PUBLICATIONS Geochemistry, Geophysics, Geosystems RESEARCH ARTICLE 10.1002/2013GC005148 Key Points: Serpentinization occurs along thermal, tectonic, and reactioninduced fractures Serpentinization may occur quasi instantly at exhumation time scales The depth and extent of serpentinization are variable in space and time Supporting Information: ReadMe Table S1_v3 text_Auxiliary_material Fig_annexe_opx_reviews Fig_annexe_EBSDmaps Correspondence to: S. Roumejon, [email protected] Citation: Roum ejon, S., and M. Cannat (2014), Serpentinization of mantle-derived peridotites at mid-ocean ridges: Mesh texture development in the context of tectonic exhumation, Geochem. Geophys. Geosyst., 15, 2354–2379, doi:10.1002/2013GC005148. Received 13 NOV 2013 Accepted 19 MAY 2014 Accepted article online 23 MAY 2014 Published online 10 JUN 2014 Serpentinization of mantle-derived peridotites at mid-ocean ridges: Mesh texture development in the context of tectonic exhumation phane Roume jon1 and Mathilde Cannat1 Ste 1 Equipe de G eosciences Marines, Institut de Physique du Globe de Paris, Sorbonne Paris Cit e, UMR7154 CNRS, Paris, France Abstract At slow spreading ridges, axial detachment faults exhume mantle-derived peridotites and hydrothermal alteration causes serpentinization in a domain extending more than 1 km next to the fault. At the microscopic scale, serpentinization progresses from a microfracture network toward the center of olivine relicts and forms a mesh texture. We present a petrographic study (SEM, EBSD, and Raman) of the serpentine mesh texture in a set of 278 abyssal serpentinized peridotites from the Mid-Atlantic and Southwest Indian Ridges. We show that serpentinization initiated along two intersecting sets of microfractures that have consistent orientations at the sample scale, and in at least one studied location, at the 100 m scale. We propose that these microfractures formed in fresh peridotites due to combined thermal and tectonic stresses and subsequently served as channels for serpentinizing fluids. Additional reaction-induced cracks developed for serpentinization extents <20%. The resulting microfracture network has a typical spacing of 60 mm but most serpentinization occurs next to a subset of these microfractures that define mesh cells 100–400 mm in size. Apparent mesh rim thickness is on average 33 6 19 mm corresponding to serpentinization extents of 70–80%. Published laboratory experiments suggest that mesh rims formation could be completed in a few years (i.e., quasi instantaneous at the plate tectonic timescale). The depth and extent of the serpentinization domain in the detachment fault’s footwall are probably variable in time and space and as a result we expect that the serpentine mesh texture at slow spreading ridges forms at variable rates with a spatially heterogeneous distribution. 1. Introduction The oceanic lithosphere formed at slow (<4 cm/yr) [e.g., Small, 1998] spreading ridges contains a variable amount of serpentinized peridotite in the first kilometers. These rocks, associated with gabbroic bodies, have been widely observed and sampled along the Mid-Atlantic Ridge [Karson et al., 1987; Cannat et al., cia et al., 2000], the Gakkel ridge [Michael et al., 2003], and the 1992, 1995b, 1997; Kelemen et al., 2004; Gra Southwest Indian Ridge [Dick, 1989; Dick et al., 2003; Seyler et al., 2003; Sauter et al., 2013]. They are exhumed along large offset normal faults, also called detachment faults [Karson, 1990; Cannat et al., 1992, Cann et al., 1997; Lavier et al., 1999; Smith et al., 2006]. Microseismicity distribution [Toomey et al., 1985; deMartin et al., 2007] reveals that the detachment faults extend down to at least 8 km on axis and geological observations show that the emergence angle of these faults at the seafloor is locally <10 [Smith et al., 2006]. Together, these results suggest that axial detachments have a concave-downward shape, corresponding to a flexure of the footwall as it is exhumed (Figure 1). Paleomagnetic data acquired on drilled samples are also consistent with rotations of 46 –80 in the footwall of Mid-Atlantic Ridge detachments [Garces and Gee, 2007; Morris et al., 2009; MacLeod et al., 2011]. Axial detachment faults can therefore be sketched as conveyor belts for the exhumation of mantle-derived peridotites (Figure 1): these rocks are progressively uplifted from the base of the brittle lithosphere toward shallower levels in which hydrothermal circulation and serpentinization are active. Serpentinization has significant geodynamic and environmental consequences. It modifies the rheology of the oceanic lithosphere [Reinen et al., 1994; Escartın et al., 1997] and may help localizing the deformation along fault planes. Serpentinization also consumes large volumes of water and produces hydrogen [Frost, 1985; Charlou et al., 2002], part of which combines with dissolved CO2 [Berndt et al., 1996] to produce methane anomalies in the water column [Charlou and Donval, 1993; Charlou et al., 1998]. These hydrogen and methane fluxes provide energy to sustain microbial systems within the substratum and at hydrothermal ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2354 Serpentinization (%) 0 50 100 dredge 10.1002/2013GC005148 ridge axis ~5 km drill hole crust 2-4km Vp mantle footwall rotation lt fau nt me he tac de S Fig12 8 km domain of axial hydrothermal circulation exhumation 0 2 4 6 8 Vp (km/s) Figure 1. Conceptual sketch of detachment faulting and serpentinization of mantle-derived peridotites at a slow spreading ridge, drawn after Cannat et al. [2010]. The detachment fault acts as a conveyor belt bringing peridotites up from the base of the brittle lithosphere and into the domain of hydrothermal circulation and serpentinization. Serpentinization is accompanied by a decrease in density and seismic velocities. (left) Due to large footwall rotation [e.g., Morris et al., 2009], crustal thickness determined by seismic experiments in the ridge flanks [after Canales et al., 2000] translates into on-axis distance from the fault. Available sampling of the exhumed ultramafics is restricted to near-seafloor levels and includes dredges, submersible sampling, and drilling (open rectangles). vents [Shock and Holland, 2004]. The heat released on axis by serpentinization reactions may also partly fuel hydrothermal circulations [Kelley et al., 2001], although this effect is probably limited to low-temperature hydrothermal systems [Lowell and Rona, 2002; Allen and Seyfried, 2004]. Serpentinization developed at midocean ridges or later on in the hinge zone of subducting oceanic plates [Ranero et al., 2003; Contreras-Reyes et al., 2008] also influences the rheology of subduction zones [Hilairet et al., 2007; Hirauchi et al., 2010] and prograde metamorphism of subducted serpentine induces fluid and chemical transfers between the slab and the mantle wedge [Ulmer and Trommsdorff, 1995]. The intensity and the spatial distribution of serpentinization reactions in the crust accreted at slow spreading mid-ocean ridges are therefore important parameters. Seismic velocities and gravity data indicate that domains of ultramafic outcrops correspond to thin but nonabsent low seismic velocity-low density crust. In domains of ultramafic seafloor, this geophysically defined crust is typically 1–4 km thick [Hooft et al., 2000; Canales et al., 2000; Cannat et al., 1995a]. It probably corresponds to partially serpentinized mantle-derived peridotites (as initially proposed by Hess [1962]), with a variable proportion of gabbroic and doleritic intrusions [Cannat, 1993]. Increasing serpentinization results in a linear decrease of both seismic velocity and density, from nearly 8 kms21 and 3300 kg m23 at 0% of serpentinization to 5 kms21 and 2500 kg m23 at 100% of serpentinization [Christensen, 1972; Miller and Christensen, 1997]. Seismic velocities models can therefore be used to predict the degree of serpentinization at depth (Figure 1). However, this geophysical approach has significant limitations [Cannat et al., 2010]: (1) it does not uniquely determine the extent of serpentinization: gabbros, dolerites, and basalts may have seismic properties similar to partially serpentinized peridotites [Miller and Christensen, 1997]; (2) the spatial resolution of seismic velocity models is poor (200–500 m) so that homogeneous and moderate serpentinization could have the same seismic signature as heterogeneous and locally complete serpentinization; and (3) seismic velocities determined in the first kilometer of the crust are typically less than values for fully serpentinized peridotites indicating that another mechanism, probably fracturing, further reduces the velocities [Detrick et al., 1994; Spudich and Orcutt, 1980; Korenaga et al., 2002; Behn and Kelemen, 2003]. Another method to constrain the intensity and spatial distribution of serpentinization at slow spreading mid-ocean ridges is to study samples of serpentinized peridotites [Dilek et al., 1997; Oufi et al., 2002; Bach et al., 2004; Fr€ uh-Green et al., 2004; Andreani et al., 2007]. The limitation of this approach is that most dredged and submersible samples represent the uppermost footwall of the detachment faults and that the deepest hole drilled so far in these terranes is only 200 m deep (ODP hole 920D [Cannat et al., 1995b]). Serpentinized peridotite samples that are currently available from slow spreading ridges therefore probably come from no more than a few hundred meters below the exhumation fault and are not necessarily representative of the deeper levels of the fault’s footwall (Figure 1). ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2355 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 Figure 2. Location of the studied ultramafic samples. For the Mid-Atlantic Ridge: ODP Site 920 at 23 210 N [Cannat et al., 1995b]; IODP Site 1274 [Kelemen et al., 2004]; and dredges of the Serpentine cruise [Picazo et al., 2012] at 13 N and 15 N. For the Southwest Indian Ridge: dredges of the Smoothseafloor cruise [Sauter et al., 2013] between 62 E and 65 E. The depth distribution of our samples from ODP Site 920 is shown in the left-hand side of figure. Sample (SMS_DR17-04-04) from the Southwest Indian Ridge is shown as an example of a typical serpentinized harzburgite. The study presented here is based on a large set of serpentinized peridotites samples from slow and ultraslow ridges (Figure 2 and Table 1), including dredged samples from the Mid-Atlantic Ridge at 13 N and 15 N (setting described in Picazo et al. [2012]), from the Southwest Indian Ridge at 62 E–65 E [Sauter et al., 2013] and a set of drilled samples from ODP Site 920 at 23 N on the Mid-Atlantic Ridge (ODP Leg 153 [Cannat et al., 1995b]), and from ODP Site 1274 at 15 N also on the MAR (ODP Leg 209 [Kelemen et al., 2004]). Some of these drilled samples have been partially oriented using paleomagnetic measurements [Hurst et al., 1997]. Previous sample studies of serpentinization at slow ridges have focused on mineralogical and geochemical aspects to unravel successive phases of serpentinization and veining [Dilek et al., 1997; Andreani et al., 2007; Fr€ uh-Green et al., 2004; Bach et al., 2004], to constrain element mobility and water-rock exchanges during serpentinization [Paulick et al., 2006; Boschi et al., 2008] and to study the magnetic properties of variably serpentinized peridotites [Oufi et al., 2002; Bach et al., 2006]. In this paper, we address the question of fluid penetration into the fresh peridotite at the initial stages of serpentinization and the subsequent development of the serpentine mesh texture, in the context of axial exhumation and detachment faulting (Figure 1). The mesh texture characterizes most serpentinized peridotites independent of their geodynamic setting (abyssal, ophiolites, subduction wedges, orogens [e.g., Francis, 1956; Prichard, 1979; Wicks and Whittaker, 1977; Viti and Mellini, 1998]). In serpentinized samples from slow spreading ridges, this mesh texture predates complex veining and local serpentine replacement [Dilek et al., 1997; Andreani et al., 2007]. Mesh cells (Figure 3) comprise a mesh core made of relict olivine, or of isotropic serpentine and occasional brucite and a mesh rim made primarily of lizardite [Wicks and Whittaker, 1977; Viti and Mellini, 1998]. Mesh cells are typically 100–500 mm in size [Wicks et al., 1977; Prichard, 1979; Oufi et al., 2002; Andreani et al., 2007; Pl€ umper et al., 2012]. Early studies of ophiolites and orogenic serpentinites have shown that serpentinization progresses from microfractures, individualizing the mesh cells, toward their centers [Maltman, 1978]. These serpentinizing microfractures therefore act as a pathway for the hydrous serpentinizing fluids from the early stages of serpentinization. The origin of these microfractures is the focus of our study, specifically for the case of serpentinization at slow spreading mid-ocean ridges. Two mechanisms have been proposed to form these serpentinizing microfractures. One hypothesis invokes anisotropic thermal contraction of the peridotite and particularly of its primary mineral constituent olivine. This process is capable of generating residual stresses sufficient to produce microcracks in mantle rocks at slow spreading ridges [deMartin et al., 2004]. Boudier et al. [2010] have proposed this mechanism for the ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2356 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 formation of serpentinizing microfractures in a sample from the Oman ophiolite. The other hypothesis invokes reaction-induced hierarchical fracturing of the olivine due to volume increase during serpentinization [Iyer et al., 2008; Jamtveit et al., 2009; Rudge et al., 2010; Kelemen et al., 2011; Kelemen and Hirth, 2012; Pl€ umper et al., 2012]. Microstructural observations in support of these hypotheses were made mostly on ophiolitic samples (although the most recent work by Pl€ umper et al. [2012] includes one sample from ODP Site 920). Our study is the first to consider a comprehensive set of serpentinized peridotites samples from slow spreading ridges. We investigate the morphology and distribution of serpentinizing microfractures at a range of scales from optical microscopy and Scanning Electron Microscopy (SEM) observations (<1 mm), to sample (>10 cm) and to drill hole scale (>100 m) and we test the control of the olivine crystallography using Electron Back Scattering Diffraction (EBSD) maps and pole figures. Our results lead us to propose that thermal contraction and tectonic stresses combine to initiate the tight network of microfractures that allows the onset of serpentinization in fresh peridotites in the footwall of slow spreading ridges detachment faults. 