Chlorine stable isotopes and halogen concentrations in convergent margins with implications for the Cl isotopes cycle in the ocean

Available online at www.sciencedirect.com

Earth and Planetary Science Letters 266 (2008) 90 – 104 www.elsevier.com/locate/epsl

Chlorine stable isotopes and halogen concentrations in convergent margins with implications for the Cl isotopes cycle in the ocean Wei Wei a,⁎, Miriam Kastner a , Arthur Spivack b a

Scripps Institution of Oceanography, University of California, San Diego, La Jolla, CA 92093, United States b Graduate School of Oceanography, University of Rhode Island, Narragansett, RI 02882, United States Received 30 July 2007; received in revised form 1 November 2007; accepted 2 November 2007 Available online 17 November 2007 Editor: R.W. Carlson

Abstract Chlorine stable isotopes (δ37Cl) and halogen concentrations (e.g. Br/Cl) in 168 pore Fluids and 23 serpentines and other solids from three subduction zones, the Nankai Trough, Costa Rica, and Mariana Forearc, provide critical information on fluid sources, flow paths, and reaction conditions. The δ37Cl values of pore fluids at the Nankai and Costa Rica subduction zones, are significantly more negative (minimum −7.8‰, 2σ ± 0.3‰) than seawater value (0‰). At Nankai Trough, the minimum δ37Cl value is situated below the décollement and evolves laterally from −7.8‰ at the most arcward ODP Site 808, to − 7.1‰ at Site 1174, ∼ 2 km seaward from Site 808, and to −5.8‰ at the reference Site 1173. At Costa Rica, along the décollement the minimum δ37Cl value evolves from −5.5‰ at the most arcward ODP Site 1040/1254, to −3.2‰ at Site 1043/1255, ∼1 km seaward, and to 0‰ at the reference Site 1039/1253. At both subduction zones, the Br/Cl ratios are higher than the seawater value (1.5 × 10− 3) and also show seaward evolutions. These pore fluids originate from greater depth arcward, at ≥ 250 °C, from hydrous mineral formation that preferentially incorporates 37Cl and excludes Br. In contrast, the δ37Cl values in the pore fluids at the Mariana serpentine mud volcanoes are higher than the seawater value (+ 0.3‰ to + 1.8‰); and the Br/Cl ratios are lower. These pore fluid values and the high Cl concentrations with positive δ37Cl values (+ 1.2 to + 6.0‰) in the serpentines, support that the upwelling pore fluid originates from dehydration of the subducting slab that releases water enriched in 37Cl, into the fluid phase. The constancy of the ocean δ37Cl over the past 200 Ma suggests that the isotopically fractionated chlorine in serpentinites and the Cl exchanged in subduction zones are efficiently recycled back into seawater. If the efficiency is b100%, the residual would be transferred to the mantle, with a maximum Cl flux between 2 to 3 × 1017 moles/Ma that would lead to an isotopic difference between the mantle and seawater over the age of the earth on the order of a few per mil. © 2007 Published by Elsevier B.V. Keywords: convergent margins; Cl stable isotopes; Br/Cl ratios; serpentines; marine Cl cycle; pore fluids

1. Introduction

⁎ Corresponding author. E-mail addresses: [email protected] (W. Wei), [email protected] (M. Kastner). 0012-821X/$ - see front matter © 2007 Published by Elsevier B.V. doi:10.1016/j.epsl.2007.11.009

Aqueous fluids play a major role in the physical and chemical evolution of subduction zones. (e.g., (Carson et al., 1990; Davis et al., 1983; Hubbert and Rubey, 1959; Kastner et al., 1991; Spivack et al., 2002). The halogens, fluorine, bromine, chlorine, and especially the

W. Wei et al. / Earth and Planetary Science Letters 266 (2008) 90–104

isotopic composition of chlorine, are excellent tracers of fluid sources, flow paths, and mixing in subduction zones as their geochemical behavior is dominated by strong partitioning into aqueous fluid phases. Chloride is the most abundant anion in seawater and fluids in crust and sediment that are subducted. The natural abundances of the two stable isotopes of chlorine, 35 Cl and 37 Cl, are approximately 76% and 24%, respectively. Hydrous minerals, such as serpentines, talc, chlorite, and amphiboles, which have relatively high concentrations of Cl, become enriched in 37Cl/35Cl upon formation by fluid–rock reactions (Schauble et al., 2003); the residual fluids thus become depleted in Cl and 37 Cl/35Cl. In contrast, dechlorination of these hydrous phases, releases 37Cl/35Cl enriched fluids. Previous studies have shown that marine sedimentary pore fluids in subduction zones have stable Cl isotopic compositions distinct from seawater, with δ37Cl (defined relative to modern seawater) values ranging from − 0.9‰ to − 7.8‰ (e.g. (Ransom et al., 1995; Spivack et al., 2002), indicating significant fractionation of Cl isotopes in their source regions or along their flow paths. The generation of these isotopically light fluids is consistent with chlorination/hydration reactions in igneous oceanic crust resulting in positive δ37Cl values in hydrous minerals (Magenheim et al., 1995; Wei et al., 2005).The apparent reactivity of Cl, manifested by the large natural isotopic variations, − 7.8 to ∼ +8‰ (Magenheim et al., 1995; Ransom et al., 1995), suggests that the other halogens may also be involved in large-scale fluid/mineral exchange. The following discussion consists of a systematic analysis of F, Cl, and Br concentrations, and Cl isotopic ratios of pore fluid samples and solid samples from three subduction zones: the Nankai Trough (Ocean Drilling Program (ODP) Legs 131 and 190), Costa Rica (ODP Legs 170 and 205), and Mariana Forearc (ODP Leg 195), and the geochemical implications in terms of identifying fluid sources, paths, and fluid–rock reactions as well as an analysis of the seawater chlorine isotopic budget.

91

2. Geologic background The Nankai, Costa Rica, and Mariana subduction zones (SZ) were chosen as the study areas because of their contrasting geologic characteristics. The Nankai Trough SZ has a significant accretionary prism whereas at the Costa Rica and Mariana SZ all sediments are underthrust. 2.1. Nankai Trough Most of the sediments are accreted, with only 25– 30% of the section, mainly hemipelagic sediments, underthrusted. The subduction rate is 20–40 mm/yr and the geothermal gradient is high, ∼ 110°C/km, due to the relatively young age (15 Ma) of the subducting plate (Table 1). The pore fluid and sediment samples are from a transect of drill sites across the trench shown in Fig. 1a and b. ODP Site 1173 is in the trench's outer-margin, 10.5 km seaward of the deformation front. Site 1174 is located 1.8 km arcward of the deformation front and 1.6 km east from Site 808, the most arcward site. The most interesting and distinct feature of the pore fluid Cl-depth profiles from Site 1173, through Site 1174, to Site 808 (Fig. 1b), is a broad low-Cl zone situated in the lower Shikoku Basin unit. Other important characteristics are, (1) relative to seawater, pore water in the low-Cl zone freshens systematically; from 8–9% at Site 1173, to 16–17% at Site 1174, and to 20–21% at Site 808, and (2) the vertical structure of the concentration profiles does not support vertical advection from basement. Chlorine stable isotopes are used to help to decipher the origin of the low-Cl fluid along this transect, in particular to determine the relative contributions of a deepsourced fluid transported laterally from the seismogenic zone versus fluid generated by in situ smectite dehydration (e.g. (Henry and Bourlange, 2004; Kastner et al., 1993; Saffer and Bekins, 1998; Underwood et al., 1993).

