple junction) Saccocia and Gillis [151] proposed a classification in two types of the breccias found in the oceanic crust:

• ‘Type 1 breccias’ (e.g., from the MARK area) characterized by the Fe-chlorite, quartz, pyrite, and anatase assemblage would be produced by hot, high salinity fluids (i.e., 3.3–10 wt% NaCl), relatively depleted in H₂S and enriched in Fe. This type of breccia is consistent with the ‘M-type alteration pipes’ containing Fe-chlorite, quartz, pyrite, and anatase of alteration pipes in the Cyprus Ophiolite complex [152].
• ‘Type 2 breccias’ (e.g., from both the MARK area and Hess Deep) characterized by the Mg-chlorite, epidote, quartz, pyrite, and titanite assemblage would be produced by hot, low salinity fluids (<0.1–5.6 wt% NaCl), relatively rich in H₂S, and enriched in Mg. In both cases, the fluid temperature was ranging 160–270 °C according the fluid inclusion study. The second type of breccia is not exactly related to Richards' et al. [152] “P-type alteration pipes” containing Mg-chlorite, quartz, pyrite, and anatase, because the latter also contains illite (K-rich fluids?), but no epidote. According to Saccocia and Gillis [151], the observed salinity variations are explained by a supercritical phase separation of seawater-derived hydothermal fluids at temperatures ranging from 450–500 °C to 550–625 °C in the Hess Deep and the Mark area, respectively. The authors predicted that Type1 alteration assemblages should exist in the up flow zones of relatively Cl-rich and H₂S-poor fluids, such as that of the TAG hydrothermal system. However, this prediction is not confirmed by our study of the core material drilled during ODP Leg 158 into the stockwork beneath and within the TAG mound. Honnorez et al. ([83], see this article, Section 6.2) observed a lower zone of earlier chloritized alteration zone characterized by a Mg-chlorite+quartz+pyrite assemblage, followed by an upper zone of later alteration zone characterized by a paragonitic phyllosilicate+quartz+pyrite+anatase assemblage without any chlorite. In contrast, the paragonitic mica observed by Honnorez et al. [83] is not present in Saccocia's and Gillis's ‘type 1’ breccias, whereas the epidote characterizing Saccocia's and Gillis's ‘type 2 breccia’ was not found in the Mg-rich chlorite bearing altered basalts from the lower zone of the TAG mound.

Illite and illite/smectite mixed-layer clay minerals appear to be common in land-based volcanogenic massive sulfide deposits and associated stockworks where the phyllosilicates are generally interpreted as the hydrothermal replacement products of rock fragments (e.g., [101,152]). The common occurrence of illite and illite/smectite (I/S) mixed-layer clay minerals in oceanic sediments is interpreted as resulting from diagenesis and/or burial metamorphism, or hydrothermal alteration of smectite which progressively recrystallizes according the classic sequence: smectite → random illite/smectite mixed-layer→regular illite/smectite mixed-layer→illite [183,184]. The proportion of illite and the degree of ordering in the I/S phyllosilicates are thought to increase with temperature.

In contrast occurrences of ‘sericitized’ metabasalts have rarely been described from active and inactive submarine hydrothermal fields in the present-day ocean. The name ‘sericite’ is placed between quotation marks because, in most cases, the mineralogy of the fine-grained, white mica is poorly defined. Na- and or K-rich micas have been identified in stockworks beneath a few volcanogenic massive sulfides deposits on land [198]. On the other hand, several occurrences of ‘sericitized’ metabasalts were found where K-rich white micas are associated with ophiolite-hosted sulfide deposits (e.g, [152,212]). In ancient crust both epidosites and ‘sercitized’ metabasalts appear to be proximally located with respect to hydrothermal fluid up-flow zones (see Section 2).