2. Sample Description For this work, we examined thin sections of 278 samples of mesh-textured serpentinized peridotites from slow spreading ridges (Figure 2 and Table 1): 45 from ODP Site 920 (MAR 23 N [Cannat et al., 1995b]), 4 from ODP Site 1274 (MAR 15 N [Kelemen et al., 2004]), 23 dredged samples from the MAR axial valley walls at 13 N and 15 N [Picazo et al., 2012], and 201 samples dredged on and off axis the Southwest Indian Ridge between 62 E and 65 E [Sauter et al., 2013]. We find that mesh textures are similar at all sampling sites and belong to one of the two types that have been described in early studies of serpentinized peridotites: the equant and the ribbon-shaped mesh textures [Francis, 1956; Maltman, 1978; Wicks, 1984]. The equant mesh texture (Figure 3a) is composed of polygonal mesh cells with mesh rims of rather homogeneous widths [Wicks et al., 1977]. In the ribbon-shaped texture (Figure 3b), one orientation of mesh rims is more developed. This ribbon-shaped texture was interpreted by Francis [1956] and Maltman [1978] as due to deformation of an initially equant mesh texture. However, Wicks [1984] argued that there was no clear evidence for deformation of the ribbon-shaped serpentine mesh and proposed instead that these two distinct textures derived from two different types of initial microfracture networks in the olivine prior to serpentinization. The equant mesh texture is the most widespread. It is expressed alone in 140 of our samples. The other 138 samples contain either a combination of equant and ribbon-shaped mesh domains (132 samples) or only ribbon-shaped mesh (6 samples). Raman analyses (Olympus FV1000 microscope coupled with a Renishaw InVia spectrometer, Institut de Physique du Globe de Paris, France) on nine selected samples indicate that, as observed in previous studies of the meshwork in serpentinized peridotites [Rumori et al., 2004; Boudier et al., 2010], lizardite is the dominant serpentine species in the mesh rims (Figure 3e). We did not identify brucite in our samples but it may occasionally occur as intergrowths in the serpentine [Bach et al., 2004]. The iron released during the olivine serpentinization crystallizes as magnetite in the mesh rims. Most magnetite grains are smaller than a few micrometers and are concentrated along the trace of serpentinizing microfractures (Figures 3f and 3g). More pronounced concentrations of magnetite also commonly delineate a broader polygonal pattern that includes up to 10 individual mesh cells (Figures 3a and 3b). Optical microscopy using a wave plate under polarized light (Figures 3c and 3d) shows that the fast crystallographic axes of serpentine in the mesh rims tend to be perpendicular to the initial microfractures. This is consistent with lizardite plates being stacked in pseudocolumns [Wicks and Zussman, 1975; Cressey, 1979] along their [001] (fast) crystallographic axis. Rumori et al. [2004] and Boudier et al. [2010] documented a similar configuration in serpentinized peridotites from the Ligurian and Oman ophiolites. Systematic observation of our samples with the wave plate shows that lizardite [001] axes are indeed strongly oriented in all samples, subperpendicular to the serpentinizing microfractures (Figures 3c and 3d). In the next section, we will take advantage of this strong crystallographic preferred orientation to visualize and map the traces of serpentinizing microfractures at the scale of thin sections. ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2357 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems ROUMEJON AND CANNAT 0 0 Protolith C 2014. American Geophysical Union. All Rights Reserved. V Dunite Very coarse grained Equant/ribbon shaped Equant Equant Equant/ribbon shaped Coarse-grained equigranular Coarse-grained equigranular Porphyroclastic Equant Equant 67 66 31 76 90 87 75 70 81 74 30 100 95 60 75 Visual Estimate Equant 45–85 46 EBSD Maps 75 30 63 42 Subdomains Equant/ribbon shaped Equant Ribbon shaped/ equant Equant/ribbon shaped Equant Mesh Texture Porphyroclastic Porphyroclastic Coarse-grained equigranular Very coarse grained Coarse-grained equigranular Coarse-grained equigranular Porphyroclastic Porphyroclastic Primary Texture 55 6 15 58 6 24 60 6 19 13 6 10 17 6 12 115 6 55 130 6 60 45 6 20 ‘‘Yellow’’ 80 6 35 ‘‘Blue’’ 57 6 22 71 6 31 50 6 15 100 6 45 140 6 50 Initial 17 6 10 15 6 9 26 6 15 19 6 13 15 6 9 21 6 13 25 6 15 26 6 15 15 6 11 15 6 8 20 6 13 Mesh Cells Area/ Perimeter 6 std 41 6 19 42 6 11 40 6 15 75 6 35 100 6 50 45 6 15 ‘‘Red’’ 40 6 16 47 6 21 55 6 20 70 6 25 80 6 35 30 6 10 ‘‘Green’’ Hierarchical Apparent Thickness of Serpentine Along Microfractures (mm) 237 6 107 366 6 166 370 6 170 360 6 170 315 6 150 320 6 110 ‘‘Yellow’’ ‘‘Blue’’ 282 6 169 324 6 136 320 6 160 295 6 120 400 6 165 135 6 60 Initial 153 6 78 164 6 69 150 6 75 265 6 140 325 6 175 170 6 95 ‘‘Red’’ 183 6 100 160 6 85 135 6 70 265 6 150 255 6 190 170 6 85 ‘‘Green’’ Hierarchical Apparent Spacing of Microfractures (mm) 10.1002/2013GC005148 2358 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License a The three values given for the serpentinization extent correspond to the proportion of fresh olivine in the microscopic domains of Figure 8, in EBSD maps (Figure S1), or are estimates made visually for the whole thin section. The mesh cells area/perimeter data are obtained as shown in Figure 8. The dimensions listed for apparent thickness of serpentine along microfractures and apparent spacing of microfractures are explained in the text and correspond to the histograms in Figure 9. b Three orthogonal thin sections. 04–66 Harzburgite 04–62 b Harzburgite 04–41 ODP 920 (23 20.3 N 45 01.0 W) B 11R2 118–127 Harzburgite (piece 16) D 14R2 114–120 Harzburgite (piece 7C) D 18R2 113–118 Harzburgite (piece 13) ODP 1274 (14 43.30 N 44 53.30 W) A 05R1 119–122 Harzburgite (piece 10) A 08R1 114–117 Dunite (piece 16) A 17R1 119–121 Harzburgite (piece 20) SE_DR08 (14 44.8’N 44 58.1’W) 07–02 Harzburgite 07–32 Harzburgite SMS_DR17 (28 33.4’N 62 21.2’E) b Harzburgite 04-04 Sample Whole TS Serpentinization % Table 1. Characteristics of the 12 Serpentinized Peridotites Selected (Out of 278 Samples) for Detailed Analyses of the Serpentine Mesh Texturea Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 Figure 3. Typical equant and ribbon-shaped serpentinization mesh textures. Abbreviations: olivine (Ol), pyroxene (Px), lizardite (Liz), magnetite (Mgt). (a) Photomicrograph of equant mesh texture in natural light (sample SE_DR08-07-02). (b) Photomicrograph of ribbon-shaped mesh texture in natural light (sample 920D-18R2, 113–118 cm, piece 13). Concentrations of magnetite along some microfracture planes underline a polygonal pattern that includes up to 10 individual mesh cells. (c and d) Detail of Figures 3a and 3b in polarized light with the addition of a wave plate revealing the preferred orientation of fast [001] crystallographic axes of lizardite subperpendicular to serpentinizing microfractures. (e) Detail of Raman spectra for serpentine in the mesh rims of Figure 3a in pale gray and Figure 3b in dark gray. Peak positions are typical of lizardite. (f and g) Scanning Electron Microscope (backscattered electron) images located by insets in Figures 3a and 3b and showing details of the mesh structure: olivine relicts, lizardite mesh rims, and serpentinizing microfractures outlined by variable amounts of magnetite. Raman spots for Figure 3e are shown as circles. Our samples show a variety of preserpentinization mineralogies: dunites (17 samples), harzburgites, or lherzolites (261 samples), some (82 samples) with evidence for magmatic impregnations and veins that have been described in previous studies [Cannat et al., 1997; Kelemen et al., 2004; Picazo et al., 2012]. Preserpentinization textures, identifiable in samples that are not fully serpentinized, range from very coarse grained with olivine grains up to 0.5 cm in size (12 samples), to coarse-grained equigranular with mosaics of equant olivine grains up to 1 mm in size (34 samples), to porphyroclastic textures with variably elongated millimeter-sized olivine grains and local recrystallization in neoblasts <0.2 mm in size (77 samples). Ceuleneer and Cannat [1997] have described these textures and the corresponding mineral preferred orientations for the ODP Site 920 samples. Most samples from the Southwest Indian Ridge show locally tight and irregular subgrain boundaries in olivine, kink-bands, and fractures in pyroxenes, local recrystallization in smaller neoblasts and occasional brittle microshear zones, suggesting plastic to semibrittle deformation in higher stress-lower temperature lithospheric conditions [Cannat et al., 2012]. Samples of mylonitic peridotites are rare at the ridge locations considered here and therefore not representative of the bulk of the serpentinizing mantle material. We chose not include these rare mylonites in our study. In most samples, pyroxenes are less altered than the olivine, and even in samples with pervasively serpentinized olivine, relics of pyroxene are frequent. Pyroxenes are altered into bastites (serpentine pseudomorphs) with occasional amphibole (tremolite). The 159 samples also show significant overprinting of the serpentine mesh texture by microscopic to mesoscopic serpentine veins. Four main veining episodes have been described in the serpentinized peridotites of ODP Site 920 [Dilek et al., 1997; Andreani et al., ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2359 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 2007] and preliminary observations in the Southwest Indian Ridge samples suggest a comparable chronology (S. Roumejon et al., manuscript in preparation, 2013). Oxygen-isotope constraints suggest that the serpentine mesh texture formed at temperatures in excess of 350 C and 280 C, respectively, in the ODP Site 920 samples [Agrinier and Cannat, 1997] and in the Southwest Indian Ridge samples [Roumejon et al., 2013]. In order to perform detailed microstructural, Raman, SEM, EBSD, and 3-D analyses of the mesh texture, we selected a set of representative samples based on four criteria: (1) samples from each studied area (Figure 2); (2) well-developed mesh textures with limited veining and adequate representation of the ribbon and equant mesh textures; (3) variable pyroxene content; and (4) a wide range of serpentinization degrees. Among our 12 selected samples (Table 1), three come from the ODP Leg 153 holes 920B and 920D, three from the ODP Leg 209 hole 1274A, two from dredges in the MAR13 N–15 N area, and four from dredges at the Southwest Indian Ridge (62 E–65 E). Because mesh textures are three-dimensional structures [e.g., Wicks et al., 1977], we studied two of the selected Southwest Indian Ridge samples for which we had sufficient material in three orthogonal thin sections. The selected samples are moderately to fully serpentinized and include 10 serpentinized harzburgites and 2 serpentinized dunites. The equant mesh texture (Figure 3a) is observed in all samples with zones of ribbonshaped mesh in four of them. The ribbon-shaped mesh dominates only in one selected sample (920D-18R2, 113–118 cm, piece 13; Figure 3b). It is associated with advanced serpentinization (85% based on image analyzes of EBSD maps; Table 1) and is in sharp contact with a domain of nearly fresh harzburgite. Serpentinization in this sample is therefore strongly heterogeneous. In most samples, serpentinization is more even, yet we observe variations at the thin section scale, with less serpentinized domains a few millimeter-wide that are commonly associated with greater concentrations of pyroxenes. The two selected serpentinized dunites exhibit very coarse-grained olivine, up to 0.5 cm in size. The harzburgitic samples have coarse-grained equigranular (five samples; Table 1) or porphyroclastic initial textures (five samples). Additional features crosscut some of the selected samples: SMS_DR17-04-62 contains brittle microshear zones up to 400 mm thick, characterized by local cataclasis with angular olivine and pyroxene clasts; SMS_DR17-04-41 is cut by a tremolite vein; and three samples selected from ODP Site 920 contain altered magmatic veins [Cannat et al., 1997]. The selected samples are representative of our larger sample set in terms of serpentine texture and mineralogy: lizardite is dominant in the mesh rims, with occasional chrysotile (identified by Raman analysis using the peak characterization of Auzende et al. [2004] and Schwartz et al. [2013]; e.g., Figure 3e) in the mesh rims of samples 920B-11R2, 118–127 cm, piece 16 and 920D-18R2, 113–118 cm, piece 13. Magnetite is found associated with mesh rim serpentine in every sample. Two samples selected from ODP Site 920 also contain needles of iron-nickel sulfide [Dilek et al., 1997]. 