Table 1 Main Geologic characteristics of the three subduction zones

Nankai Trough Costa Rica Mariana

Fate of incoming sediments

Subduction rate (mm/yr)

Most sediment accreted

20–40

No accretionary prism; all sediments subducted No accretionary prism

Age of subducting plate (Ma)

HF (mW/m2)

Temp Gradient (oC/km)

15

125–130

110

85

24

10–30

∼10

40

N34

13–99

9.2 (HoleE, Site 1200) to 72 (HoleF, Site 1200)

HF = Heat Flow. Data are from ODP Leg 190, 170/205, and 195 Initial Reports.

92

W. Wei et al. / Earth and Planetary Science Letters 266 (2008) 90–104

2.2. Costa Rica The décollement and underthrust sediments were penetrated at Sites 1043 (1255) located ∼ 0.5 km arcward from the trench and at Site 1040 (1254), ∼ 1.1 km arcward from Site 1043 (Fig. 1c). There is no accretionary prism and the underthrust section is lithologically identical to the reference site section, Site 1039 (1253), ∼1.5 km seaward of the trench; it thins from ∼ 380 m at the reference site to ∼ 285 m at Site 1040 (1254). The wedge sediments above the décollement consist mostly of silty clay with volcanic ash. The subduction rate at Costa Rica is ∼85 mm/yr and the geothermal gradient is ∼10°C/km (ODP Leg 170/205 Initial Reports). The presence of a distinct and active fluid flow regime along the décollement is clearly indicated by its chemistry: the fluid has lower than seawater concentrations of Cl, Na, K, and Mg, higher Ca, Li, and Sr concentrations, is enriched in C3–C6 hydrocarbons, and has relatively non-radiogenic 87Sr/86Sr and 7Li/6Li ratios, suggesting that it originates from mineral–fluid reactions at ≥ 150 °C while the temperature at the décollement is ∼ 2–4 °C (Chan and Kastner, 2000; Silver et al., 2000). 2.3. Mariana forearc On the Mariana intra-oceanic SZ, cold (∼ 2 °C) springs with lower than seawater salinity are associated with serpentinite mud volcanoes (Fig. 1d). Carbonate chimneys with brucite, blue amphiboles, phengite, and hydro-garnets are also present. Water released from the downgoing Pacific Plate, venting from a depth of 15– 27 km, hydrates the overlying mantle wedge and converts depleted harzburgite to lower-density serpentinite. The resulting serpentinite mud, containing variably serpentinized harzburgite clasts, ascends buoyantly along fractures and extrudes at the seafloor, where it forms large (30 km in diameter, 2 km high) mud volcanoes along the outer Mariana forearc, in a band that extends from 50 to 120 km behind the trench axis (Fryer et al., 1985, 1995, 2000; Fryer and Mottl, 1992; Mottl, 2003; Mottl et al., 2004). The subduction rate is 40 mm/yr (Table 1). The South Chamorro Seamount, located on the southern Mariana forearc ∼ 85 km arcward from the trench (Fig. 1d), was drilled on ODP Leg 195, Site 1200. It exhibits the second most highly-altered fluid relative to seawater yet detected in the Mariana system (Mottl, 2003). This is the only known site of active blueschist mud volcanism. Hole 1200E is located ∼10 m north of

Hole 1200A (Fig. 1d) where a cold seep was identified. Hole 1200B is ∼20 m east of Holes 1200A and 1200E, and Holes 1200F and 1200D are ∼ 20 and 80 m north of Hole 1200E, respectively, forming a transect northward from the seep at Hole 1200A (Fig. 1d). The deep upwelling fluids in both the South Chamorro Seamount (ODP Leg 195) and Conical Seamount (ODP Leg 125) have distinctly higher than seawater pH. The detailed chemistry is summarized in (Benton, 1997; Benton et al., 2001; Mottl, 2003; Mottl et al., 1992, 2004). 3. Analytical methods 3.1. Halogen concentration analysis Chloride concentrations of pore fluid samples were determined by titration with AgNO3 (0.1% precision based on duplicate analyses of samples and standards). Fluoride and Br concentrations were determined by eluent-suppressed ion chromatography (IC) with a Dionex DX-120 (external precision is ∼3%). Pore fluid fluoride concentrations at Site 1039 were also analyzed by Saether (Kastner et al., 2006), using an ion-selective electrode and ∼ 3.6% precision. Pore fluid bromide concentrations at Site 808 were also analyzed by colorimetric methods (You et al., 1993) and the results obtained by the two methods agree well. 3.2. Column exchange preparation for chlorine isotopic analysis Pore fluids, Cl is the dominant anion and isobaric interference and ionization suppression due to other anions during mass spectrometry is negligible for most pore fluids (Xiao and Zhang, 1992). Therefore, the pore fluid samples were directly converted to CsCl solution using a Cs-form of AG50-X4 resin in a short (∼ 4 cm high) ion exchange column. The pore fluid samples from Mariana that had relatively high alkalinity, were passed through an H+-form resin to neutralize OH-and CO32- which is degassed as CO2, before the Cs-column exchange. Solids, the solution obtained from the chemical purification procedures was first loaded into a short H+-form resin to remove CO32- and OH-. The residual solution that consists primarily of HCl, was directly loaded onto a column with Cs-form cation exchange resin (DOWEX 50w x8, 200-400 mesh). The column was rinsed with Milli Q water until the pH is 7. The yield and total amount of Cl are determined by ion chromatography. The yield was 100 ± 5%, thus eliminating isotope frac-

W. Wei et al. / Earth and Planetary Science Letters 266 (2008) 90–104

tionation during sample preparation (Chan et al., 1992). The solution is evaporated almost to dryness and the dissolved Cl is loaded onto the filament. 3.3. Chlorine isotope analysis Chlorine isotopes were analyzed by thermal ionization mass spectrometer (TIMS) using a VG 336, by the Cs2Cl+ method described in (Magenheim et al., 1994; Numata et al., 2001; Xiao et al., 1995). Daly detection was utilized with a gain of ∼ 1013. Two micrograms of Cl was used per analysis, equivalent to ∼ 0.1 μl seawater.

93

Cl/35Cl ratios are reported as δ37Cl values in per mil (‰) deviations from seawater (‘Standard' Mean Ocean Cl: SMOC), expressed as: h
i 
 d37 Cl ¼ 37 Cl=35 Cl sample 37 Cl=35 Cl smoc 1 37

 1000;

ð1Þ

The standard was filtered seawater from the scientific pier at Scripps Institution of Oceanography, San Diego. The typical in-run precision for a single analysis is 0.15‰ and the external reproducibility (1σ) is 0.3‰ based on 160 standard analyses from 2001 to 2005.

Fig. 1. (a) Seismic reflection profile of the Muroto transect at Nankai Trough, showing locations of Ocean Drilling Program (ODP) Sites 808, 1174 and 1173. Fig is from ODP Leg 190 Initial Reports. (b) Cl concentration (mM)-depth profiles along the Muroto transect at Nankai Trough. Fig is from ODP Leg 190 Initial Reports. (c) Seismic reflection profile of the Costa Rica Transect off shore Nicoya Peninsula, showing locations of ODP Sites 1039/1253, 1043/1255, 1040/1254. The vertical green and red lines indicate the depth of drilling on ODP Legs 170 and 205, respectively; the blue line indicates the décollement. Fig is from ODP Leg 170 and 205 Initial Reports. (d) Location map of seamounts at Marina subduction zone with the pap of drilled holes A, B, D, E, F, and of the spring at ODP Site 1200, drilled on the summit of South Chamorro seamount. Fig is from ODP Leg 195 Initial Reports.