Paragonitic micas were documented in oceanic basalts of the Izu-Bonin-Mariana fore-arc region, in the western Pacific [8]. The lowermost basement of the 41 My old crust drilled during ODP Leg 125 is made up of an altered boninite-andesite sheeted dyke complex. The rocks are variably recrystallized to chlorite, smectite, chlorite/smectite mixed-layer minerals (including corrensite), quartz, albite, K-feldspar, and rare magnetite. The 5 m thick section of basement, near the bottom of the drill hole is intensely hydrothermally altered and mineralized, but it does not contain smectite or chlorite/smectite mixed-layer minerals. It contains ‘sericitized’ 1–1.5 cm thick halos adjacent to quartz veins where the rock is replaced by “mixtures of paragonite, mixed-layer mica-smectite, and pyrophyllite”, associated with quartz, albite, K-feldspar, and minor chlorite. Microprobe analyses of the mica mixtures indicate dioctahedral aluminous micas containing small amounts of Mg, and subequal amounts of K and Na. It is not certain whether these white mica mixtures consist of submicroscopically intergrown K- and Na-rich phases, or metastable intermediate NaK mica [8]. Pyrite containing small inclusions of pyrrhotite, chalcopyrite and sphalerite is disseminated in both the host chloritized rocks and the sericitized halos.

Metabasalts containing white micas diversely called ‘sericite’, ‘illite’, or ‘muscovite’ have been rarely described from active and inactive submarine hydrothermal fields in the present-day ocean and, in general, they only represent minor or trace constituents of the altered basalts An ‘illitic mineral’ was found to partly replace plagioclase in basalt fragments associated with the Green Seamount suflide deposit, 12 km off the East Pacific Rise [6]. Mg-rich chlorite, beidellite or saponite also participated to the replacement of the basalt with titanite and pyrite, but a single clay mineral was present within any individual basalt fragment. Rectorite, a regular illite/smectite mixed-layer clay mineral (I/S) was identified in the inactive massive sulfide deposit from the Green Sea Mount where it fills pore spaces between quartz and sulfide where it was followed by opal [2]. The I/S phyllosilicate represents on average 3% (maximum 18%) of the modal composition of the quartz+sulfide+phyllosilicate ore samples. It was interpreted as having directly precipitated from hydrothermal solutions. The I/S phyllosilicate with R1-type ordering contains 25–45% smectite. One sample also contained discrete smectite. In addition I/S replaces plagioclase phenocryts in metabasalt fragments (see above [6]). O isotope analyses indicate that the I/S formed at temperatures of 220 or 270 °C assuming formation from 0% seawater or a +2% enriched hydrothermal fluid, respectively. On the same basis, prior quartz and later opal would have formed at temperatures of 230 or 320 °C and 70 or 170 °C, respectively [6]. Green Sea Mount I/S phyllosilicates have similar chemical composition to those related to Kuroko massive sulfide deposits (i.e., they fall on the muscovite-pyrophyllite side of Hower and Mowatt, [87], ternary diagram), but different from those of diagenetic origin in marine sediments (i.e., the latter are more ‘celadonitic’ and fall near the centre of Hower and Mowatt, [87], ternary diagram). Authigenic clay particles with a ‘sericitic’ chemical composition have been identified in the altered host rock of a chloritic vein in metabasalts from DSDP Hole 504B [109]. The white mica was associated with albite partially replacing the primary plagioclase phenocysts, whereas the olivine phenocrysts were replaced by chlorite. In the last two cases ‘sericite’ or ‘illite’ were not identified by X-ray diffraction but on the basis of the chemical compositions obtained by electron microprobe analyses.