3. Geometry and Distribution of Serpentinizing Microfractures at Thin Section Scale Optical microscopy images of equant serpentine mesh texture in natural or polarized light may give the impression that serpentinizing microfractures are randomly oriented. This impression is reinforced by the polygonal pattern outlined by more pronounced magnetite concentrations that commonly encloses up to dozens of mesh cells (Figure 3a). However, optical microscopy images obtained under polarized light with the addition of a wave plate systematically reveal that the lizardite in the mesh rims, that tends to orient with its [001] fast optical axis perpendicular to serpentinizing fractures (Figures 3c and 3d), has a pronounced crystallographic preferred orientation (CPO) at the scale of the whole thin section (Figure 4). In these images, polarization colors change from yellow to blue when the preferred orientation of the [001] fast axis of lizardite is, respectively, parallel and perpendicular to the fast axis of the wave plate (Figures 3c and 3d). Intermediate orientations lead to magenta colors. This method therefore underlines serpentinizing microfractures, including those with low magnetite concentrations that are less visible otherwise. Using this approach, we have been able to map the trace of serpentinizing microfractures in the 12 selected samples, and in 21 additional partially oriented drilled samples from ODP Site 920 (Table 1 and Table S1 in supporting information). In every case, we found four orientations of microfractures, forming two sets of ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2360 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 Figure 4. Mapping the traces of serpentinizing microfractures. Sample SMS_DR17-04-66 (serpentinized dunite, equant mesh texture). (a) Scan of the whole thin section, natural light. (b and c) Thin section maps with the serpentine mesh texture in pale gray, pyroxenes, bastites, and late serpentine veins in darker gray, and the traces of serpentinizing microfractures in yellow, blue, red, or green. For design clarity, only two orientations of microfractures are represented in each map. (d and e) Photomicrographs of the 2.5 mm by 3.5 mm area shown as inset in Figures 4a–4c). Polarized light with wave plate inserted. Lizardite with fast [001] axis parallel or perpendicular to wave plate axis appears, respectively, in yellow or in blue. These colors are used to trace microfractures that have lizardite rims > 10 mm in thickness and draw their average orientation in the thin section maps of Figures 4b and 4c. The ‘‘yellow’’ and ‘‘red’’ orientations in Figure 4b correspond to the blue and yellow colors in Figure 4d; the ‘‘blue’’ and ‘‘green’’ orientations in Figure 4c correspond to the blue and yellow colors in Figure 4e. By convention, we use blue or yellow colors to draw microfractures that can be traced over a few millimeters (see text). mutually perpendicular orientations. We now use four of the selected samples to describe this microfracture pattern. Our approach is illustrated in Figure 4 in the case of a serpentinized coarse-grained dunite (Southwest Indian Ridge sample SMS_DR17-04-66). We explore the whole thin section (Figure 4a) at regular steps (about 5 mm) using a graduated stage holder and turn the stage relative to the wave plate (in Figures 4d and 4e, we show the fast axis of the wave plate turning relative to the thin section instead) until one set of serpentinizing microfractures turns blue. We note that lizardite along perpendicular microfractures then turns yellow. We measure the apparent strike of these microfractures and draw them in the thin section map (Figure 4b). We then turn the stage again until another set of serpentinizing microfractures turns blue. In the thin section considered in Figure 4, we note that this occurs after a rotation of the stage by 40 , and that again lizardite associated with another orientation of perpendicular microfractures turns yellow (Figures 4d and 4e). We measure and draw the apparent strike of this other set of mutually perpendicular microfractures in the thin section map (Figure 4c). We then check for other orientations of the stage that would correspond to a clear lizardite preferred orientation, and move the stage to the next step. This method produces about 20 apparent strike measurements per thin section for each microfracture orientation, with standard deviations 65 to 615 . We also checked the local variability of these apparent strikes in 3.5 mm by 2.6 mm microphotographs using a regular grid of lines spaced by 200 mm and making measurements at the intersection with each grid line. This method yields more than 60 measurements per microphotograph for each microfracture orientation, and standard deviations are 67 to 617 . ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2361 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems A A B a SMS_DR17-04-66 10.1002/2013GC005148 B C b SMS_DR17-04-04 C Figure 5. Three-dimensional reconstruction of the ‘‘blue’’ and ‘‘yellow’’ serpentinizing microfracture planes mapped in three orthogonal thin sections for (a) sample SMS_DR17-04-66 and (b) sample SMS_DR17-04-04. As sketched in the insets, these microfractures exhibit an overall planar morphology, despite thin section-scale variations of apparent strikes with standard deviations up to 617 . The angle 6 standard deviation between the ‘‘blue’’ and ‘‘yellow’’ microfractures for sections A, B, and C are, respectively, 42 6 11 , 55 6 15 , and 54 6 11 in Figure 5a, and 43 6 14 , 37 6 8 , and 41 6 9 in Figure 5b. The arrow in Figure 5b means that thin section A corresponds to the lower face of the cube. In the serpentinized coarse-grained dunite presented in Figure 4, the traces of two families of microfractures that make an apparent angle of 40 (42 6 11 ; the yellow and blue traces in Figures 4b and 4c) are actually directly visible on the hand-held thin section (Figure 4a). The ‘‘wave plate’’ method allows to show Figure 6. Mapping the traces of serpentinizing microfractures in (a) sample SE_DR08-07-02 (serpentinized harzburgite, equant mesh texture) and (b) sample 920D-18R2, 113–118 cm, piece 13 (serpentinized harzburgite, ribbon-shaped mesh texture). As in Figure 4, for each sample, the first figure is a scan of the whole thin section, the two figures below that are maps of the traced microfractures, shown in pairs (yellow and red, and blue and green) for clarity, and the two larger figures to the right are detailed microphotographs in polarized light with a wave plate inserted. Additional explanations may be found in the caption for Figure 4. ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2362 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 that these two families are associated with strongly oriented lizardite rims and are perpendicular to two other orientations of serpentinizing microfractures that are only visible at the thin section scale (the red and green traces in Figures 4b and 4c). The step-by-step mapping approach also permits to assess the continuity of each family of microfractures at thin section scale. This allows us to show that two orientations of microfractures, ‘‘red’’ in Figure 4b and ‘‘green’’ in Figure 4c, have shorter traces, while the other two, ‘‘yellow’’ in Figure 4b and ‘‘blue’’ in Figure 4c, commonly link and extend up to a few centimeters. In figures following this one, we retain the same colors for serpentine in the same sets of microfractures, with the more continuous sets in blue and yellow, and the shorter sets in red and green. For this sample and for Southwest Indian Ridge serpentinized harzburgite sample SMS_DR17-04-04, we mapped three orthogonal thin sections (Figure 5) and show that the traces of the more continuous ‘‘blue’’ and ‘‘yellow’’ serpentinizing microfractures in each orthogonal section are consistent with a dominantly planar morphology (inset cubes in Figure 5). The orientation of the traces of serpentinizing microfractures in the serpentinized harzburgite thin section shown in Figure 5b is more variable than in the serpentinized dunite of Figure 4, particularly in the vicinity of partially serpentinized pyroxenes. Microfractures with traces parallel to the edge of a pyroxene show greater continuity and have thicker associated mesh rims. The two serpentinized harzburgites presented in Figures 3 and 6 display similar results. Sample SE_DR08-07-02 (Figure 6a) is fully serpentinized and has an equant mesh texture. Sample 920D-18R2, 113–118 cm, piece 13 (Figure 6b) presents a ribbon-shaped mesh texture in a fully serpentinized region, in sharp contact with much less serpentinized peridotite. In both cases, we find the two sets of mutually perpendicular serpentinizing microfractures. Each set comprises serpentinizing microfractures that tend to be more continuous at an apparent angle of 38 6 8 and 43 6 11 , respectively (in yellow and blue; Figure 6), and shorter perpendicular serpentinizing microfractures (in red and green). In the ribbon texture, one orientation of the more continuous microfractures is strongly predominant (represented in blue in Figure 6b). In each of our 40 mapped samples, we found a similar pattern with a maximum apparent angle of 55 between the traces of the ‘‘yellow’’ and ‘‘blue’’ serpentinizing microfractures and standard deviations lower than 617 . Although we could not make orthogonal thin sections in most samples because they were too small, this apparent angular relationship indicates that the ‘‘yellow’’ and ‘‘blue’’ microfractures make true angles 55 . In order to further characterize the geometry of serpentinizing microfractures, we also tried the U-stage method described by Boudier et al. [2010]. In this method, the section is tilted so as to align the two edges of the microcrack over the thickness on the thin section (30 mm). As a consequence, microfractures that have a rim of lizardite thicker than 10 mm cannot be measured if their dip in the thin section is less than 60 . This method is therefore fit to explore the fine-scale geometry of microfractures in weakly serpentinized peridotites [Boudier et al., 2010] but it introduces a bias for thin section-scale statistical studies of microfractures in significantly serpentinized samples. 4. SEM Images of Serpentinizing Microfractures The SEM image (Zeiss Supra 55P SEM, Universite Pierre et Marie Curie, France) in Figure 7a shows a detail of the ribbon-shaped mesh texture described in Figures 3b and 6b (920D-18R2, 113–118 cm, piece 13). Serpentinizing microfractures that control the ribbon texture (‘‘blue’’ microfractures in Figure 6b) are spaced by 50 mm or less and extend over more than the height of the image. Serpentinizing fractures with the orientation of the ‘‘yellow’’ family in Figure 6b are less abundant and appear shorter. Both families are associated with short perpendicular serpentinizing microfractures (the red and green families in Figure 6b) creating Tjunctions and limiting four-sided olivine domains. Some of these short fractures develop from etch pits in the olivine (Figure 7a) and are therefore posterior to the onset of serpentinization. Following Iyer et al. [2008], Kelemen and Hirth [2012], and Pl€ umper et al. [2012], we interpret these textural characteristics as indicators of a hierarchical microfracturing process associated with volume increase during serpentinization. This reaction-induced fracturing process produces the red and green orientations of microfractures and initiates along the more continuous ‘‘blue’’ and ‘‘yellow’’ microfractures that we will therefore describe as initial serpentinizing microfractures. The SEM image in Figure 7b shows a broader view of the same sample, in the poorly serpentinized domain visible in the left part of the thin section in Figure 6b. It shows that the initial serpentinizing microfractures ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2363 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 Figure 7. Scanning Electron Microscopy (backscattered electron) images of serpentinizing microfractures. Abbreviations: olivine (Ol), orthopyroxene (Opx), serpentine (Serp), microfracture (microfract). Insets show the overall orientation of the traces of ‘‘blue’’ and ‘‘yellow’’ serpentinizing microfractures at thin section scale. (a) Detail of the ribbon-shaped mesh texture in sample 920D-18R2, 113–118 cm, piece 13. Maps of serpentinizing microfractures for this thin section are shown in Figure 6b. The ‘‘blue’’ and less common ‘‘yellow’’ initial microfractures (see text) extend over more than the image width. The ‘‘red’’ and ‘‘green’’ hierarchical microfractures are shorter and at right angles to the ‘‘blue’’ and ‘‘yellow’’ microfractures. Some hierarchical microfractures seem to be related to etch pits in the olivine. (b) A broader view of the same sample in a less serpentinized domain, showing initial microfractures cutting olivine-pyroxene contacts with no apparent displacement. (c) Detail of the equant mesh texture in sample SE_DR08-07-02. Maps of serpentinizing microfractures for this thin section are shown in Figure 6a. Segments of the ‘‘blue’’ and ‘‘yellow’’ initial serpentinizing microfractures branch with no apparent displacement. (d) Detail of the equant mesh texture in sample SMS-DR17-04-62. This serpentinized harzburgite displays several brittle shear zones <100 mm in width, containing angular clasts of the primary mineralogy (olivine and pyroxene). One of these is outlined by arrows. The ‘‘blue’’ and ‘‘yellow’’ initial serpentinizing microfractures cut across these shear zones. (e) Detail of Figure 7d showing segments of the ‘‘blue’’ initial serpentinizing microfractures that cut across olivine and pyroxene clasts in microshear zone. (f) Detail of the equant mesh texture in sample SMS_DR17-04-04. Maps of serpentinizing microfractures for this thin section are shown in Figure 5b. The thickness of serpentine varies between <2 mm along microfractures in the olivine relict and >100 mm along magnetite-lined microfractures that bound the mesh cells. also cut into pyroxenes and that the olivine-pyroxene contacts are not displaced. We make similar observations in Figure 7c that shows a SEM detail of sample SE_DR08-07-02, a serpentinized harzburgite with an equant mesh texture (Figures 3a and 6a). In this sample, the two families of initial fractures are equally developed, forming anastomosing relays and delimiting elongated 50–200 mm wide lozenge-shaped ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2364 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 domains, with not visible displacement. Orthogonal hierarchical microfractures are observable in the more serpentinized right part of the SEM image. Lizardite rims developed along these initial and hierarchical microfractures outline the overall equant meshwork. Figures 7d and 7e show SEM images of sample SMS-DR17-04-62, a serpentinized harzburgite with welldeveloped brittle microshear zones. Most of these shear zones are less than 100 mm wide and contain angular clasts of the primary mineralogy (olivine and pyroxene). They are distinct from the anastomosing network of initial serpentinizing microfractures. In Figure 7e, we present a detail of a less serpentinized part of the sample in which initial serpentinizing microfractures of the ‘‘blue’’ family, with spacing of 50 mm, are continuous across the olivine and pyroxene clasts that form the shear zone. These crosscutting relations indicate that the formation of the brittle microshear zones predated the formation of the serpentinizing microfractures. Our SEM study also reveals that the amount of serpentine associated with the different orientations of serpentinizing microfractures may be variable: in samples with a ribbon-shaped mesh, those initial serpentinizing microfractures that are parallel to the ribbons tend to have thicker serpentinized rims (Figure 7a). We also note that in all of our samples some serpentinizing microfractures have a very thin serpentine coating. This is true in moderately serpentinized domains (Figures 7b and 7c) but this is also observed in extensively serpentinized samples where some microfractures are coated by <2 mm of serpentine (the thinnest fractures in the olivine relict of Figure 7f), while other are lined by serpentine rims 100 mm thick or more (the broad lizardite rims in Figure 7f). In the next section, we further examine the scales of initial and hierarchical microfracturing and their relation to the characteristic length scales of the mesh cells. 