94

W. Wei et al. / Earth and Planetary Science Letters 266 (2008) 90–104

Fig. 1 (continued).

For solid samples, halogens were extracted from silicates by the pyrohydrolysis method (Magenheim et al., 1994). Procedural blanks were established for

Cl for each sample analyzed. The Cl blank is less than 0.1 μg/ml of solution, which is less than 1% of the amount of Cl that was extracted from

W. Wei et al. / Earth and Planetary Science Letters 266 (2008) 90–104

samples with Cl concentrations of N 20 μg (30– 800 ppm). Chlorine was isolated and purified by precipitating AgCl from fluids extracted from solid samples (Numata et al., 2001) re-dissolving the AgCl in ammonia. This purification eliminates ionization suppression from other anions in the mass spectrometer. 4. Results and discussion 4.1. Nankai Trough Most of Site 808 pore fluid δ37Cl data were published in (Ransom et al., 1995). The new pore fluid δ37Cl data for Sites 1173 and 1174, together with newly acquired data for Site 808, provide a 2-D across-trench section through the Nankai Trough Muroto transect (Fig. 1b). This is the first high resolution, chlorine isotope transect for a subduction zone (Fig. 2). Pore fluid δ37Cl are significantly more negative than seawater and exhibit the largest range observed in any environment yet studied (Fig. 2). The δ37Cl profiles also show a systematic evolution along the three sites. In the lower Shikoku Basin Facies, at stratigraphically equivalent depths (based on the lithology) below the décollement at Sites 808 and 1174, or the proto-décollement at

95

the reference Site 1173, δ37Cl values decrease to a minimum of − 7.8‰ at the most arcward Site 808, to − 7.1‰ at Site 1174 and to − 5.8‰ at Site 1173. At all three sites, there is no simple correlation between Cl concentrations and isotopic compositions; δ37Cl is depleted through most of the section while concentrations are only depleted below the turbidite facies section. Two in situ reactions that can lead to low-Cl concentrations are smectite dehydration and/or smectite to illite transformation. The quantity of smectite transformed to illite has been indirectly inferred based on cation exchange capacity (CEC) measurements (Henry and Bourlange, 2004). In the Lower Shikoku Basin facies, CEC is greater at Site 1173 than at the two other sites, 1174 and 808, which have similar CEC values. Based on these observations it was thus argued that the low-Cl concentrations may be primarily due to water release during the smectite to illite transformation with the sediment behaving essentially as a closed system (no advection of pore fluids and minimal diffusion). However, based on pore fluid compositions we consider this unlikely. Illite is a potassium-rich clay while smectite is not. Thus a source of potassium is required for the production of illite from smectite. The principal source of potassium in these sediments is the pore fluid. In this

Fig. 2. δ37Cl (‰)-depth profiles in pore fluids (solid circles) and Cl concentrations of corresponding solids (■) at the Nankai Trough transect. The décollement is situated in the lower Shikoku Basin. Seawater value is by definition 0‰ therefore not shown. The error bars represent external errors.

96

W. Wei et al. / Earth and Planetary Science Letters 266 (2008) 90–104

transect, the maximum pore fluid K concentration is ∼ 12 mM. Based on a stoichiometry of 12 waters per K (mole/mole), the maximum amount of water that can be produced is then given by,

fractionation factor is unrealistic. For example, in a closed sediment system, the fractionation factor between fluid and solid for irreversible Cl uptake is approximately given by,

ðH2 OÞmax =ðH2 OÞinitial ¼ 12  ½K

ai103 ðd  d0 Þlnf

ð2Þ

Thus the maximum dilution due to water associated with the smectite to illite transformation is ∼ 0.25% or ∼1/80 of the observed dilution. The quantity of water released via smectite dehydration can also be estimated based on measured bound water or CEC if it is assumed that all of the ion exchange capacity of sediment is due to smectite. Based on this and assuming that there are 12 water molecules per exchangeable cation in smectite, the dilution is given by: 12  CEC  q  ð1  LÞL

ð3Þ

In which ρ is the density of the solid and ø is porosity. Using the CEC data of Henry and Bourlange, (2004) and a typical porosity of 0.35 in the Lower Shikoku Basin facies a potential dilution of ∼ 25% is calculated. This suggests that dehydration may significantly contribute to the lowering of chlorinity, but will not affect the δ37Cl value, which is consistent with the lack of a systematic relationship between chloride concentration and isotopic composition. Conversely, alteration of ash to authigenic smectite can cause increases in Cl concentration because of the high water content in smectite, but with only minor fractionation in the fluid δ37Cl because of the very low Cl concentration in smectite. Ash alteration will, however, fractionate the oxygen isotopes of the fluid. Typically clays with δ18O of ∼ 27‰ enriched from the fluid will be produced and the fluid will be depleted in δ18O (Savin and Epstein, 1970). The δ18O of the pore fluids at Sites 1174 and 808 are in the range of − 3 to − 5‰ through the entire section, indicating that hydration may also occur (Kastner et al., 1993; Wei et al., 2003). The addition of dissolved Sr with a low 87Sr/86Sr, decreasing from seawater value 0.7092 to 0.7080 (Kastner et al., 1993; Wei et al., 2003), is also observed and is consistent with ash alteration. The negative δ37Cl of the pore fluid can in principle be due to either the advection of an isotopically fractionated fluid generated at greater depth or by in situ reaction. Although in situ fractionation can be due to the release of fractionated Cl from or uptake into solid phases, both processes are very unlikely because the sediments do not contain enough Cl (Fig. 2) to be either a significant source or sink of Cl and the implied

ð4Þ

with f ¼ Clsolid q

ð1  /Þ /

ð5Þ

in which Clsolid is the Cl concentration in the solid, ρ is its density (∼ 2.65 g/cm3) and ø is porosity (∼ 35%). The typical solid Cl concentrations at Sites 808 and 1174 are 30 ppm. This leads to a required fractionation of 800‰ (α = 1.8), an extraordinarily large value. Hence, seaward-evolved δ37Cl vs depth profiles suggest that a deep-sourced fluid is laterally advected from arc to sea through Site 808, 1174, and probably to Site 1173. A common fluid may have advected into both the Shikoku and turbidite sections and then been modified by dehydration and/or hydration reactions, or the fluids Cl concentration may have been modified along its flow path; thus the Cl concentration variations may have been generated before or after the fluid advected into the analyzed sections. Because of the regional high geothermal gradient of ∼11 °C/100 m (Table 1), the dehydration/hydration reactions in the Shikoku and turbidite sections most likely have differed. In order to estimate the contributions of the hydration or dehydration reactions acting as the third fluid source after the two end-member mixing, concentration of a conservative element, that is not involved in low temperature fluid-sediment reactions, is required. Such data, however, are not yet available and should be pursued in future research. Although non-conservative, Br concentrations provide further evidence of an advected fluid, as shown below. Br/Cl ratios higher than the seawater value in the shallow turbidite section are attributed to organic matter diagenesis (Price and Calvert, 1977), where organic matter is more abundant (TOC N 1.0%) (Fig. 1 in the Appendix) (2001). At all these sites, Br concentrations decrease with depth in the Shikoku Basin facies but Br/ Cl ratios do not return to seawater value, neither below the turbidite section nor below the décollement. At the depth of the Cl concentration minimum below the décollement the ratios are 0.2 to 0.4 × 10− 3 higher than the seawater value of 1.5 × 10− 3 (Fig. 1 in the Appendix), with Br concentration of ∼ 800 μM, slightly lower than the seawater value. Organic matter is mostly absent at this depth, at ∼100 °C (2001) thus not affecting Br concentration. Pore fluid dilution by smectite dehydration