Well crystallized boehmite and anatase, and irregular mixed-layer chlorite-Al-rich (most likely a bedeillitic) smectite along with discrete chlorite and smectite were respectively identified from outer and inner 0.2–1 mm thick alteration crusts replacing two basalt samples dredged from the Gorda Rise [86]. In the inner crust minor quartz and phillipsite, possibly lizardite and kaolinite, and traces of talc were also found. The Al-rich secondary minerals are thought to result from the residual Al-enrichment (not the precipitation from Al-rich hydrothermal fluids) during the formation of the crusts. Minor amounts of boehmite were also identified in hydrothermal precipitates and surficial alteration crusts on basalts from the same general area [85], and from 8 km south of this area [42]. Unfortunately, all of the basalt samples studied by these authors were dredged from taluses which crumbled from the east wall of the ridge axial valley. Hence their age is unknown, even though the authors think that they are relatively young because of the freshness of the glassy pillow rims [85]. Mg-rich minerals, i.e., the discrete chlorite and smectite, and a now disappeared mixed-layer chlorite–smectite, are considered to have formed during an initial hydrothermal alteration stage of the basalts in seawater-dominated conditions. Then, during a second stage, the temperature of the fluid rose and leaching by the extremely low pH solutions prevailed. As a result, extensive leaching of all elements except Al and Ti promoted the formation of boehmite and anatase in an outer crust, whereas a beidellitic smectite–chlorite mixed-layer clay mineral replaced the first formed clays in the inner crust. However, the relationship between the study samples and any active hydrothermal vents was not demonstrated. The high temperature conditions of alteration were inferred from the similarity with other cases of Al-rich clay minerals described in the literature as resulting from high temperature hydrothermal alteration. They were mainly compared to the Al-rich phyllosilicates identified in alteration crusts on oceanic basalts collected from the East Pacific Rise [73]. Two samples were collected by submersible from an hydrothermally altered outcrop about 1.5 km away from an active black smoker vent field, near the axis of the East Pacific Rise, at 21°N. One sample was coated with X-ray amorphous material, whereas the other one was covered by a 0.5–0.75 mm thick crust of white phyllosilicate whose surface was stained by Fe-oxide compounds and coated with a scattering of tiny pyrrhotite crystals. Detailed mineralogical study indicated that the mineralogy of the hydrothermal alteration layer changes from impure, partly dehydrated beidellitic smectite adjacent to the basalt to a mixture of beidellitic smectite and randomly interstratified Al-rich chlorite/beidellite (90% smectite, 10% chlorite mixed layer), plus some discrete chlorite and possible minor amesite in the exterior of the crust. The basalt from the interior of the sample, i.e., away from the alteration crusts, appeared to be fresh but its$+6.5‰\delta$¹⁸O value indicates a slight, low-to-moderate temperature. According to Haymon and Kastner [73] the hydrothermal clay minerals formed by complete replacement of the basaltic glass and plagioclase microphenocrysts within 1 mm of the surfaces exposed to the venting of black smokers. The hydrothermal fluids were extremely reducing, acidic, and Mg-poor, and they interacted with the basalt and normal seawater. The Al-rich phyllosilicates formed at temperatures estimated to be in the 290–360 °C range, and may be considered as residual phases remaining after intense basalt leaching.

All of these seafloor occurrences of Al-rich phyllosilicates have been interpreted as being related to black smokers hydrothermal activity, and having formed from relatively high temperature (i.e., in the 220–360 °C range) solutions with a high degree of supersaturation. The Al-phyllosilicates resulted either from the replacement of basalt fragments or direct precipitation with other hydrothermal minerals such as pyrite, quartz and opal. Such cases confirm the mobility of Al and Ti during rock/water interactions in hydrothermal conditions when solutions are very acidic and reducing. However one should keep in mind that such phyllosilicates are probably metastable phases which disappear after the hydrothermal activity which generated them ceases.

Zierenberg et al. [213] describe the formation of a silicified cap-rock through the silicification/precipitation of siliceous crusts covering and cementing large areas of basaltic talus associated with an active hydrothermal field. The Sea Cliff hydrothermal field is located along a rift-bounding normal fault on the northern segment of the Gorda Ridge. The maximum temperature of 247 °C of the active field was measured in chimneys made up of anhydrite, with minor sulfides and amorphous silica but most venting at the Sea Cliff field is diffuse and at much lower temperature.