5. Distribution and Spacing of Serpentinizing Microfractures In order to investigate the characteristic scales of the mesh cells, and to link these scales with the advancement of the serpentinization reactions, we used microphotographs (magnification 1003) from 39 selected 1.4 mm by 1 mm domains in nine samples (Table 1 and Figure 8). We included microphotographs from orthogonal thin sections when available. In each domain, we traced manually all the visible serpentinizing microfractures and we used image processing to delineate the contours of olivine relicts (Figures 8a and 8b). SEM images made at yet greater magnification reveal a few additional cracks with very thin (<2 mm) serpentine rims that are not visible with the optical microscope but we find that the 1.4 mm by 1 mm microphotographs adequately capture the geometry and scales of the serpentinizing microfracture network in our samples. We derived a local serpentinization extent for each domain from the ratio of olivine surface to the total surface enclosed by the mapped microfractures. These serpentinization extents range between 18% and 100% for the 39 selected domains and may vary significantly within each sample. We could not find domains with serpentinization extents <18%. Figures 8a and 8b show examples of the microfracturing patterns obtained for two serpentinized harzburgites with an equant mesh texture (Table 1). The size of the mapped domains is close to that of a typical olivine grain in these samples and the microfracture network at this scale may be at a small angle to the average orientations measured for the thin section (inset in Figures 8a and 8b shows average orientation of the ‘‘blue’’ and ‘‘yellow’’ initial serpentinizing microfractures in each sample). In Figure 8c, we plot the mean area over perimeter ratio for the cells defined by the microfracture network in each domain, as a function of the local serpentinization extent. This ratio is equivalent to the half radius of a circle of same area than the cell. Between 18% and 70% of serpentinization, the mean value of this ratio is around 15 mm in all of our samples (suggesting mean diameters around 60 mm for the unit cells of this microfracture network), with large but consistent standard deviations. This stability of unit cell dimensions for serpentinization rates between 18% and 70% indicates that little additional microfracturing of the protolith, reaction-induced or otherwise, occurred after 20% of serpentinization. For serpentinization extents >70%, the mean ratio of cell area over cell perimeter increases to >25 mm, yielding mean diameters >100 mm for the microfracture unit cells at >90% serpentinization (Figure 8c). This increase is due to the overgrowth of serpentine rims from nearby cells over the smaller microcracks and to the full alteration of the smaller olivine relicts. ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2365 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 Figure 8. Characteristic dimensions of the serpentinizing microfractures network in selected samples. (a and b) Examples of the microphotographs and line drawings used to obtain the statistical data plotted in (c) and (d). Photomicrographs are taken under natural or polarized light (magnification 1003) and cover 1.4 mm by 1 mm domains. Several domains are analyzed in each sample. Microfractures are drawn manually. Local serpentinization extents (S) are derived from the surface proportion of olivine over the total surface enclosed by the microfractures. Insets show the overall orientation of the traces of ‘‘blue’’ and ‘‘yellow’’ serpentinizing microfractures at thin section scale. (c) Mean area over perimeter ratio for the cells defined by the microfracture network in each 1.4 mm by 1 mm domain, as a function of the local serpentinization extent. Vertical error bars: one standard deviation. (d) Complementary cumulative distribution function for the cell areas in eight samples for which we measured >150 cells. Symbols highlight distribution functions for three representative samples: 920D 14R2, 114–120 cm, piece 7C (534 measured cells); SMS_DR17-04-66 (1151 measured cells); and SE_DR08-07-32 (186 measured cells). Mesh cell areas in these samples follow lognormal distributions with fitting parameters (mean 6 one standard deviation) of, respectively, 8.07 6 0.99, 8.32 6 1.1, and 8.81 6 1.15 mm2. The complementary cumulative distribution function for the mesh cell areas (measured on microphotographs as shown in Figures 8a and 8b) is shown in Figure 8d for eight samples. For samples SMS_DR17-0466 and SMS DR17-04-04, the cumulative distributions are similar for cell areas measured in the three orthogonal thin sections. Figure 8d highlights three samples (920D 14R2, 114–120 cm, piece 7C; SMS_DR17-04-66; and SE_DR08-07-32) with serpentinization extents of 63%, 67%, and 87%, respectively (Table 1). The y axis in Figure 8d gives the percent of mesh cells having an area larger than the value indicated on the x axis (as underlined by the 50%, 10%, and 1% lines), showing that different samples contain different proportions of medium to large mesh cells. For example, only 10% of the mesh cells have areas larger than 104 mm2 (equivalent to a 60 mm diameter for an ideal circular cell) in sample 920D 14R2, 114–120 cm, piece 7C, whereas this proportion is of 20% and 40%, respectively, in samples SMS_DR17-04-66 and SE_DR08-07-32. The proportion of larger mesh cells is positively correlated with the average serpentinization extent as also seen in Figure 8c and consistent with the overgrowth of serpentine rims at the expense of the smallest microcracks in the more serpentinized samples. The measured distributions in each sample show a close fit to lognormal cumulated distribution curves (Figure 8d). The mean and standard deviation of the natural logarithm of the measured cell areas are similar in the samples represented in Figure 8d: 8.07 6 0.99, 8.32 6 1.1, and 8.81 6 1.15 mm2, respectively, in samples 920D 14R2, 114–120 cm, piece 7C, SMS_DR17-04-66, and SE_DR08-07-32. Converting these values back into units of distance yield the following areas: 3197, 4105, and 6701 mm2. These lognormal distributions are similar to the distribution found by Pl€ umper et al. [2012], who used a similar method for measuring mesh cell areas in a serpentinized harzburgite from ODP Site 920 Hole D. ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2366 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems “yellow” initial microfractures 15 nb. of data 65 measures average = 50±15 µm av. spacing = 320±160 µm 15 25 20 91 measures average = 40±15 µm av. spacing = 150±75 µm 40 147 measures average = 55±20 µm av. spacing = 135±70 µm 30 10 20 10 5 5 0 20 28 measures average = 45±20 µm av. spacing = 320±110 µm 6 15 4 10 2 5 0 “green” hierarchical microfractures “red” hierarchical microfractures 15 0 8 b 920D 18R2 #13 20 38 measures average = 55±15 µm av. spacing = 370±170 µm 10 nb. of data SMS_DR17-04-66 a “blue” initial microfractures 10.1002/2013GC005148 25 65 105 145 185 225 0 10 5 118 measures average = 80±35 µm av. spacing = 135±60 µm 0 15 38 measures average = 45±15 µm av. spacing = 170±95 µm 10 0 10 8 19 measures average = 30±10 µm av. spacing = 170±85 µm 6 4 5 2 25 65 105 145 185 225 0 25 65 105 145 185 225 0 25 65 105 145 185 225 apparent thickness of lizardite rims along microfractures (µm) Figure 9. Histograms of the apparent width of serpentine measured across initial and hierarchical serpentinizing microfractures in samples (a) SMS_DR17-04-66 (equant mesh texture; microphotographs and microfractures maps in Figure 4) and (b) 920D-18R2, 113–118 cm, piece 13 (ribbon-shaped mesh texture; microphotographs and microfractures maps in Figure 6b). Measurements were performed along regularly spaced grid lines on several 3.5 mm by 2.6 mm microphotographs taken under polarized light with the wave plate inserted. Microfractures associated with serpentine width <10 mm are not taken into account. The plots also give the total number of measurements per sample, the average apparent serpentine thickness, and the average apparent spacing of the corresponding microfractures. 6. Thickness and Spacing of Serpentine Mesh Rims The wave plate method is effective to produce thin section-scale maps of those serpentinizing microfractures that correspond to apparent lizardite thickness >10 mm (Figures 4–6). We measured this apparent thickness of lizardite and the apparent spacing of these microfractures in the four samples illustrated in Figures 5 and 6 (Table 1). For two of these samples, measurements were carried out in three orthogonal sections, yielding comparable results. Measurements were made in 3.5 mm by 2.6 mm microphotographs taken under polarized light with the wave plate inserted (partly shown in Figures 4d and 46). We measured the apparent serpentine thicknesses at the intersections with a regular grid of lines spaced by 400 mm. This method allows us to measure lizardite rims with apparent thickness >10 mm. The measured values (Figure 9) correspond to the full apparent thickness of the serpentine domains associated with serpentinizing microfractures, approximately equivalent to twice the apparent thickness of the mesh rims that surround intervening mesh cores (Figure 3). If we assume that the rims replace preexisting olivine, these apparent rim thickness values provide an approximation for the extent of the serpentinization reactions associated with each serpentinizing microfracture. Histograms of the apparent thickness of lizardite along fractures in the serpentinized dunite sample with an equant mesh texture (SMS_DR17-04-66; Figure 9a) display similar 20–100 mm ranges across the four sets of serpentinizing microfractures with a mean value around 50 mm. By contrast, in the serpentinized harzburgite with a ribbon-shaped mesh texture (920D-18R2, 113–118 cm, piece 13; Figure 9b) those initial microfractures that parallel the ribbons (the ‘‘blue’’ microfractures; Figure 6b) develop wider serpentine rims, ranging between 20 mm and more than 200 mm in apparent thickness with a mean value of 80 mm. Serpentine thicknesses along the other three orientations of serpentinizing microfractures in this sample are thinner (20–90 mm with a mean value of 40 mm). In samples SMS_DR17-04-04 and SE_DR08-07-02 (Figures 5b and 6a) that have a dominantly equant mesh texture, the widest mesh rims are found in the vicinity of pyroxenes and are associated with those microfractures that are subparallel to the former olivine-pyroxene contacts. Depending on the orientation of these contacts, this leads to larger mesh rim thicknesses associated with one or the other of the four families of serpentinizing microfractures and produces ribbon-shaped textured domains. The ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2367 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 average apparent thickness of serpentine along microfractures in our four selected samples (Table 1) is 66 6 38 mm. This corresponds to an average thickness of the mesh rims of 33 6 19 mm. The apparent spacing of the serpentinizing microfractures that correspond to these well-developed serpentine rims is also variable. In serpentinized dunite sample SMS_DR17-04-66 (equant mesh texture; Figure 8a), the average apparent spacing is similar for the ‘‘blue’’ and ‘‘yellow’’ families of initial microfractures (370 6 170 mm and 320 6 160 mm, respectively). It is more than twice the spacing between the shorter hierarchical (red and green) microfractures (150 6 70 and 135 6 75 mm; Figure 9a). Measurements of the apparent spacing of the serpentinizing microfractures that correspond to well-developed mesh rims in two other equant mesh-textured samples (SMS_DR17-04-04 and SE_DR08-07-02; equant serpentine mesh textures) give similar results for initial microfractures (Table 1), and higher values (250–350 mm) for the spacing of hierarchical microfractures. In the ribbon-shaped textured serpentinized harzburgite sample 920D-18R2, 113–118 cm, piece 13 (Figure 9b), the average spacing of microfractures that correspond to well-developed serpentine rims is 135 6 60 mm for the orientation paralleling the ribbons (‘‘blue’’ orientation) and 320 6 110 mm for the ‘‘yellow’’ orientation. For hierarchical (red and green) microfractures, the spacing is similar (170 6 95 and 170 6 85 mm), and similar to that measured in the dunitic equant mesh-textured sample (Figure 9a). Ribbons are thus found to form along tightly spaced initial microfracture planes, consistently with the hypothesis of a link between ribbon-shaped mesh texture and a different microfracture pattern in the fresh olivine prior to serpentinization [Wicks, 1984]. It should also be noted that microscopic domains of ribbonshaped mesh develop next to pyroxenes or bastites in samples that have a dominantly equant mesh texture (Figures 5b and 6a). In this case, it is possible that the pyroxenes create local perturbations either before or during serpentinization. 