W. Wei et al. / Earth and Planetary Science Letters 266 (2008) 90–104

alone does not fractionate Br from Cl to change the Br/Cl ratio. Unlike the smaller Cl, the Br appears to be excluded from the high temperature hydrous minerals, as indicated by Br concentrations in all solid samples from Nankai that are below the detection limit of 5 ppm by I.C. Thus, in addition to fractionating Cl isotopes, the formation of hydrous minerals at elevated temperatures also fractionates Br from Cl, thus increasing the Br/Cl ratios in the fluid, as observed. Fluoride concentrations in the pore fluids remain near the seawater value throughout the turbidite facies, increase dramatically in the upper Shikoku basin pore fluids, and reach maximum values in the lower Shikoku Basin section (Fig. 2 in the Appendix). The maximum F concentration decreases from 3320 μM at Site 808, to 2568 μM at Site 1174, to 2111 μM at Site 1173 as compared to 86 μM in seawater. These values are significantly higher than previously reported values from other ODP sites [e.g. Froelich et al. (1991); Gieskes et al. (2002)]. As the solid phases in both the Turbidite section and the Lower Shikoku contain abundant F, increasing from approximately 450 ppm in the Turbidite section to approximately 1000 ppm in the Lower Shikoku, mobilization from the solids with increasing temperature may cause the increase in pore fluid concentrations. 4.2. Costa Rica SZ At the reference Site 1039/1253, δ37Cl values of pore fluids are within ± 0.5‰ of the seawater value through-

97

out the sediment section (Fig. 3). Arcward across the trench, at Sites 1043/1255 and 1040/1254 (Fig. 1c), δ37Cl values at the fracture zone (FZ), situated at 90 m and 150 m above the décollement at Site 1043/1255 and Site 1040/1254, show minima of − 3.0‰ and − 3.2‰, respectively. Similarly, at the décollement zone of both sites, δ37Cl minimum values are − 3.0‰ at Site 1043/ 1255, and − 5.5‰ at Site1040/1254 (Fig. 3). In both the FZ and décollement, the δ37Cl minimum values are more negative at the more arcward site. This arcward trend of Cl depletion and 35Cl enrichment, together with the anomalies of other geochemical tracers, such as Li and Sr concentration and isotope ratios, indicate vigorous fluid advection within the décollement and fault zone that transports species generated at temperatures of N∼ 150 °C (Chan and Kastner, 2000; Kastner et al., 2006; Silver et al., 2000), at greater depth arcward. The negative δ37Cl of the deep-sourced fluid also implies that as at the Nankai SZ, high temperature hydrous minerals formation (N ∼ 150 °C) sequester Cl and fractionate its isotopes. Below the décollement, in the underthrust sediment section that is lithologically equivalent to that of the reference Site 1039/1253, pore fluid δ37Cl values are close to the seawater value, within ± 0.5‰. The Br/Cl ratios at the FZ and décollement also increase arcward. At the more seaward Site 1043/1255, Br/Cl only shows a slight increase to 2.8 × 10− 3 at the FZ and 1.8 × 10− 3 at the décollement, compared to ∼ 1.5 × 10− 3 in the sediment section above the FZ and below the décollement, whereas at the more arcward

Fig. 3. δ37Cl (‰)-depth profiles in pore fluids at Costa Rica subduction zone transect. Note the different depth scales at the three sites. Seawater value in A, B, and C is denoted by “x”. U1, U2, U3, and T1 indicate lithostratigraphic units. U1: diatom ooze with ash; U2: silty clay with ash; U3: calcareous ooze; T1: silty clay and breccia.

98

W. Wei et al. / Earth and Planetary Science Letters 266 (2008) 90–104

Site 1040/1254, the maximum value is 5.8 × 10− 3 at the FZ and 5.5 × 10− 3 at the décollement (Fig. 3 in the Appendix). Between the FZ and décollement, the approximate constant and higher than seawater Br/Cl value of ∼ 3.8 × 10− 3 may be caused by advection of or diffusion from the décollement and the FZ. Similar to Nankai, the high Br/Cl ratios observed at Costa Rica are consistent with the conclusion that high temperature hydrous mineral formation fractionates Br and Cl. The maximum F concentrations are 1300 μM at the FZ and 1400 μM at the décollement at the most arcward Site 1040/1254 (Fig. 4 in the Appendix). These maxima are lower than at the Nankai SZ (2000–3000 μM), possibly due to the different F concentration in the source solids: 400–500 ppm at Costa Rica and up to 900 ppm at Nankai. Above the FZ and below the décollement, F concentrations are close to that in seawater. In summary, as at the Nankai SZ, a chemically distinct fluid having a high temperature source arcward is laterally injected along the FZ and décollement. This fluid has a lower than seawater Cl concentration, negative δ37Cl, high F concentration and Br/Cl ratio. Because of dilution with in situ pore fluids, the signal becomes weaker seaward along the décollement. 4.3. Mariana mud volcanoes Chloride concentrations in the pore fluids from all holes drilled at South Chamorro Seamount are lower than bottom seawater in this region by up to 7% (ODP Leg 195 Initial Reports). In contrast to the negative δ37Cl values in the pore fluids at Nankai and Costa Rica subduction zones, pore fluid δ37Cl values at ODP Site 1200, serpentinite mud volcano, range between 0.3‰ at 0.9 mbsf, to +1.8‰ at just 70 mbsf in Hole A (Fig. 4), the deepest uncontaminated sample obtained. It is important to note that this maximum drilling depth (70 mbsf) is much shallower than at Nankai (1200 mbsf) or Costa Rica (700 mbsf). The observed positive δ37Cl (up to +1.8‰) suggests that the fluid–solid reactions, which affect Cl isotope ratios at depth are distinct at the Mariana versus at the Nankai and Costa Rica SZ. In the associated serpentinite mud samples from Hole 1200 E, Cl concentrations range from 110 to 200 ppm and the δ37Cl from + 1.0 to + 4.2‰ (Fig. 4 and Table 2). The serpentinized igneous basement samples from Holes 1200 A and B are similar; both are enriched in Cl up to 400–500 ppm, with δ37Cl value of + 1.7‰ (1200 B, Table 2). The serpentinite from Bluemoon Seamount contains 521 ppm Cl and the δ37Cl value is +1.9‰. The δ37Cl values in serpentines from the Mariana seamounts are close to the theoretical estimate of

Fig. 4. δ37Cl (‰)-depth profile in pore fluids and serpentines at the Mariana subduction zone, South Chamorro seamount ODP Site 1200.