Two samples of strongly altered basaltic hyaloclastite collected from a talus covering a large normal-fault scarp were found to be altered by hydrothermal solutions in a mixing zone between ascending hydrothermal fluid and entrained seawater. During an initial alteration stage both crystalline and glassy basaltic clasts of the hyaloclastites were replaced by a Mg-rich smectite, i.e., stevensite, mixed-layered smectite/chlorite, and talc, at temperature of about 220 °C as inferred from O isotope analysis. High water/rock ratio prevailed during the Mg metasomatism that lead to the extreme leaching resulting in the removal of even the least soluble elements such as Al and Ti. More advanced alteration promoted silicification of the already altered basaltic clasts. The hyaloclastite cement is mainly amorphous silica and chalcedony, with traces of barite and pyrite. ‘Gosts’ of dissolved crystals in the cement were ascribed to anhydrite by the authors. Silicification proceeded at temperature ranging from 110 to 85 °C. Rare clasts of ‘massive sulfide’, i.e., mainly pyrite with minor sphalerite and chalcopyrite, are associated with the siliceous crusts. Finally microbial mats partly preserved in cavities near the upper surface of the crusts contain unusual sulfides and sulfosalts such as pearcite (Ag_14.7−XCu_1.3−XAs₂S₁₁), proustite (Ag₃AsS₃), besides more normal chalcopyrite and galena. It suggests that microbial activity might have precipitated Ag sulfides and sulfosalts in the siliceous crusts.

In this paper we have proposed that ocean floor metamorphism and hydrothermal alteration are two end members of a wide range of water–rock interactions under hydrothermal conditions, with varying degrees of intensity and under different conditions. They respectively correspond to distal and non-pervasive, vs. proximal and pervasive cases of hydrothermal reactions between oceanic crustal rocks and discharges of seawater-derived solutions.

Two main case histories were chosen to illustrate the two end-members, and they were compared to other published examples of alteration processes generating similarly altered oceanic rocks. On one hand the studies of metabasalts and metadolerites in the more than 2 km long drill Hole 504B of DSDP/ODP, and of metagabbros in ODP Hole 735B represent the reference studies of the non-pervasive and distal hydrothermal alteration (i.e., so-called ‘hydrothermal metamorphism’) through Layer 2 and Layer 3 of the oceanic crust, respectively. Hydrothermal processes diffusively affect Layer 2 rocks of the upper oceanic crust, at temperatures ranging from 400 to 150 °C, or locally transform the gabbros of the deeper oceanic Layer 3 along fissures, at temperatures ranging from 800 to 450 °C. Metabasalts and metagabbros are ubiquitous components of the oceanic crust resulting from such non-pervasive hydrothermal processes. Such metabasites are so commonly and largely widespread on the present-day sea floor that they probably form an important part of the oceanic crust.

On the other hand strongly replaced basalts (e.g., epidosites, sericitized or chloritized basalts), and gabbros (e.g., rodingites), are scarcely found in the oceanic crust. The hydrothermal alteration process which pervasively and intensively affects the crustal rocks at temperatures ranging from 250 up to 370 °C, is located in close vicinity of hydrothermal fluid up-flow zones. Therefore hydrothermally replaced basalts are probably not significant crustal constituents because they are too restricted in space to the immediate vicinity of focused discharges of hydrothermal waters. The horizontally and vertically zoned distribution of hydrothermal alterations within and beneath the TAG active hydrothermal field and mound, near the axis of the Mid-Atlantic Ridge, illustrates a typical example where the basaltic stockwork beneath a 4.5 Mton sulfide deposit has been totally hydrothermally replaced by quartz+pyrite+paragonite assemblages within 25 m of the up-flow zone of the active black smokers, and strongly chloritized deeper and farther away form the upflow zone. Analogous alteration zonations are well known around land-based VMS (Volcanic-hosted Massive Sulfide deposits) that are mined all over the world as Cu, Zn, Au and/or Pb ores.