7. Serpentinizing Microfractures and Olivine Preferred Orientation In this section, we explore the relationship between olivine crystallography and the geometry of initial serpentinizing microfractures. Olivine has anisotropic thermal expansion coefficients, with the largest value along the [010] crystallographic axis and the lowest along the [100] axis [Suzuki, 1975; Bouhifd et al., 1996]. The [010] axis also corresponds to the direction of minimal energy required to fracture the olivine grains [Swain and Atkinson, 1978]. Thus, microfractures that form in response to stresses generated by thermal contraction of the exhumed mantle material are expected to preferentially form parallel to the easy (010) olivine cleavage planes, as observed in a sample from the Oman ophiolite by Dewandel et al. [2003] and Boudier et al. [2010]. Mapping serpentinizing microfractures at thin section scale in our abyssal samples (Figures 4–6) has already revealed a more complicated pattern, with two intersecting sets of initial microfractures. We used Electron Back Scattering Diffraction (EBSD) in six selected samples in order to check whether these families of initial serpentinizing microfractures formed at a systematic angle to the olivine fabric. The analyses were carried out at the SEM-EBSD facility at Geosciences Montpellier (Universite Montpellier 2, France), using a JEOL JSM 5600 SEM. The EBSD maps (Figure S1) cover between 2.63 and 5.78 cm2. They are based on up to 500,000 data points attributed to olivine, pyroxene or serpentine and spaced by 35 or 50 mm. The proportion of points that correspond to primary minerals and to serpentine is used to derive a serpentinization extent at the EBSD map scale. This extent varies between 45% and 85% (Table 1). The EBSD data reveal a moderate to strong olivine crystallographic preferred orientation (CPO) in each sample (Figure 10). The interpretation of these CPOs in terms of plastic slip systems in olivine is detailed in the supporting information. In order to compare the pole figures of olivine CPOs with the orientation of initial serpentinizing microfractures in each sample, we consider the average strike of the traces of these microfractures in each thin section (as determined using the wave plate method; Figures 4–6). In most samples, we can only compare these strikes with the olivine CPO because we do not have measurements of the microfracture dips. We only have this information for the two samples (SMS-DR17-04-04 and SMS-DR17-04-66; Figures 10a and 10c) in which we have been able to cut and study three orthogonal sections. In sample SMS-DR17-04-04 (Figure 10a), we find that while the ‘‘blue’’ family of serpentinizing microfractures is close to parallel to the preferred orientation of the olivine (010) crystallographic planes (i.e., close to perpendicular to the preferred orientation of the [010] axes), there is a significant angle between olivine (010) ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2368 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems a SMS_DR17-04-04 [010]ol d SE_DR08-07-32 [100]ol 10 Area = 3.32 cm N = 133536 nol = 12886 8 [001]ol b SMS_DR17-04-62 [010]ol 6 8 6 [001]ol 4 2 2 e 920D 14R2 #7C 10 8 [001]ol Area = 2.63 cm N = 105984 nol = 14156 [100]ol 4 [100]ol Area = 5.78 cm N = 473800 nol = 102168 [010]ol 10.1002/2013GC005148 6 [010]ol [100]ol 8 Area = 2.63 cm N = 105984 nol = 42188 6 [001]ol 2 2 c SMS_DR17-04-66 Area = 4.12 cm N = 165600 nol = 34100 [010]ol [100]ol f 920D 18R2 #13 15 [001]ol 4 4 10 Area = 4.24 cm N = 510600 nol = 110176 5 [010]ol [100]ol 5 4 [001]ol 3 2 1 0 Figure 10. Electron Backscattered Diffraction (EBSD) results showing polar Crystallographic Preferred Orientation (CPO) diagrams for olivine in six selected samples (lower hemisphere projection). Mean apparent strikes and dips (standard deviations < 617 ) of the ‘‘blue’’ and ‘‘yellow’’ initial serpentinizing microfractures at thin section scale are shown as either yellow and blue planes in (a) and (c) for samples in which we analyzed three orthogonal thin sections (Figure 5), or as traces in the plane of the studied thin section: yellow and blue rectangles in (b), (d), (e), and (f). The trace of elongated spinel grains is shown by open rectangles. For each sample, we specify the analyzed area, the total number of EBSD data points (N), and the number of data points attributed to olivine (nol). Corresponding EBSD maps and orthopyroxene CPOs are shown in Figures S1 and S2, respectively. and the microfractures of the ‘‘yellow’’ family. In sample SMS-DR17-04-66 (Figure 10c), ‘‘blue’’ initial serpentinizing microfractures tend to be nearly orthogonal to the (010) planes in olivine, and there is a significant angle between olivine (010) and the ‘‘yellow’’ family. In sample SE-DR08-07-32 (Figure 10d), the preferred strike of the olivine (010) planes is parallel to the average strike of the ‘‘blue’’ microfractures. In the two analyzed ODP site 920 samples, the average strikes of the ‘‘blue’’ and ‘‘yellow’’ serpentinizing microfractures tend to make small angles with the girdles outlined by olivine [010] axes (Figures 10e and 10f). This indicates that these two families of initial microfractures preferentially make large angles with the olivine (010) planes. Sample SMS-DR17-04-62 shows a similar pattern (Figure 10b). Therefore, in the six samples considered for this study, we did not find a consistent geometrical relationship between the olivine CPO and the initial serpentinizing microfractures. The lack of a systematic relationship between olivine CPO and the geometry of initial serpentinizing microfractures is also observed at the scale of individual olivine grains. In section 3, we described the use of a U-stage to measure thin serpentinizing microfractures within olivine relicts. We did not find a systematic relationship between the strike and dip of these microfractures and the crystallography of the host olivine. 8. Partial Reorientation of the Initial Serpentinizing Microfractures at ODP Site 920, Using Paleomagnetic Constraints Long pieces of core drilled at ODP Site 920 are parallel to the drilling direction and therefore close to vertical. However, these pieces can rotate about the vertical axis during recovery. A reorientation of these long pieces of core at ODP site 920 was proposed based on paleomagnetic data [Cannat et al., 1995b; Hurst et al., 1997]. ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2369 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 It has significant uncertainties in terms of absolute reorientation in the geographic frame [Hurst et al., 1997], but provides robust constraints on the reorientation of long pieces of core relative to one another. Harzburgites drilled at ODP site 920 are crosscut by successive generations of nearly planar serpentine veins [Dilek et al., 1997; Andreani et al., 2007]. Among them, the so-called ‘‘V2 veins’’ are the most widespread. They are white on the core cut surfaces (Figure 11a), commonly sigmoidal in shape, and filled with chrysotile fibers [Dilek et al., 1997; Andreani et al., 2007] that grow perpendicular to the vein contacts (Figure 11c). The reorientation of these V2 veins based on core demagnetization data [Hurst et al., 1997] shows that they have a statistically consistent geometry, with a 30 average dip throughout the 200 m deep Hole 920D. Hole 920B, drilled only 20 m away, shows a similar pattern but is only 100 m deep. Hurst et al. [1997] further proposed that if the magnetic declination was 0 when the magnetization was acquired, V2 veins in the cored section can be reoriented to a dominant north to northwest strike and an eastward 30 dip (Figure 11d). In this section, we propose a reorientation, at the drill hole scale, of the initial serpentinizing microfractures at ODP site 920. For this we examined the geometrical relations between the V2 veins and the microscopic serpentinizing microfractures. We did this in 19 samples that contain V2 veins and correspond to core pieces sufficiently long to have preserved the orientation of the vertical (Table S1). Onboard ODP Leg 153, cores were preferentially cut orthogonal to macroscopic V2 vein planes so that the dip of these veins in the cut plane of the long pieces of core is the true dip. Our 19 selected samples are serpentinized harzburgites with coarse-grained equigranular or porphyroclastic primary textures and equant or ribbon-shaped serpentine mesh textures. Serpentinization extents are >50%. Figure 11a to c shows one of these samples, with a ribbon-shaped serpentine mesh texture. In thin section, it is clear that the V2 veins are parallel to the dominant (‘‘blue’’) set of serpentinizing microfractures (Figure 11c). We found this to be the case in all samples, with the V2 veins parallel to one of the two orientations of initial serpentinizing microfractures (‘‘blue’’ or ‘‘yellow’’). Since V2 veins have consistent dips and orientations at the 200 m scale of ODP Hole 920D [Hurst et al., 1997] (Figure 11d), this must be also the case for at least one set of initial serpentinizing microfractures. 9. Discussion In this study, we examined the serpentine meshwork in 278 samples of serpentinized peridotites from several locations along slow spreading ridges (Figure 2). We then carried out a detailed microstructural, SEM, EBSD, and Raman study of the serpentine mesh in a selection of representative samples (Table 1). We find that the characteristic dimensions of the serpentine mesh and the geometry of the network of microfractures that controls the formation of this mesh are similar at all studied locations. This suggests that mechanisms of microfracturing and serpentine mesh texture development are also similar. The observed microfracture network is consistently made of two sets of initial serpentinizing microfractures that make an angle of <55 and commonly extend across several adjacent grains of olivine and pyroxene. Each set of initial microfractures is associated with shorter orthogonal microfractures that we interpret, following [Iyer et al., 2008; Jamtveit et al., 2009; Kelemen and Hirth, 2012; Pl€ umper et al., 2012], as hierarchical cracks due to serpentinization-induced volume increase. Another characteristic of all samples is that, although serpentinizing microfractures form tight networks (typical spacing 60 mm for microfractures that have a rim of serpentine >2 mm; section 5 and Figure 8), those microfractures that have serpentine thicknesses >10 mm and thus may have experienced larger fluid/rock ratios, due to higher time-integrated fluid flux, are more widely spaced (100–400 mm; section 5 and Figure 9). In this discussion, we address two questions: (1) what processes cause the initial microfracturing of the peridotite? (2) What are the characteristic scales of the fluid pathways from the onset of serpentinization to the fully developed mesh texture? For both questions, we consider how the processes involved may fit in with the tectonic and hydrothermal evolution of mid-ocean ridge detachment faults. 9.1. Initial Microfracturing of the Peridotite 9.1.1. Observational Constraints and Possible Processes We consider the four following observational constraints: (1) The initial and the hierarchical serpentinizing microfractures are pervasive and closely spaced in all samples: 60 mm based on the statistical analysis illustrated in Figure 8c, consistent with SEM observations (Figure 7). This spacing is independent of, and much ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2370 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 Figure 11. Relative reorientation of initial serpentinizing microfractures at ODP site 920. (a) Example of a long piece of core (sample 920D 13R3, 129–135 cm, piece 19; see Table S1) cut perpendicular to macroscopic V2 serpentine veins. (b) Thin section located in Figure 11a. Scan, photographed in polarized light, showing a well-developed ribbon-shaped mesh texture. (c) Microphotograph located in Figure 11b, photographed in polarized light with the wave plate inserted. Inset shows the overall orientation of the traces of ‘‘blue’’ and ‘‘yellow’’ serpentinizing microfractures at thin section scale. Serpentine ribbons (fast axis of lizardite subperpendicular to axis of wave plate) and the corresponding ‘‘blue’’ microfractures are parallel to V2 veins (fast axis of chrysotile subparallel to axis of wave plate). (d) Reoriented azimuths and dips of V2 veins in ODP hole 920D based on paleomagnetic data [Hurst et al., 1997]. The proposed absolute reorientation has significant uncertainties [Hurst et al., 1997]. However, the relative reorientation is robust and indicates that V2 veins have a consistent orientation and dip over the 200 m of the drilled section. We have examined seven of these reoriented V2-bearing samples (shown as diamonds in Figure 11d) and confirmed that V2 veins are parallel to one orientation of initial serpentinizing microfractures. less than, the size of the original olivine grains. (2) Initial serpentinizing microfractures develop in two orientations that intersect at angles < 55 . SEM images (Figure 7) show that there is neither displacement across these fractures nor any indication for a chronology in the development of the two orientations. We therefore propose that they formed at the same time (conjugate microfractures). (3) Initial serpentinizing microfractures have consistent geometries over thin section and sample scales. In the one location where we had access to suitable sampling (ODP drill site 920), these microfractures appear to have consistent trends and dips over a 200 m thick downhole section. (4) We did not find a systematic relationship between the trend and dip of initial serpentinizing microfractures and the olivine CPO. Our findings therefore differ from the results of Boudier et al. [2010] who found a preferred orientation of serpentinizing microfractures perpendicular to the olivine [010] axis (highest value of thermal expansion coefficients [Suzuki, 1975; Bouhifd et al., 1996]) in a sample from the Oman ophiolite. We now examine how these constraints fit with the two proposed methods for formation of serpentinizing fractures in peridotites: anisotropic thermal contraction of olivine [deMartin et al., 2004; Boudier et al., 2010] and reaction-induced hierarchical fracturing due to volume increase during serpentinization [Iyer et al., 2008; Jamtveit et al., 2009; Rudge et al., 2010; Kelemen and Hirth, 2012; Pl€ umper et al., 2012]. Because the abyssal samples we are considering come from the footwall of large offset faults (Figure 1), we will also examine microfracturing due to tectonic deviatoric stresses as a third possible mechanism. ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2371 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 9.1.2. Reaction-Induced Microfracturing Initial serpentinizing microfractures are not mutually perpendicular and do not form T-junctions such as expected for local reaction-induced hierarchical microfracturing [Iyer et al., 2008; Jamtveit et al., 2009; Pl€ umper et al., 2012]. They could be extensional cracks caused by serpentinization of neighboring peridotite domains as described by O’Hanley [1992] and many other references cited in the previous paragraph. However, they are very closely spaced which would suggest that these neighboring domains would be both small and nearby. Such closely spaced cracks are, for example, observed in plagioclase-rich microdomains next to serpentinized olivine grains in troctolitic rocks [e.g., Jamtveit and Hammer, 2012], but they are radial at thin section scale and therefore not consistent with observational constraint 3 (constant orientations of initial serpentinizing microfractures at sample and, at least at ODP Site 920, 100 m scale). Initial serpentinizing microfractures are also conjugate which do not fit well with them being purely extensional cracks. 