δ37Cl of 2–3‰ in silicates formed in equilibrium with seawater (Schauble et al., 2003). The different Cl contents of the serpentines suggest that Cl partition coefficients may also somewhat depend on the particular mineralogy and chemistry of serpentine, One brucite sample separated from a calcite chimney at Conical Seamount has a Cl concentration of 428 ppm and a δ37Cl value of + 8.4‰, the most positive δ37Cl value analyzed so far in any reported Cl-containing mineral (Table 1 in the Appendix). It appears that at high temperatures and pressures when hydrous minerals dehydrate, in addition to water, Cl with positive δ37Cl is also released into the fluid phase. The positive δ37Cl in the pore fluids supports that at the Mariana subduction zone, the origin of the upwelling pore fluids is the dehydration of the subducting serpentinized crust (Mottl et al., 2003, 2004). 4.4. The Br/Cl ratios in the three SZs The pore fluid Br/Cl versus 1/Cl data from all ODP Site 1200 holes trend along a straight line (Fig. 5a), consistent with mixing between two fluids; an ascending fluid and slightly modified bottom seawater. In general, in Hole 1200D that is farthest from the spring, Br/Cl ratios are higher and closer to the seawater value, the upwelling fluid signal is more diluted by mixing with seawater as compared with Holes E, F, and A that are closer to the spring. In Fig. 5a all 1/Cl values are higher than the seawater value because of Cl depletion

W. Wei et al. / Earth and Planetary Science Letters 266 (2008) 90–104

99

Table 2 Fluorine and chlorine concentrations (ppm) and δ37Cl (‰) values in serpentines, analyzed in this study Serpentine

Location

Sample description

F (ppm)

Cl (ppm)

δ37Cl (‰) a

Antigorite

N. California

47

140

2.2

Chrysotile Serpentine muds Site 1200 Hole E Core 3H1, 11.4 mbsf

from S. Africa Marianna

N95% serpentinized Dunite(ol) and Harzburgite (ol + opx) N95% serpentine

0

415

2.3

430

186

3.2

Core 5H1, 17.7 mbsf Core 7H1, 27.0 mbsf Core 10H2, 54.9 mbsf Site 1200 Hole B b, Core 2W2, 33.2 mbsf Serpentine bulk sample Brucite Separate from Calcite Serpentine bulk sample

S. S. S. S.

30 27 78 5

200 146 127 400

1.4 4.2 1.2 1.7

56 244

521 428

1.9 8.4

12–109 (n = 4)

445–745 (n = 4)

− 0.1 to + 1.8 (n = 2)

92–100 (n = 2)

290 (n = 1)

+6.0 (n = 1)

Serpentine powdered

S. Chamorro seamount Chamorro seamount Chamorro seamount Chamorro seamount Chamorro seamount

Bluemoon seamount Conical Seamount Hess Deep (ODP Leg 147 Site 895)

Hess Deep (ODP Leg 147 Site 895)

Serpentinized peridotite with aragonite Serpentinized harzburgite Serpentinized mylonite Serpentinized harzburgite Altered igneous basement rock N95% serpentine Carbonate chimney at Conical Seamount N95% serpentinized from harzburgites and dunites; only washed outside without crushing powdered to fine grains

n = number of samples analyzed. a Error is 0.6‰ (2σ). b Serpentines from Hole 1200A and 1200B are both altered igneous basement rocks; δ37Cl value of 1200 A is not available.

with depth. Below 5 mbsf, the lower than the seawater Br/Cl values indicate the depletion of Br relative to Cl due to dehydration/dechlorination of hydrous minerals

releasing Cl enriched in 37Cl and water, but not Br. In contrast, at the Nankai and Costa Rica SZ, where hydrous mineral formation influences the pore fluid Br

Fig. 5. Br/Cl vs 1/Cl plots of pore fluids at (a) Mariana, (b) Nankai, and (c) Costa Rica subduction zones. In (a), the arrow demonstrates the direction away from the spring. The closer to the spring (Hole F, E, and A), the lower the Br/Cl ratios is. Hole D is the furthest from the spring and has Br/Cl ratios approaching the seawater value. The dotted horizontal line indicates the seawater Br/Cl value and the symbol “x” indicates seawater value for both Br/Cl and 1/Cl. Note that at Mariana most data points are below the seawater line, whereas at Nankai and Costa Rica, the data points are above the seawater line.

100

W. Wei et al. / Earth and Planetary Science Letters 266 (2008) 90–104

and Cl concentrations and the δ37Cl values, the pore fluids have higher than seawater Br/Cl ratios and negative δ37Cl values (Fig. 5b and c). Another interesting observation is that the maxima in Br/Cl ratios at the fracture zone and décollement at the Costa Rica sites 1040 and 1043 are higher than those at Nankai, suggesting a higher organic carbon content in the fluid source region at the Costa Rica SZ than at Nankai. 4.5. In situ versus deep sourced fluid at SZs The depth-profiles of the various geochemical tracers indicate that a deeply sourced fluid is laterally advected seaward at both Nankai and Costa Rica SZ. Pore fluids are a mixture of more deeply sourced and in situ fluids (defined as the original sedimentary fluid modified by local reactions). The relative amount of each of these fluid components is reflected in the δ37Cl: d37 Clm ¼ d37 Clin situ fin situ þ d37 Cldeepsourced fdeepsourced ð6Þ in which δ37Clm is the Cl isotope composition of the analyzed fluid, f is the fraction of the chloride from the in situ fluid and/or of the deep-sourced fluid as indicated by the sub-script label. Diffusion can be ignored if the characteristic length scale for diffusion (d) is less than the width of the low chlorinity anomaly or the width of the gradient zone (d = 2(Dt)1/2), in which d is diffusion distance, D is diffusion coefficient of Cl in sediment, and t is time) (Henry and Bourlange, 2004). It is valid for Nankai subduction zone (d = 250 m with t = 500 kyr and D = 1.4 × 10− 9 m2/s (Henry and Bourlange, 2004)). For the Nankai SZ, the minimum and average δ37Cl increase across the transect from 808 through 1174 to 1173 (Fig. 2). These trends reflect a decreasing fraction of the Cl from the deeply sourced fluid. Assuming the deeply sourced fluid has a uniform isotopic composition, the δ37Cl of the pore fluid is δ37Cldeep-sourcedfdeep-sourced, derived from Eq. (7), as δ37Clin situ is approximately 0‰. Because the isotopic composition of the deep sourced fluid must be equal to or less than the minimum value at Site 808, the relative fractions of Cl from the deep source and original pore fluid can be estimated. Assuming that the minimum δ37Cl value at Nankai Site 808 represents the composition of the deeply sourced fluid, the fraction of the average Cl from this source is 100% at Site 808, ∼ 90% at Site 1174, and ∼ 75% at Site 1173. If the actual deeply sourced fluid has a lower δ37Cl, these values will be lower accordingly. At the Costa Rica SZ the fresher than seawater pore fluid along the décollement and fracture zone is a mix-

ture of fluid advected from the underthrust sediments, as suggested by Saffer and Screaton (2003), with Cl concentration close to the seawater value, and a deeply sourced fluid flowing along the décollement. Because of the difference in the diffusion coefficients of the Cl isotopes (D37Cl/D35Cl = 0.99857) (Richter et al., 2006), they may fractionate by diffusion in the fluid advected from the underthrust sediment (Eggenkamp et al., 1994). Based on a solution of the transient advection-diffusion equation in one dimension for the underthrust section (Crank, 1975; Desaulniers et al., 1986), the δ37Cl value of the in situ fluid should be ∼− 1‰ at both Sites 1040/ 1254 and 1043/1255, assuming no mixing with the deep-sourced fluid. The observed δ37Cl values along the Costa Rica décollement, of − 3.2‰ at Site 1043 to −5.5‰ at Site 1040, indicate that the deep-sourced fluid has δ37Cl that is isotopically depleted as at Nankai, presumably due to fractionation associated with high temperature hydrous mineral formation. The relative proportions of the deeply-sourced and in situ fluids are calculated assuming that the Cl concentration of the deep sourced fluid is equal to the measured value of 480 mM at the décollement at Site 1040. The isotopic data are used to test the consistency of this estimate (Eq. (6)). Based on concentration alone, the contributions at Site 1043 are 60% deeply-sourced and 40% advected from the underthrust sediments. These proportions should result in a fluid that has δ37Cl of − 3.7‰, which is within the uncertainty of the measured value of − 3.2‰ at the décollement at Site 1043. The arc to sea trend of decreasing percentage of the deep sourced fluid along the décollement, (Site 1040 to Site 1043), is also reflected in the Br and F concentrations. 4.6. The importance of serpentinization in the oceanic chlorine and water cycles Serpentine, Mg3Si2O5(OH)4, has three different forms: lizardite, chrysotile, and antigorite (O'Hanley, 1996). It contains ∼ 13 wt.% H2O and hundreds to thousands of ppm of Cl (e.g. (Carlson and Miller, 2003; Ruepke et al., 2002; Seno et al., 2001), hence, they are the most important Cl and H2O solid phase reservoir in the subducting plate and the mantle wedge. During subduction, Cl and water in serpentines are carried into the SZ and at depth are providing a particularly fertile water and Cl source, enriched in 37Cl for magma generated in subduction-related arc volcanoes (Ruepke et al., 2002; Scambelluri et al., 1995; Thompson, 1992; Ulmer and Trommsdorff, 1995). The δ37Cl value of