In summary major differences exist between the metabasalts or metagabbros affected by the weaker and widespread ‘ocean-floor metamorphism’, and the intensely replaced and spatially restricted rocks collected near active or fossil upwelling zones of hydrothermal discharges, but all result from various degrees of intensity of hydrothermal interactions between oceanic crustal rocks and seawater derived hydrothermal solutions.

Even though the secondary mineral assemblages of hydrothermally altered oceanic rocks are very similar to their regional metamorphic counterparts, Eskola's concept of metamorphic facies cannot be applied to hydrothermally altered oceanic rocks because they are almost always the products of transient open systems whose temperature, solution chemistry and flow (hence the water/rock ratio) keep rapidly evolving through time. Metastable assemblages corresponding to two or more successive generations of secondary minerals can often be observed within the same hydrothermally altered rocks, and they generally coexist with primary basaltic relicts.

The degree of high temperature alteration of crustal rocks by hydrothermal fluids, and the amount of ductile deformation may be related to the spreading rate of the oceanic ridge that generates the crust. At fast spreading ridges, the absence of high temperature ductile shear zones prevents any water/rock interactions at temperatures higher than that of Lister's ‘cracking front’, i.e., 500 °C, generating permeability through brittle fracturing. In contrast ductile synkinematic deformation in slow spreading ridges allows water/rock interactions at temperatures up to 800 °C in the gabbros forming the deeper Layer 3 of the oceanic crust.

Oceanic metabasalts exceptionally exhibit deformation and schistosity, indicating that static conditions generally prevailed during the hydrothermal process. It also confirms that hydrothermal alteration is dominated by circulation of fluids along primary permeability pathways such as thermal contraction cracks, interpillow or inter-clast voids, etc, or as a result of hydrofracturing. On the other hand, oceanic metagabbros can display textures due to synkinematic ductile deformations without metamorphic recrystallization, in anhydrous conditions, at temperatures ranging from 900 to 800 °C. They grade into gneissic amphibolites resulting from recrystallizations in presence of high water partial pressure, at temperatures ranging from 750 to 500 °C. Finally, when lower crustal temperatures are reached (i.e., <300 °C), the gabbros are deformed in brittle conditions, and secondary minerals of the greenschist facies assemblages similar to those of the metabasalts form in open fissures and their host rocks. It appears that layer 2 is mainly cooled by hydrothermal convection which alters the basalts and dolerites at medium to low temperatures, under static conditions. Conversely layer 3 is cooled by conduction and its gabbros are altered, at first, at high temperature, during plastic deformation, and later at medium temperature, under static conditions.

Finally, later fissures opened as the plate moved become the path-ways of seawater dominated solutions from which carbonate-rich veins are precipitated at temperatures <150 °C in the superficial basalts within 10–15 My of the crust formation. In the gabbros recovered in DSDP/ODP Hole 735B, carbonate-rich veins formed at temperatures >150 °C when the crust was unroofed, and then at 10 °C when the crust was uplifted by tectonic movements related to the near Atlantis Transform Fault. The carbonate veins are not properly hydrothermal in nature.

The ‘reactor’ at the base of the hydrothermal system at the ridge axis is thought to be located in the root zone of the dykes within the upper gabbros representing the top of axial magma chambers at about 2 km depth, i.e., in the transition crustal region between layers 2 and 3. This is where the conductive boundary is supposed to have prevailed during the peak of the hydrothermal activity.

The various cases of hydrothermal alteration of oceanic crustal rocks presented in the paper do not include metabasites from transform faults which can be strongly deformed and recrystallized into mylonites, ultramylonites and phyllonites during recurrent polymetamorphic events related to shearing along transcurrent fracture zones such as the Romanche Transform, in the Equatorial Atlantic. In addition, hydrothermal processes, end-products and systems discussed in the present paper are restricted to oceanic crust generated at or near spreading axes occurring in present-day oceans. Ancient crustal material (i.e., ophiolites) and present-day off-axis processes (e.g., hot spots) were not considered in the review.