9.1.3. Thermal Cracking The observation that serpentinizing microfractures do not have a systematic preferred orientation relative to olivine CPOs (observational constraint 4) is different from observations in Oman that Boudier et al. [2010] suggested were consistent with their thermal cracking hypothesis. The close spacing and pervasive distribution of initial serpentinizing microfractures are, however, potentially consistent with a role for stresses induced by thermal contraction. Fredrich and Wong [1986] published a thorough review of thermal cracking experiments on a variety of rock types and developed a 2-D micromechanical interpretation. Another extensive review including more recent experimental work may be found in deMartin et al. [2004] who specifically addressed thermal cracking of peridotites at slow spreading mid-ocean ridges. Micromechanical models that adequately predict experimental results propose that stresses due to thermal contraction mismatch (in polymineral aggregates or in monomineral aggregates with thermal contraction anisotropy) build up at grain boundary and intragranular flaws as temperature decreases and viscous stress dissipation slows. Cracking occurs when a critical stress intensity factor is reached at these local stress singularities. Crack density is expected to increase with the differential of temperature applied and crack length increases with grain size. Although crack initiation is modeled as occurring preferentially at grain boundaries, intragranular and transgranular cracks (as observed in our samples, e.g., Figure 7) are frequent in the experimental material. 9.1.4. Tectonic Cracking A role for tectonic deviatoric stresses is consistent with the observation that initial serpentinizing microfractures form two sets intersecting at less than 55 . This would fit an interpretation of these microfractures as conjugate Riedel structures formed in a dominantly non coaxial deviatoric stress field [e.g., Katz et al., 2004]. The absence of displacement along these microfractures would then argue for a formation at stress levels lower than the yield stress [e.g., Tapponnier and Brace, 1976]. Peridotites cooling into the brittle regime in the region closest to the detachment (Figure 1) may undergo such pervasive cracking while deviatoric stresses build up to ultimately equal the rock strength and cause localized failure. We find, however, that this far field tectonic stresses hypothesis is difficult to reconcile with initial serpentinizing microfractures being so closely and pervasively spaced in all the samples we studied. This characteristic seems more consistent with a crystal-scale mechanism for stress build up, which points back to anisotropic thermal contraction as a suitable mechanism. In order to reconcile these seemingly contradictory constraints, we propose as a working hypothesis that initial serpentinizing microfractures in our slow spreading ridge samples formed due to local stress concentrations caused by a combination of anisotropic thermal contraction and deviatoric tectonic stresses lower than the rock strength. 9.1.5. Formation of Initial Serpentinizing Microfractures Next to Mid-Ocean Ridge Detachments The addition of tectonic and thermal stresses proposed in the previous paragraph is actually a predictable consequence of tectonic exhumation. As exhumation proceeds, the axial detachment fault acts as conveyor belt bringing newly fractured peridotites from the mantle to the domain of active hydrothermal circulation (Figure 1). Strain localizes in the fault, but the footwall is subjected to significant shear stresses and to bending stresses [e.g., Lavier et al., 1999], while the temperature of a given volume of peridotite decreases, leading to thermal contraction. In Figure 12, we follow deMartin et al. [2004] and infer that the base of the elastic domain in peridotites on axis corresponds with the 800 C isotherm. We further infer that the temperature interval for thermal stress build up is between 800 C and 400 C, the maximum temperature for initial serpentinization [e.g., Agrinier and Cannat, 1997; Fr€ uh-Green et al., 2004]. Contraction due to cooling over ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2372 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 onset of serpentinization, hierarchical microfracturing, mesh texture development oth ui ds fault r hyd Domain of active serpentinization ~500 µm er mal fl crust initial microfracturing Brittle domain 400°C mantle Ductile domain 800°C exhumation onset of microfracturing Figure 12. Conceptual sketch of microfracturing and development of a serpentine mesh texture in the footwall of a mid-oceanic detachment fault. The sketch corresponds to the domain within a few hundred meters of the fault, as indicated in the inset in Figure 1. Fresh peridotite is progressively exhumed and passes from the ductile domain into the brittle domain. There, the combination of anisotropic thermal contraction and tectonic stresses leads to pervasive microfracturing. At temperatures <400 C, fluid infiltration along these initial microfractures causes serpentinization and the formation of additional, hierarchical microfractures. Provided that fluid supply is sufficient, the serpentine mesh texture may then be fully developed (i.e., to serpentinization extents of 70–80%) in times that could be as short as a few years (see text). Alternatively, local heterogeneities in fluid fluxes may cause domains of fresher peridotites to be preserved. Stresses due to serpentinization-induced volume increase and/or tectonics may cause cracks to develop and allow enhanced fluid fluxes and hydrothermal cooling in previously less permeable rocks. As a result, we expect that the mesh-texture formation stage of serpentinization at slow spreading ridges occurs in a nonsteady state and spatially heterogeneous fashion during tectonic exhumation of the peridotites. the 800 C–400 C temperature interval should generate 1% porosity [Bouhifd et al., 1996]. If our hypothesis is correct, this will be complemented by a component of tectonic dilatancy. Initial serpentinizing microfractures that have thin serpentine rims (and have therefore retained their original geometry; e.g., Figure 7) typically extend a few hundred mm and their dominant sets commonly crosscut each other. This would generate a connected porosity open to fluid transport once the cracked peridotites get into the hydrothermal domain. This permeable network of conjugate microfractures has a preferred orientation, at least at scales of a few decimeters, and possibly of hundreds of meters (c.f. ODP Site 920). If our interpretation is correct, this preferred orientation would be influenced both by preexisting olivine CPOs, and by the orientation of tectonic stress, in proportion of the misorientation and of the relative magnitude of tectonic versus thermal stresses. The reorientation proposed (Figure 11) for initial serpentinizing microfractures at ODP Site 920 is therefore not necessarily representative of what happens at mid-ocean ridge detachment faults in general. However, since it is our only example, in which the present-day dip of one set of initial microfractures is known, it is worth noting that in this particular instance this dip is moderate. If the absolute reorientation of the ODP site 920 cores discussed by Hurst et al. [1997] is correct, this set of microfractures would have a dominantly north to north-northwest strike and an eastern dip, near parallel to the emerging detachment fault at this location [Karson and Lawrence, 1997]. The samples from the Southwest Indian Ridge present crosscutting relationships (Figures 7 d and 7e) that suggest a separation in time between the formation of discrete brittle shear zones in the peridotite and the later formation of the initial serpentinizing microfractures. A working hypothesis is that the observed brittle shear zones formed at the final stages of plastic deformation. Subsequent deformation would then focus in the detachment fault, and tectonic stresses would decrease below yield strength in the bulk of the footwall, allowing for thermal stresses to accumulate with progressive cooling as sketched in Figure 12. ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2373 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 The thickness of the layer, over which cracking is inferred (Figure 12), is expected to be variable, depending on the efficiency and geometry of hydrothermal heat extraction. Its depth is at least 8 km because microseismicity typically extends to this depth at the axis of the Mid-Atlantic Ridge [Toomey et al., 1985; Wolfe et al., 1995; deMartin et al., 2007]. A recent microseismicity study at the Knipovich Ridge [Schlindwein et al., 2013] suggests that the brittle to ductile transition could be as deep as 20 km along some portions of ultraslow spreading ridges. deMartin et al. [2004] estimated a maximum depth of 4–6 km for thermal cracking at a slow spreading mid-ocean ridge based on the balance between the thermal stress intensities they calculated with the micromechanical model of Fredrich and Wong [1986] and confining pressure (that counteracts extensional thermal stresses). This approach, however, does not take detachment faulting and the associated tectonic stresses into account. The yield strength of rocks in the brittle domain increases with depth [Byerlee, 1978] and exceeds the confining pressure (Goetze’s criterion) [Kohlstedt et al., 1995]. Tectonic loading in the footwall of the detachment does not reach the yield strength except in the fault zone(s), yet peridotites that are being exhumed into the brittle domain (Figure 12) experience deviatoric stresses that can partially counteract confining pressure. Introducing tectonic stresses into the picture therefore offers a way to account for the formation of initial serpentinizing fractures on axis at depths greater than estimated by deMartin et al. [2004]. The region sketched in Figure 12 represents only the portion of the detachment’s footwall that is closest to the fault: i.e., the domain that will become the shallow seismically defined crust as exhumation proceeds (Figure 1). We do not have the samples to document serpentinization-related processes in the footwall far from the fault, because ultramafic samples from slow spreading ridges have so far been recovered either at the seafloor, or in shallow drill holes (Figure 1). 9.2. From the Onset of Serpentinization to the Fully Developed Mesh Texture As serpentinization initiates, volume change will induce additional, reaction-driven cracking, producing an increasingly interconnected, three-dimensional network of closely spaced microfractures for fluid infiltration. Our results suggest that, in the case of abyssal serpentinized peridotites from slow spreading ridges, hierarchical reaction-driven microfractures combine with the initial microfracture network. The average dimension (60 mm) of the domains (or cells) defined by these combined microfractures does not appear to vary for serpentinization extents between 18% (the minimum local serpentinization extents found in our set of samples) and 70% (Figure 8c). This indicates that reaction-induced microfracturing mainly occurs between 0 and 20% serpentinization. Since the mesh rim formation stage of serpentinization is probably isochemical for major elements [Coleman and Keith, 1971; Shervais et al., 2005; Bach et al., 2006; Paulick et al., 2006] there should still be an increase of volume as the reaction proceeds past 20%. A possible explanation is that hierarchical fracturing of olivine relicts is suppressed, when the extent of serpentinization exceeds 20%, because the stresses induced in the small olivine relicts (diameter < 60 mm on average) by the ongoing reaction are dissipated by deformation of the adjacent serpentine rims and by frictional sliding along the abundant network of microfractures. Not all the initial and hierarchical microfractures are coated by well-developed serpentine rims. Quite a few have very thin rims (2 mm), and the network of microfractures with well-developed rims (i.e., >10 mm) defines larger olivine domains, or cells, typically between 100 and 400 mm in diameter as reported in the literature for serpentine mesh in samples from a variety of geodynamic contexts [Wicks et al., 1977; Prichard, 1979; Viti and Mellini, 1998]. Using mesh rim width as a proxy for the local extent of serpentinization, this suggests that all microfractures do not act as equally efficient fluid pathways during meshwork-forming serpentinization. This may be because they represent poorly connected portions of the permeability network from the start, or because they close due to reaction-induced volume increase after incipient serpentinization [e.g., Emmanuel and Berkowitz, 2006]. In any case, the relevant mesh cell dimensions (100–400 mm in diameter) during meshwork-forming serpentinization are 2–4 times greater than the scales of initial and hierarchical microfracturing (60 mm in diameter). Another characteristic of mesh-textured serpentinites is that lizardite pseudocolumns forming the mesh rims rarely reach the mesh cell centers (Figure 3). In section 6, we measured the apparent thickness of serpentine along microfractures in several well-serpentinized samples (Table 1 and Figure 9) and found an average value of 66 6 38 mm, corresponding to serpentinization extents of 70–80%. Isotropic serpentine fills the mesh cores at a later stage, replacing the olivine relicts [Dungan, 1977; Wicks et al., 1977; Viti and Mellini, 1998]. Viti and Mellini [1998] proposed that mesh rims form in conditions of low fluid/rock ratio ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2374 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 and cease to grow because the available fluid has been consumed, while the replacement of the olivine cores by small serpentine crystals forming isotropic assemblages occurs later at conditions of higher fluid/rock ratio. Given that lizardite pseudocolumns grow from the microfractures into the serpentinizing olivine grains [e.g., Rumori et al., 2004; Boudier et al., 2010], and assuming that none of the lizardite grew in an open microfracture via precipitation from a fluid bringing Mg, Fe and Si into the rock from some other rock volume, the widths of serpentine mesh rims represent twice the mean distance from permeable microfracture to the reaction front where lizardite replaces olivine. If some of the lizardite did precipitate directly from fluid in an open fracture, and/or if the lizardite volume is significantly larger than the volume of olivine that has been replaced in a given sample, then the widths of the serpentine mesh rims represent upper bounds on the thickness of the olivine that has been replaced. In turn, these upper bounds, divided by the time interval of lizardite formation, will yield upper bounds on the rate of olivine replacement. Experiments carried out at 250 C–350 C by Malvoisin et al. [2012] with aggregates of 50–65 mm sized grains of San Carlos olivine have reached 80% serpentinization in about 500 days. Using these rates for natural samples would lead, if fluid supply is adequate, to the formation of a well-developed mesh texture (i.e., full width of serpentine rims between 30 and 140 mm, Table 1, and serpentinization extents 70–80%) in 500 to 1100 days. Allowing for some direct precipitation of lizardite would not change the order of magnitude of these rates that are fast compared to exhumation rates: 17 mm yr21 (Southwest Indian Ridge) or 24 mm yr21 (northern MidAtlantic Ridge) in the conveyor belt configuration of Figure 1. At these plate tectonic timescales, the formation of the serpentine mesh texture can therefore be modeled as a quasi-instantaneous process occurring when the microfractured peridotites come into contact with hydrous fluids in the domain of hydrothermal circulation (Figure 12). Alternatively, the serpentine could have formed over a longer time period, controlled by small and intermittent supplies of aqueous fluids. 