W. Wei et al. / Earth and Planetary Science Letters 266 (2008) 90–104

+ 1.7‰ of the 747 °C gas condensate recently collected from Nicaragua Momotombo volcano (Shaw et al., 2003) reported in Table 1 in the Appendix probably reflects such a Cl recycling process. In serpentinite, Cl may occur: (1) as a substitute for structural OH−(Sanford, 1981); (2) situated structurally between the basal layers perpendicular to the c-axis (Miura et al., 1981); and (3) as finely-disseminated NaCl along grain boundaries (Sharp and Barnes, 2004). The first involves distinct chlorine isotope fractionations (Schauble et al., 2003) and the latter two will show no isotopic fractionation as they simply represent disseminated NaCl crystals deposited in interstices (Kaufmann et al., 1984). In order to distinguish between these modes of occurrence analytically, each of four individual serpentine bulk samples from Hess Deep was split and the splits were prepared in two distinct ways and analyzed separately: (1) one non-crushed piece was washed and sonicated up to 10 times until no chloride was detected in the wash-solution, followed by routine pyrohydrolysis for Cl extraction. This extracted Cl consists of the structural Cl in the serpentine plus possibly trapped pore fluid Cl that may be present as finelydisseminated NaCl along grain boundaries, as suggested by Sharp and Barnes (2004); (2) The second piece of each serpentine sample was crushed, finely powdered, thoroughly washed and sonicated, up to 10 times, until no chloride was detected in the washing solution, followed by routine pyrohydrolysis for Cl extraction. The residual powdered serpentine should only contain structurally bound Cl. Sharp and Barnes (2004) followed an overall similar sample preparation protocol, except that in their procedure, samples were washed only four times and Cl in the wash solution was not determined. In the wash solutions, the total Cl content was ∼ 0.09 to 0.11 wt% of the bulk samples (n = 4) prepared by procedure (1), and ∼0.28 to 0.45 wt.% in the powdered samples (n = 2) prepared by procedure (2). The δ37Cl was 0 ± 0.5‰ in the wash solutions prepared by both procedures, which indicates that the water-soluble fraction of Cl in serpentines has the seawater δ37Cl value. The δ37Cl values of the pyrohydrolyzed uncrushed samples of procedure (1) ranged from − 0.1 to +1.8‰ and Cl content is 447–745 ppm (Table 2). The δ37Cl value of the crushed samples of procedure (2), is ∼ +6.0‰ and Cl content is 290 ppm, indicating structurally controlled chlorine isotope fractionation. Thus, the lower δ37Cl values of − 0.1 to+ 1.8‰ in the uncrushed serpentines, compared with structural δ37Cl of ∼ +6.0‰, suggests that the uncrushed sample values represent a mixture of soluble NaCl and structurally bound Cl (insoluble Cl). Based on the results from the

101

above procedures (1) and (2), the high Cl concentration of ∼ 0.2 wt% and δ37Cl value of ∼ 0.5‰ in the watersoluble fraction of Cl, reported by Sharp and Barnes (2004), most likely represents a mixture of soluble Cl with seawater δ37Cl and of structural Cl with higher positive δ37Cl values. Using their data on Cl concentrations and δ37Cl in serpentines, Sharp and Barnes (2004) have estimated that serpentines carry 12 × 1012 g/yr of Cl into subduction zones, undoubtedly contributing an important Cl flux in this tectonic environment. They also suggested that the δ37Cl of the ocean has increased with time, assuming a δ37Cl value + 4.7‰ for MORB and a δ37Cl serpentinite equal to the seawater value (0‰). Their estimates, however, did not include: the subduction zone pore fluid refluxes into the ocean with negative δ37Cl values (− 7.8 to −5.5‰), as observed at Nankai and Costa Rica; the positive δ37Cl of hydrothermal fluids (+ 2.4 to + 4.1‰); the efficient recycling of dehydrated slab fluids (+ 1.8‰); and that serpentines have positive δ37Cl values (+ 1.4 to +6.0‰), which suggest that serpentines may be an important 37Cl sink. With the newly acquired data in this study, reported in Table 2, the oceanic Cl and Cl stable isotope cycle is re-examined. 5. The oceanic Cl and Cl isotope cycle Changes in the δ37Cl of seawater over time can be used to evaluate the fluxes of Cl between the mantle and the major exogenous reservoirs, seawater and evaporites. The isotopic balance for seawater is given by
  P
37 37 d d37 Cl sw i Fi d Cli  d Clsw ¼ Isw dt

ð7Þ

Where Fi and δ37Cli are the magnitudes and δ37Cl of the chlorine fluxes into and out of the ocean, δ37Clsw is the δ37Cl of seawater and Isw is the total seawater inventory of Cl. The dominant processes that affect the δ37Clsw are: (1) serpentinization of the oceanic crust; and (2) reactions in subduction zones that can potentially drive changes in the δ37Cl of seawater. For example, if all of the Cl taken up into serpentinites and exchanged in subduction zones were permanently sequestered from the exogenous reservoirs, δ37Clsw could be driven to more negative values with time. Using the Nankai data, this could lead to a change in seawater of about − 7‰ over approximately 200 Ma, using the estimated recycling time of seawater through subduction zones (Kastner et al., 1991).

102

W. Wei et al. / Earth and Planetary Science Letters 266 (2008) 90–104

As the δ37Cl of evaporites is not significantly different than that of seawater, the left-hand-side of Eq. (7), the time derivative of seawater δ37Cl, is contained in the δ37Cl history of evaporites. The δ37Cl of evaporites has been 0 ± 0.5‰ (Eggenkamp and Coleman, 1995) over the past 200 million years. Thus, the derivative in Eq. (8) can be taken as less than 0.0025 per mil/Ma over this time interval. The simplest interpretation of this constancy of δ37Cl, over the past 200 Ma, is that the isotopically fractionated chlorine lost from fluids into serpentinites and Cl exchanged in convergent margins is efficiently recycled back into seawater. This recycling presumably occurs via devolatilization into fluids and volcanic gasses (such as those in the Marianas mud volcanoes) and during arc volcanism, as suggested by the δ37Cl value of +1.7‰ of the 747 °C gas condensate recently collected from Nicaragua Momotombo volcano (Table 1 in the Appendix). While taking the derivative as exactly zero implies a nearly 100% recycling efficiency, the small range in the δ37Cl of evaporites over the past 200 Ma does allow for the possibility of a lower efficiency. If it is assumed that there is no fractionation associated with devolatilization, the maximum flux into the mantle is given by: Fmantle



 d d37 Cl sw Fserp þ Fcmex  ⁎Isw ⁎
¼ dt Fserp d37 Clserp þ Fcmex d37 Clcmex

ð8Þ

Based on this study, we estimate a representative structural Cl content of 400 ppm and δ37Cl value of +1.2 to +6.0‰ in serpentine (Table 2). With a crustal production rate of 6.0 ± 0.8 × 1016 g/yr (Mottl, 2003) and with ∼ 15% being serpentinized (Carlson and Miller, 2003; Peacock, 1990), the Cl flux into serpentines is estimated to be 9.6 × 1016 mol/Ma. To estimate the subduction zone exchange flux (Fcmex), we assume that the Cl concentrations and isotope ratios at Nankai are representative of subduction zones (Kastner et al., 1991; Spivack et al., 2002), and that the entire volume of the ocean is recycled through subduction zones in 200 Ma. This gives a flux of 3.2 × 1018 mol/Ma with δ37Cl of − 5.5 to − 7.8‰. This flux is most likely a maximum value. As the subduction zone flux is approximately 30 times larger than the flux into serpentine, the uncertainty in the serpentine isotopic composition does not strongly affect the calculations and Eq. (8) can be approximated by
 d d37 Cl sw ⁎Isw dt Fmantle c 37 d Clcmex

ð9Þ

which only depends on δ37Clcmex and not the estimated fluxes. Thus the maximum flux into the mantle is estimated to be between 2 to 3 × 1017 moles/Ma. A flux of this magnitude integrated over the age of the earth could lead to an isotopic difference between the mantle and seawater on the order of a few per mil. 6. Conclusions At the Nankai (Muroto transect) and Costa Rica SZ, the high resolution negative δ37Cl and Cl concentration data, as well as the higher than seawater Br/Cl molar ratios in the pore fluids, suggest vigorous arc to sea lateral fluid advection from a deep-source; the fraction of the deeply sourced fluid decreases from arc to sea. At the Mariana SZ serpentine mud volcanoes, the upwelling pore fluids have higher than seawater δ37Cl values, with a maximum of + 1.8‰, and lower than the seawater Br/Cl ratios, manifesting the dehydration reactions at the fluid source that release enriched 37Cl into the fluid phase, but not Br. (8)The constancy of δ37Cl in the ocean over the past 200 Ma, derived from the evaporite data, suggests that the isotopically fractionated chlorine lost from fluids into serpentinites and Cl exchanged in convergent margins is efficiently recycled back into seawater. It does not preclude the possibility that Cl is transferred back to mantle. The maximum Cl flux into the mantle is estimated to be between 2 to 3 × 1017 mol/Ma. Integrated over the age of the earth, it could lead to an isotopic difference on the order of a few per mil between the mantle and seawater. Acknowledgement This research was supported by the National Science Foundation through research grant OCE04-53064 and by the Academic Senate of the University California, San Diego for the Ion Chromatograph. The authors greatly acknowledge Mr. Edward Roggenstein for his analytical assistance with the TIMS at the University of Rhode Island. The authors also thank Dr. William Seyfried and Dr. Kang Ding for providing hydrothermal fluid samples, Dr. Patty Fryer for providing Mariana serpentine samples, and Dr. David Hilton for providing the gas condensate samples from Nicaragua. The authors thank Dr. Barbara Ransom for her help and advice on the Cl isotope project and Gretchen Robinson for the laboratory management at Scripps Institution of Oceanography. We also thank two anonymous reviewers for their constructive comments.

W. Wei et al. / Earth and Planetary Science Letters 266 (2008) 90–104

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j. epsl.2007.11.009. References Benton, L.D., 1997. Origin and evolution of serpentine seamount fluids, Mariana and Izu-Bonin forearcs; implications for the recycling of subducted material. Thesis, University of Tulsa, Tulsa, OK, United States (USA), 209 pp. Benton, L.D., Ryan, J.G., Tera, F., 2001. Boron isotope systematics of slab fluids as inferred from a serpentine seamount, Mariana Forearc. Earth Planet. Sci. Lett. 187 (3–4), 273–282. Carlson, R.L., Miller, D.J., 2003. Mantle wedge water contents estimated from seismic velocities in partially serpentinized peridotites. Geophys. Res. Lett. 30 (5), 4. Carson, B., Suess, E., Strasser, J.C., 1990. Fluid flow and mass flux determinations at vent sites on the Cascadia margin accretionary prism. J. Geophys. Res. B, Solid Earth Planets 95 (6), 8891–8897. Chan, L.H., Kastner, M., 2000. Lithium isotopic compositions of pore fluids and sediments in the Costa Rica subduction zone: implications for fluid processes and sediment contribution to the arc volcanoes. Earth Planet. Sci. Lett. 183 (1–2), 275–290. Chan, L.H., Edmond, J.M., Thompson, G., Gillis, K., 1992. Lithium isotopic composition of submarine basalts: implications for the lithium cycle in the oceans. Earth Planet. Sci. Lett. 108 (1–3), 151–160. Crank, J., 1975. The mathematics of diffusion. Oxford University press. Davis, D., Suppe, J., Dahlen, F.A., 1983. Mechanics of fold-andthrust belts and accretionary wedges. J. Geophys. Res. B 88 (2), 1153–1172. Desaulniers, D.E., Kaufmann, R.S., Cherry, J.A., Bentley, H.W., 1986. (super 37 Cl–(super 35 Cl variations in a diffusion-controlled groundwater system. Geochim. Cosmochim. Acta 50 (8), 1757–1764. Eggenkamp, H.G.M., Coleman, M.L., 1995. d37Cl variations in selected minerals: a possible tool for exploration. J. Geochem. Explor. 55, 249–255. Eggenkamp, H.G.M., Middelburg, J.J., Kreulen, R., 1994. Preferential diffusion of 35Cl relative to 37Cl in sediments of Kau Bay, Halmahera, Indonesia. Chem. Geol. 116 (3–4), 317–325. Froelich, P.N., Mortlock, R.A., Mefferd, M., Powers, J., 1991. Interstitial water chemistry: abyssal South Atlantic and East Georgia Basins, Islas Orcadas and Meteor Rises. Proc. ODP, Sci. Results 114, 719–731. Fryer, P., Mottl, M.J., 1992. Lithology, mineralogy, and origin of serpentine muds recovered from Conical and Torishima forearc seamounts; results of Leg 125 drilling. Proc. Ocean Drill. Program Sci. Res. 125, 343–362. Fryer, P., Ambos, E.L., Hussong, D.M., 1985. Origin and emplacement of Mariana forearc seamounts. Geology 13 (11), 774–777. Fryer, P., Lockwood, J.P., Becker, N., Phipps, S., Todd, C.S., 2000. Significance of serpentine mud volcanism in convergent margins. Spec. Pap. — Geol. Soc. Am. 349, 35–51. Fryer, P., Mottl, M., Johnson, L., Haggerty, J., Phipps, S., Maekawa, H., 1995. Serpentine bodies in the forearcs of western Pacific con-