9.3. Time and Space Heterogeneity of Fluid Circulations and the Development of the Serpentine Mesh Texture In our study, we considered evidence for fluid availability at the microscopic and sample scales (<30 cm). We show that microfracture densities are more or less constant for serpentinization extents between 18% and 70% (Figure 8c), indicating that microfracturing alone does not guarantee a fluid supply sufficient for mesh texture formation to proceed to completion. Sample 920D-18R2, 113–118 cm, piece 13 is a good example, with an extensively serpentinized ribbon-shaped mesh texture on one side of the thin section (Figure 6b) in sharp contact with a less serpentinized domain that displays similarly spaced microfractures (Figures 7a and 7b). This indicates that at the spatial scales documented here (i.e., <30 cm), fluid flux is variable. Variations of the serpentinization extents at the larger scale of the drill hole at ODP Site 920 [Dilek et al., 1997] also suggest heterogeneities in fluid flux. The limit of the hydrothermally cooled domain where serpentinization occurs (Figure 12) is therefore probably complex, with strong topography at a variety of scales. Spatial and temporal heterogeneity of the fluid circulation patterns is predicted by porous flow models of hydrothermal circulation in detachment faults and in their footwall [e.g., Fontaine et al., 2009]. These effects will be enhanced by complexities of the permeable structure, such as due to tectonic stresses keeping specific orientations of microfractures open. Time and space heterogeneity of fluid distribution should also be enhanced by serpentinization-induced volume increase in fluid-rich domains. This will promote cracking at a range of scales [O’Hanley, 1992; Jamtveit et al., 2008; 2009; Kelemen and Hirth, 2012; Jamtveit and Hammer, 2012] and create channels for fluids into previously impermeable domains. In this way, the base of the serpentinization domain interface sketched in Figure 12 will evolve dynamically as serpentinization proceeds. Finally, it is probable that tectonic stresses that help to create the initial microfractures in the fresh peridotites are lower in regions of the footwall that are away from the detachment fault (Figure 1). As a result, thermal stresses there would be more effectively counteracted by confining pressure and the initial microfracture network may be less pervasive. All these factors combined argue for large time and space heterogeneity of the mesh texture forming serpentinization in the mantle-derived peridotites exhumed at slow spreading ridges. 10. Conclusions Our microstructural and petrographic study of the serpentine meshwork was carried out on 278 variably serpentinized abyssal peridotites sampled at or near the seafloor at the slow spreading Mid-Atlantic and ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2375 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 Southwest Indian ridges. Serpentinization of these samples occurred in the footwall within a few hundred meters of axial detachment faults. Our study reveals that the serpentine mesh texture developed from a network of microfractures that has consistent characteristics and dimensions in all samples. This network comprises intersecting sets of initial serpentinizing microfractures that commonly extend across several grains of olivine and into adjacent pyroxenes. These initial microfractures have a preferred orientation at sample scale (and at 100 m scale at ODP Site 920). We propose that they form due to stresses caused by a combination of anisotropic thermal contraction and tectonic stresses. These microfractures generate a connected porosity capable of transporting hydrothermal fluids. The onset of serpentinization at temperatures <400 C leads to additional reaction-driven (hierarchical) cracking, forming a crack network with a typical spacing of 60 mm. This study shows that all microfractures do not act as equally efficient fluid pathways during meshworkforming serpentinization. This is illustrated by differences in the thickness of the associated serpentine rims. Those microfractures that have serpentine thicknesses >10 mm are spaced by 100–400 mm. The relevant cell dimensions during meshwork-forming serpentinization are therefore 2–4 times greater than the scales of initial and hierarchical microfracturing. Serpentine thickness, representing twice the distance between the permeable microfracture and the reaction front, can be used as an upper bound for the advancement of mesh-forming serpentinization. In the most serpentinized samples, lizardite selvages on microfractures have an average apparent thickness of 33 6 19 mm, corresponding to a serpentinization extent of 70–80%. Based on published laboratory experiments, the formation of these mesh rims could be completed in a few years and therefore be quasi instantaneous at the plate tectonic timescale. We propose that formation of the serpentine mesh texture in ultramafic rocks exhumed at slow spreading ridges occurs near the base of the hydrothermally cooled domain adjacent to the detachment faults (Figure 12). We envision that the domain of active serpentinization is limited by the 400 C isotherm and that this limit is complex, with strong topography at a variety of scales. We expect that it is also highly variable in time, due both to timedependant hydrothermal circulation in the detachment footwall, and to reaction-induced cracking allowing for enhanced fluid flux into previously less permeable domains. As a result, we expect that mesh-texture formation during serpentinization at slow spreading ridges occurs in a nonsteady state and spatially heterogeneous fashion during tectonic exhumation of the peridotites. Acknowledgments The authors thank Françoise Boudier, David Mainprice, Benedicte M enez, Mark Van Zuilen, Gianreto Manatschal, and Javier Escartın for their help and insight. This manuscript benefited from constructive reviews from Peter B. Kelemen, Bruno Reynard, and an anonymous reviewer. This work was supported by the Agence Nationale de la Recherche (Rift2Ridge project). This is IPGP contribution number 3502. ROUMEJON AND CANNAT References Agrinier, P., and M. Cannat (1997), Oxygen-isotope constraints on serpentinization processes in ultramafic rocks from the mid-Atlantic ridge (23 N), Proc. Ocean Drill. Program Sci. Results, 153, 381–388. Allen, D. E., and W. E. Seyfried (2004), Serpentinization and heat generation: Constraints from Lost City and Rainbow hydrothermal systems, Geochim. Cosmochim. Acta, 68(6), 1347–1354, doi:10.1016/j.gca.2003.09.003. Andreani, M., C. Mevel, A.-M. Boullier, and J. Escartın (2007), Dynamic control on serpentine crystallization in veins: Constraints on hydration processes in oceanic peridotites, Geochem. Geophys. Geosyst., 8, Q02012, doi:10.1029/2006GC001373. Auzende, A. L., I. Daniel, B. Reynard, C. Lemaire, and F. Guyot (2004), High-pressure behaviour of serpentine minerals: A Raman spectroscopy study, Phys. Chem. Miner., 31, 269–277, doi:10.1007/s00269-004-0384-0. Bach, W., C. J. Garrido, H. Paulick, J. Harvey, and M. Rosner (2004), Seawater-peridotite interactions: First insights from ODP Leg 209, MAR 15 N, Geochem. Geophys. Geosyst., 5, Q09F26, doi:10.1029/2004GC000744. Bach, W., H. Paulick, C. J. Garrido, B. Ildefonse, W. P. Meurer, and S. E. Humphris (2006), Unraveling the sequence of serpentinization reactions: Petrography, mineral chemistry, and petrophysics of serpentinites from MAR 15 N (ODP Leg 209, Site 1274), Geophys. Res. Lett., 33, L13306, doi:10.1029/2006GL025681. Behn, M. D., and P. B. Kelemen (2003), Relationship between seismic P-wave velocity and the composition of anhydrous igneous and meta-igneous rocks, Geochem. Geophys. Geosyst., 4(5), 1041, doi:10.1029/2002GC000393. Berndt, M. E., D. E. Allen, and W. E. Seyfried (1996), Reduction of CO2 during serpentinization of olivine at 300 C and 500 bar, Geology, 24(4), 351–354, doi:10.1130/0091-7613(1996)024<0351:ROCDSO>2.3.CO;2. Boschi, C., A. Dini, G. L. Fr€ uh-Green, and D. S. Kelley (2008), Isotopic and element exchange during serpentinization and metasomatism at the Atlantis Massif (MAR 30 N): Insights from B and Sr isotope data, Geochim. Cosmochim. Acta, 72(7), 1801–1823, doi:10.1016/ j.gca.2008.01.013. Boudier, F., A. Baronnet, and D. Mainprice (2010), Serpentine mineral replacements of natural olivine and their seismic implications: Oceanic lizardite versus subduction-related antigorite, J. Petrol., 51, 495–512, doi:10.1093/petrology/egp049. Bouhifd, M., D. Andrault, G. Fiquet, and P. Richet (1996), Thermal expansion of forsterite up to melting point, Geophys. Res. Lett., 23(10), 1143–1146. Byerlee, J. (1978), Friction of rocks, Pure Appl. Geophys., 116(4-5), 615–626. Canales, J. P., J. A. Collins, J. Escartın, and R. S. Detrick (2000), Seismic structure across the rift valley of the Mid-Atlantic Ridge at 23 20’ (MARK area): Implications for crustal accretion processes at slow spreading ridges, J. Geophys. Res., 105(B12), 28,411–28,425. C 2014. American Geophysical Union. All Rights Reserved. V 2376 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 Cann, J., D. K. Blackman, D. K. Smith, E. McAllister, B. Janssen, S. Mello, E. Avgerinos, A. R. Pascoe, and J. Escartın (1997), Corrugated slip surfaces formed at ridge-transform intersections on the Mid-Atlantic Ridge, Nature, 385, 329–332. Cannat, M. (1993), Emplacement of mantle rocks in the seafloor at mid-ocean ridges, J. Geophys. Res., 98(B3), 4163–4172. Cannat, M., D. Bideau, and H. Bougault (1992), Serpentinized peridotites and gabbros in the Mid-Atlantic Ridge axial valley at 15 370 N and 16 520 N, Earth Planet. Sci. Lett., 109, 87–106. Cannat, M., et al. (1995a), Thin crust, ultramafic exposures, and rugged faulting patterns at the Mid-Atlantic Ridge (22 –24 N), Geology, 23(1), 49–52, doi:10.1130/0091-7613(1995)023<0049. Cannat, M., J. A. Karson, D. J. Miller and the Shipboard Scientific Party of ODP Leg 153 (1995b), Proceedings of the Ocean Drilling Program, Initial Reports, vol. 153, Ocean Drill. Program, College Station, Tex. Cannat, M., Y. Lagabrielle, H. Bougault, J. Casey, N. de Coutures, L. Dmitriev, and Y. Fouquet (1997), Ultramafic and gabbroic exposures at the Mid-Atlantic Ridge: Geological mapping in the 15 N region, Tectonophysics, 279, 193–213. Cannat, M., F. Fontaine, and J. Escartın (2010), Serpentinization and associated hydrogen and methane fluxes at slow spreading ridges, in Diversity of Hydrothermal Systems on Slow Spreading Ocean Ridges, Geophys. Monogr. Ser., edited by P. A. Rona et al., pp. 241–264, AGU, Washington, D. C., doi:10.1029/2008GM000760. Cannat, M., D. Sauter, and S. Roum ejon (2012), Formation of an ultramafic seafloor at the Southwest Indian Ridge 62 -65 E: Internal structure of detachment faults and sparse volcanism documented by sidescan sonar and dredges, Abstract OS11E-03TI presented at 2012 Fall Meeting, AGU, San Francisco, Calif. Carter, N. L., and H. G. Av e Lallemant (1970), High temperature flow of dunite and peridotite, Geol. Soc. Am. Bull., 81(8), 2181–2202. Ceuleneer, G., and M. Cannat (1997), High-temperature ductile deformation of site 920 peridotites, Proc. Ocean Drill. Program Sci. Results, 153, 23–34. Charlou, J.-l., and J.-p. Donval (1993), Hydrothermal methane venting between 12 N and 26 N along the mid-atlantic ridge, J. Geophys. Res., 98(B6), 9625–9642. Charlou, J.-L., Y. Fouquet, H. Bougault, J.-P. Donval, J. Etoubleau, P. Jean-baptiste, A. Dapoigny, P. Appriou, and P. A. Rona (1998), Intense CH 4 plumes generated by serpentinization of ultramafic rocks at the intersection of the 15 200 N fracture zone and the Mid-Atlantic Ridge, Geochim. Cosmochim. Acta, 62(13), 2323–2333. Charlou, J.-L., J.-P. Donval, Y. Fouquet, P. Jean-baptiste, and N. Holm (2002), Geochemistry of high H2 and CH4 vent fluids issuing from ultramafic rocks at the Rainbow hydrothermal field (36 14’N, MAR), Chem. Geol., 191, 345–359. Christensen, N. I. (1972), The abundance of serpentinites in the oceanic crust, J. Geol., 80, 709–719. Coleman, R. G., and T. E. Keith (1971), A chemical study of serpentinization—Burro Moutain, California, J. Petrol., 12(2), 311–328. Contreras-Reyes, E., I. Grevemeyer, E. R. Flueh, and C. Reichert (2008), Upper lithospheric structure of the subduction zone offshore of southern Arauco peninsula, Chile, at 38 S, J. Geophys. Res., 113, B07303, doi:10.1029/2007JB005569. Cressey, B. A. (1979), Electron microscopy of serpentinite textures, Can. Mineral., 17, 741–756. deMartin, B. J., G. Hirth, and B. Evans (2004), Experimental constraints on thermal cracking of peridotite at oceanic spreading centers, in Mid-Ocean Ridges: Hydrothermal Interactions Between the Lithosphere and Oceans, vol. 148, pp. 167–185, Geophysical Monograph Series, doi:10.1029/148GM07. deMartin, B. J., R. A. Sohn, J. P. Canales, and S. E. Humphris (2007), Kinematics and geometry of active detachment faulting beneath the Trans-Atlantic Geotraverse (TAG) hydrothermal field on the Mid-Atlantic Ridge, Geology, 35(8), 711–714, doi:10.1130/G23718A.1. Detrick, R. S., J. A. Collins, R. Stephen, and S. Swift (1994), In situ evidence for the nature of the seismic layer 2/3 boundary in oceanic crust, Nature, 370(6487), 288–290. Dewandel, B., F. Boudier, H. Kern, W. Warsi, and D. Mainprice (2003), Seismic wave velocity and anisotropy of serpentinized peridotite in the Oman ophiolite, Tectonophysics, 370(1-4), 77–94, doi:10.1016/S0040-1951(03)00178-1. Dick, H. (1989), Abyssal peridotites, very slow spreading ridges and ocean ridge magmatism, Geol. Soc. Spec. Publ., 42, 71–105. Dick, H. J. B., J. Lin, and H. Schouten (2003), An ultraslow-spreading class of ocean ridge, Nature, 426(6965), 405–412, doi:10.1038/ nature02128. Dilek, Y., A. Coulton, and S. D. Hurst (1997), Serpentinization and hydrothermal veining in peridotites at site 920 in the MARK area, Proc. Ocean Drill. Program Sci. Results, 153, 35–59. Dungan, M. A. (1977), Metastability in serpentine-olivine equilibria, Am. Mineral., 62, 1018–1029. Emmanuel, S., and B. Berkowitz (2006), Suppression and stimulation of seafloor hydrothermal convection by exothermic mineral hydration, Earth Planet. Sci. Lett., 243, 657–668, doi:10.1016/j.epsl.2006.01.028. Escartın, J., G. Hirth, and B. Evans (1997), Effects of serpentinization on the lithospheric strength and the style of normal faulting at slowspreading ridges, Earth Planet. Sci. Lett., 151, 181–189. Fontaine, F. J., J. Escartın, and M. Cannat (2009), Hydrothermal heat extraction at the center of slow-spreading mid-ocean ridges: Controls by fault systems and focused magmatism, EOS Trans. AGU, 90(52), Fall Meet. Suppl., Abstract #OS11B-06. Francis, G. H. (1956), The serpentinite mass in Glen Urquhart, Inverness-Shire, Scotland, Am. J. Sci., 254, 201–226. Fredrich, J., and T. Wong (1986), Micromechanics of thermally induced cracking in three crustal rocks, J. Geophys. Res., 91(B12), 12,743– 12,764. Frost, B. R. (1985), On the stability of sulfides, oxides, and native metals in serpentinite, J. Petrol., 26(1), 31–63. Fr€ uh-Green, G. L., J. A. Connolly, A. Plas, D. S. Kelley, and B. Grob ety (2004), Serpentinization of oceanic peridotites: Implications for geochemical cycles and biological activity, in The Subseafloor Biosphere at Mid-Ocean Ridges, Geophys. Monogr. Ser., vol. 144, edited by W. S. Wilcock et al., pp. 119–136, AGU, Washington, D. C., doi:10.1029/144GM08. Garc es, M., and J. Gee (2007), Paleomagnetic evidence of large footwall rotations associated with low-angle faults at the Mid-Atlantic Ridge, Geology, 35(3), 279–282, doi:10.1130/G23165A.1. Gracia, E., J.-L. Charlou, J. Radford-Knoery, and L. M. Parson (2000), Non-transform offsets along the Mid-Atlantic Ridge south of the Azores (38 N–34 N): Ultramafic exposures and hosting of hydrothermal vents, Earth Planet. Sci. Lett., 177, 89–103. Hess, H. H. (1962), History of Ocean Basins, Petrologic Studies, 4, 599–620. Hilairet, N., B. Reynard, Y. Wang, I. Daniel, S. Merkel, N. Hilairet, N. Nishiyama, and S. Petitgirard (2007), High-pressure creep of serpentine, interseismic deformation, and initiation of subduction, Science, 318(5858), 1910–1913. Hirauchi, K., K. Michibayashi, H. Ueda, and I. Katayama (2010), Spatial variations in antigorite fabric across a serpentinite subduction channel: Insights from the Ohmachi Seamount, Izu-Bonin frontal arc, Earth Planet. Sci. Lett., 299(1-2), 196–206, doi:10.1016/j.epsl.2010.08.035. Hooft, E., R. Detrick, D. Toomey, J. Collins, and J. Lin (2000), Crustal thickness and structure along three contrasting spreading segments of the Mid-Atlantic Ridge, 33.5 -35 N, J. Geophys. Res., 105(B4), 8205–8226. ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2377 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 Hurst, S., J. Gee, and R. Lawrence (1997), Data report: Reorientation of structural features at Sites 920 to 924 using remanent magnetization and magnetic characteristics, Proc. Ocean Drill. Program Sci. Results, 153, 547–559. Iyer, K., B. Jamtveit, J. Mathiesen, A. Malthe-Sïrenssen, and J. Feder (2008), Reaction-assisted hierarchical fracturing during serpentinization, Earth Planet. Sci. Lett., 267, 503–516, doi:10.1016/j.epsl.2007.11.060. Jamtveit, B., and Ø. Hammer (2012), Hierarchical fracturing during weathering and serpentinisation, Geochem. Perspect., 1(2006), 418–432. Jamtveit, B. r., A. Malthe-Sïrenssen, and O. Kostenko (2008), Reaction enhanced permeability during retrogressive metamorphism, Earth Planet. Sci. Lett., 267, 620–627, doi:10.1016/j.epsl.2007.12.016. Jamtveit, B., C. V. Putnis, and A. Malthe-Sïrenssen (2009), Reaction induced fracturing during replacement processes, Contrib. Mineral. Petrol., 157, 127–133, doi:10.1007/s00410-008-0324-y. Karson, J. A. (1990), Seafloor spreading on the Mid-Atlantic Ridge: Implications for the structure of ophiolites and oceanic lithosphere produced in slow-spreading environments, in Proceedings of the Symposium TROODOS 1987, edited by J. Malpas et al., pp. 547–555, Geol. Surv. Dep., Nicosia, Cyprus. Karson, J. A., and R. M. Lawrence (1997), Tectonic setting of serpentinite exposures on the western median valley wall of the MARK area in the vicinity of site 920, Proc. Ocean Drill. Program Sci. Results, 153, 5–21. Karson, J. A., et al. (1987), Along-axis variations in seafloor spreading in the MARK area, Nature, 328(6132), 681–685, doi:10.1038/328681a0. Katz, Y., R. Weinberger, and A. Aydin (2004), Geometry and kinematic evolution of Riedel shear structures, Capitol Reef National Park, Utah, J. Struct. Geol., 26, 491–501, doi:10.1016/j.jsg.2003.08.003. Kelemen, P. B., and G. Hirth (2012), Reaction-driven cracking during retrograde metamorphism: Olivine hydration and carbonation, Earth Planet. Sci. Lett., 345–348, 81–89, doi:10.1016/j.epsl.2012.06.018. Kelemen, P. B., E. Kikawa, D. J. Miller, and Shipboard Scientific Party of ODP Leg 209 (2004), Proceedings of the Ocean Drilling Program Initial Reports, vol. 209, Ocean Drill. Program, College Station, Tex. Kelemen, P. B., J. Matter, E. E. Streit, J. F. Rudge, W. B. Curry, and J. Blusztajn (2011), Rates and mechanisms of mineral carbonation in peridotite: Natural processes and recipes for enhanced, in situ CO2 capture and storage, Annu. Rev. Earth Planet. Sci., 39(1), 545–576, doi: 10.1146/annurev-earth-092010-152509. Kelley, D. S., et al. (2001), An off-axis hydrothermal vent field near the Mid-Atlantic Ridge at 30 N, Nature, 412(6843), 145–149, doi:10.1038/ 35084000. Kohlstedt, D. L., B. Evans, and S. J. Mackwell (1995), Strength of the lithosphere: Constraints imposed by laboratory experiments, J. Geophys. Res., 100(B9), 17,587–17,602. Korenaga, J., P. B. Kelemen, and W. S. Holbrook (2002), Methods for resolving the origin of large igneous provinces from crustal seismology, J. Geophys. Res., 107(B9), 2178, doi:10.1029/2001JB001030. Lowell, R. and P. Rona (2002), Seafloor hydrothermal systems driven by the serpentinization of peridotite, Geophys. Res. Lett., 29, 26-1–26-4. Lavier, L. L., W. R. Buck, and A. N. B. Poliakov (1999), Self-consistent rolling-hinge model for the evolution of large-offset low-angle normal faults, Geology, 27(12), 1127–1130, doi:10.1130/0091-7613(1999)027<1127:SCRHMF>2.3.CO;2. MacLeod, C. J., J. Carlut, J. Escartın, H. Horen, and A. Morris (2011), Quantitative constraint on footwall rotations at the 15 450 N oceanic core complex, Mid-Atlantic Ridge: Implications for oceanic detachment fault processes, Geochem. Geophys. Geosyst., 12, Q0AG03, doi: 10.1029/2011GC003503. Maltman, A. J. (1978), Serpentinite textures in Anglesey, North Wales, United Kingdom, Geol. Soc. Am. Bull., 89, 972–980, doi:10.1130/00167606(1978)89<972:STIANW>2.0.CO;2. Malvoisin, B., F. Brunet, J. Carlut, S. Roum ejon, and M. Cannat (2012), Serpentinization of oceanic peridotites: 2. Kinetics and processes of San Carlos olivine hydrothermal alteration, J. Geophys. Res., 117, B04102, doi:10.1029/2011JB008842. Michael, P. J., et al. (2003), Magmatic and amagmatic seafloor generation at the ultraslow-spreading Gakkel ridge, Arctic Ocean, Nature, 423(6943), 956–961, doi:10.1038/nature01704. Miller, D. J., and N. I. Christensen (1997), Seismic velocities of lower crustal and upper mantle rocks form the slow-spreading Mid-Atlantic Ridge, south of the Kane transform zone (MARK), Proc. Ocean Drill. Program Sci. Results, 153, 437–454. Morris, A., J. S. Gee, N. Pressling, B. E. John, C. J. MacLeod, C. B. Grimes, and R. C. Searle (2009), Footwall rotation in an oceanic core complex quantified using reoriented Integrated Ocean Drilling Program core samples, Earth Planet. Sci. Lett., 287(1-2), 217–228, doi:10.1016/ j.epsl.2009.08.007. O’Hanley, D. S. (1992), Solution to the volume problem in serpentinization, Geology, 20, 705–708. Oufi, O., M. Cannat, and H. Horen (2002), Magnetic properties of variably serpentinized abyssal peridotites, J. Geophys. Res., 107(B5), doi: 10.1029/2001JB000549. Paulick, H., W. Bach, M. Godard, J. C. M. De Hoog, G. Suhr, and J. Harvey (2006), Geochemistry of abyssal peridotites (Mid-Atlantic Ridge, 15 200 N, ODP Leg 209): Implications for fluid/rock interaction in slow spreading environments, Chem. Geol., 234(3-4), 179–210, doi: 10.1016/j.chemgeo.2006.04.011. Picazo, S., M. Cannat, A. Delacour, J. Escartın, S. Roum ejon, and S. Silantyev (2012), Deformation associated with the denudation of mantlederived rocks at the Mid-Atlantic Ridge 13 –15 N: The role of magmatic injections and hydrothermal alteration, Geochem. Geophys. Geosyst., 13, Q04G09, doi:10.1029/2012GC004121. Pl€ umper, O., A. Rïyne, A. Magras o, and B. Jamtveit (2012), The interface-scale mechanism of reaction-induced fracturing during serpentinization, Geology, 40(12), 1103–1106, doi:10.1130/G33390.1. Prichard, H. M. (1979), A petrographic study of the process of serpentinisation in ophiolites and the ocean crust, Contrib. Mineral. Petrol., 68, 231–241. Ranero, C. R., J. P. Morgan, K. McIntosh, and C. Reichert (2003), Bending-related faulting and mantle serpentinization at the Middle America trench, Nature, 425(6956), 367–373, doi:10.1038/nature01961. Reinen, L. A., J. D. Weeks, and T. E. Tullis (1994), The frictional behavior of lizardite and antigorite serpentinites: Experiments, constitutive models, and implications for natural faults, Pure Appl. Geophys., 143(1-3), 317–358, doi:10.1007/BF00874334. Roum ejon, S., M. Cannat, P. Agrinier, M. Godard, M. Andreani (2013), Microfracturing and fluid pathways in serpentinizing abyssal peridotites along the Southwest Indian Ridge (62 -65 E), Abstract MR22A-05 presented at 2013 Fall Meeting, AGU, San Francisco, Calif. Rudge, J. F., P. B. Kelemen, and M. Spiegelman (2010), A simple model of reaction-induced cracking applied to serpentinization and carbonation of peridotite, Earth Planet. Sci. Lett., 291, 215–227, doi:10.1016/j.epsl.2010.01.016. Rumori, C., M. Mellini, and C. Viti (2004), Oriented, non-topotactic olivine serpentine replacement in mesh-textured, serpentinized peridotites, Eur. J. Mineral., 16, 731–741, doi:10.1127/0935-1221/2004/0016-0731. Sauter, D., et al. (2013), Continuous exhumation of mantle-derived rocks at the Southwest Indian Ridge for 11 million years, Nat. Geosci., 6(4), 314–320, doi:10.1038/ngeo1771. ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2378 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems 10.1002/2013GC005148 Schlindwein, V., A. Demuth, W. H. Geissler, and W. Jokat (2013), Seismic gap beneath Logachev Seamount: Indicator for melt focusing at an ultraslow mid-ocean ridge?, Geophys. Res. Lett., 40(9), 1703–1707, doi:10.1002/grl.50329. Schwartz, S., S. Guillot, B. Reynard, R. Lafay, B. Debret, C. Nicollet, P. Lanari, and A. L. Auzende (2013), Pressure–temperature estimates of the lizardite/antigorite transition in high pressure serpentinites, Lithos, doi:10.1016/j.lithos.2012.11.023. Seyler, M., M. Cannat, and C. M evel (2003), Evidence for major-element heterogeneity in the mantle source of abyssal peridotites from the Southwest Indian Ridge (52 to 68 E), Geochem. Geophys. Geosyst., 4(2), 9101, doi:10.1029/2002GC000305. Shervais, J. W., P. Kolesar, and K. Andreasen (2005), A field and chemical study of serpentinization—Stonyford, California: Chemical flux and mass balance, Int. Geol. Rev., 47(1), 1–23, doi:10.2747/0020-6814.47.1.1. Shock, E. L., and M. E. Holland (2004), Geochemical energy sources that support the subsurface biosphere, in Subseafloor Biosphere at MidOcean Ridges, Geophysical Monograph Series, pp. 153–165. Small, C. (1998), Global systematics of mid-ocean ridge morphology, in Faulting and Magmatism at Mid-Ocean Ridges, Geophys. Monogr. Ser., vol. 106, edited by W. R. Buck et al., pp. 1–25, AGU, Washington, D. C., doi:10.1029/GM106p0001. Smith, D. K., J. R. Cann, and J. Escartın (2006), Widespread active detachment faulting and core complex formation near 13 N on the MidAtlantic Ridge, Nature, 442(7101), 440–443, doi:10.1038/nature04950. Spudich, P., and J. Orcutt (1980), Petrology and porosity of an oceanic crustal site: Results from wave form modeling of seismic refraction data, J. Geophys. Res., 85(B3), 1409–1433. Suzuki, I. (1975), Thermal expansion of periclase and olivine, and their anharmonic properties, J. Phys. Earth, 23(2), 145–159. Swain, M. V., and B. K. Atkinson (1978), Fracture surface energy of olivine, Pure Appl. Geophys., 116, 866–872. Tapponnier, P., and W. F. Brace (1976), Development of stress-induced microcracks in Westerly granite, Int. J. Rock Mech. Min. Sci., 13, 103– 112. Toomey, D. R., S. C. Solomon, G. M. Purdy, and M. H. Murray (1985), Microearthquakes beneath the median valley of the mid-atlantic ridge near 23 N: Hypocenters and focal mechanisms, J. Geophys. Res., 90(4), 5443–5458. Ulmer, P., and V. Trommsdorff (1995), Serpentine stability to mantle depths and subduction-related magmatism, Science, 268(5212), 858– 861, doi:10.1126/science.268.5212.858. Viti, C., and M. Mellini (1998), Mesh textures and bastites in the Elba retrograde serpentinites, Eur. J. Mineral., 10, 1341–1359. Wicks, F. J. (1984), Deformation histories as recorded by serpentinites; III, fracture patterns developed prior to serpentinization, Can. Mineral., 22, 205–209. Wicks, F. J., and E. J. W. Whittaker (1977), Serpentine textures and serpentinization, Can. Mineral., 15, 459–488. Wicks, F. J., and J. Zussman (1975), Microbeam X-ray diffraction patterns of the serpentine minerals, Can. Mineral., 13(3), 244–258. Wicks, F. J., E. J. W. Whittaker, and J. Zussman (1977), An Idealized Model for serpentine textures after olivine, Can. Mineral., 15, 446–458. Wolfe, C. J., G. M. Purdy, D. R. Toomey, and S. C. Solomon (1995), Microearthquake characteristics and crustal velocity structure at 29 N on the Mid-Atlantic Ridge: The architecture of a slow spreading segment, J. Geophys. Res., 100(B12), 24,449–24,472. ROUMEJON AND CANNAT C 2014. American Geophysical Union. All Rights Reserved. V 2379 15252027, 2014, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2013GC005148 by Cochrane France, Wiley Online Library on [28/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Geochemistry, Geophysics, Geosystems