103

vergent margins; origin and associated fluids. Geophys. Monogr. 88, 239–279. Gieskes, J.M., Simoneit, B.R.T., Goodfellow, W.D., Baker, P.A., Mahn, C., 2002. Hydrothermal geochemistry of sediments and pore waters in Escanaba Trough—ODP Leg 169. Appl. Geochemi. 17 (11), 1435–1456. Henry, P., Bourlange, S., 2004. Smectite and fluid budget at Nankai ODP sites derived from cation exchange capacity. Earth Planet. Sci. Lett. 219 (1–2), 129–145. Hubbert, M.K., Rubey, W.W., 1959. Mechanics of fluid-filled porous solids and its application to overthrust faulting, [Part] 1 of Role of fluid pressure in mechanics of overthrust faulting. Geol. Soc. Amer. Bull. 70 (2), 115–166. Kastner, M., Elderfield, H., Jenkins, W.J., Gieskes, J.M., Gamo, T., 1993. Geochemical and isotopic evidence for fluid flow in the western Nankai subduction zone, Japan. Proc. Ocean Drill. Program Sci. Results 131, 397–413. Kastner, M., Elderfield, H., Martin, J.B., 1991. Fluids in convergent margins; what do we know about their composition, origin, role in diagenesis and importance for oceanic chemical fluxes? Philos. Trans. R. Soc. Lond. Ser. A: Math. Phys. Sci. 335 (1638), 243–259. Kastner, M., Solomon, E., Wei, W., Chan, L.-H., Saether, O.M., 2006. Data report: Chemical and isotopic compositions of pore fluids and sediments from across the Middle America Trench, offshore Costa Rica. In: Morris, J.D., Villinger, H.W., Klaus, A. (Eds.), Proc. ODP, Sci. Results, vol. 205. Kaufmann, R., Long, A., Bentley, H., Davis, S., 1984. Natural chlorine isotope variations. Nature 309 (5966), 338–340. Magenheim, A.J., Spivack, A.J., Michael, P.J., Gieskes, J.M., 1995. Chlorine stable isotope composition of the oceanic crust; implications for Earth's distribution of chlorine. Earth Planet. Sci. Lett. 131 (3–4), 427–432. Magenheim, A.J., Spivack, A.J., Volpe, C., Ransom, B., 1994. Precise determination of stable chlorine isotopic ratios in low-concentration natural samples. Geochim. Cosmochim. Acta 58 (14), 3117–3121. Miura, Y., Rucklidge, J., Nord, G.L., 1981. The occurrence of chlorine in serpentine minerals. Contrib. Mineral. Petrol. 76 (1), 17–23. Mottl, M.J., 2003. Partitioning of energy and mass fluxes between mid-ocean ridge axes and flanks at high and low temperature. In: Halbach, P.E., Tunnicliffe, V., Hein, J.R. (Eds.), Energy and Mass Transfer in Marine Hydrothermal Systems. Dahlem University Press, Berlin, pp. 271–286. Mottl, M.J., Fryer, P., Pearce, J.A., Stokking, L.B., Ali, J.R., al, e., 1992. Pore waters from serpentinite seamounts in the Mariana and Izu–Bonin forearcs, Leg 125; evidence for volatiles from the subducting slab. Proc. Ocean Drill. Program Sci. Results 125, 373–385. Mottl, M.J., Komor, S.C., Fryer, P.C., Moyer, L., 2003. Deep-slab fluids fuel extremophilic Archaea on a Mariana forearc serpentinite mud volcano: Ocean Drilling Program Leg 195. Geochem. Geophys. Geosyst. 4 (11). doi:10.1029/2003GC000588. Mottl, M.J., Wheat, C.G., Fryer, P., Gharib, J., Martin, J.B., 2004. Chemistry of springs across the Mariana forearc shows progressive devolatilization of the subducting plate. Geochimi. Cosmochim. Acta 68 (23), 4915–4933. Numata, M., Nakamura, N., Gamo, T., 2001. Precise measurement of chlorine stable isotopic ratios by thermal ionization mass spectrometry. Geochem. J. 35 (2), 89–100. O'Hanley, D.S., 1996. Serpentinites: records of tectonic and petrological history.

104

W. Wei et al. / Earth and Planetary Science Letters 266 (2008) 90–104

Peacock, S.M., 1990. Fluid processes in subduction zones. Science 248 (4953), 329–337. Price, N.B., Calvert, S.E., 1977. The contrasting geochemical behaviours of iodine and bromine in recent sediments from the Namibian shelf. Geochim. Cosmochim. Acta 41 (12), 1769–1775. Ransom, B., Spivack, A.J., Kastner, M., 1995. Stable Cl isotopes in subduction-zone pore waters; implications for fluid–rock reactions and the cycling of chlorine. Geology 23 (8), 715–718. Richter, F.M., Mendybaev, R.A., Christensen, J.N., Hutcheon, I.D., Williams, R.W., Sturchio, N.C., Beloso, J.A.D., 2006. Kinetic isotopic fractionation during diffusion of ionic species in water. Geochim. Cosmochim. Acta 70 (2), 277–289. Ruepke, L.H., Morgan, J.P., Hort, M., 2002. Does slab serpentinization and deserpentinization create the primary HIMU mantle component? Geochim. Cosmochim. Acta 66 (15A), 656. Saffer, D.M., Bekins, B.A., 1998. Episodic fluid flow in the Nankai accretionary complex; timescale, geochemistry, flow rates, and fluid budget. J. Geophys. Res. 103 (B12), 30,351–30,370. Saffer, D.M., Screaton, E.J., 2003. Fluid flow at the toe of convergent margins; interpretation of sharp pore-water geochemical gradients. Earth Planet. Sci. Lett. 213 (3–4), 261–270. Sanford, R.F., 1981. Mineralogical and chemical effects of hydration reactions and applications to serpentinization. Am. Mineral. 66 (3–4), 290–297. Savin, S.M., Epstein, S., 1970. The oxygen and hydrogen isotope geochemistry of clay minerals. Geochim. Cosmochim. Acta 34 (1), 25–42. Scambelluri, M., Muentener, O., Hermann, J., Piccardo, G.B., Trommsdorff, V., 1995. Subduction of water into the mantle; history of an Alpine peridotite. Geology 23 (5), 459–462. Schauble, E.A., Rossman, G.R., Taylor, J.H.P., 2003. Theoretical estimates of equilibrium chlorine-isotope fractionations. Geochim. Cosmochim. Acta 67 (17), 3267–3281. Seno, T.D.Z., Kobayashi, Y., Nakamura, M., 2001. Dehydration of serpentinized mantle: Seismic evidence from southwest Japan. Earth Planets Space 53, 861–871. Sharp, Z.D., Barnes, J.D., 2004. Water-soluble chlorides in massive sea-floor serpentinites; a source of chloride in subduction zones. Earth Planet. Sci. Lett. 226 (1–2), 243–254.

Shaw, A.M., Hilton, D.R., Fischer, T.P., Walker, J.A., Alvarado, G.E., 2003. Contrasting He–C relationships in Nicaragua and Costa Rica: insights into C cycling through subduction zones. Earth Planet. Sci. Lett. 214 (3–4), 499–513. Silver, E., Kastner, M., Fisher, A., Morris, J., McIntosh, K., Saffer, D., 2000. Fluid flow paths in the Middle America Trench and Costa Rica margin. Geology 28 (8), 679–682. Spivack, A.J., Kastner, M., Ransom, B., 2002. Elemental and isotopic chloride geochemistry and fluid flow in the Nankai Trough. Geophys. Res. Lett. 29 (14). Thompson, A.B., 1992. Water in the Earth's upper mantle. Nature 358 (6384), 295–302. Ulmer, P., Trommsdorff, V., 1995. Serpentine stability to mantle depths and subduction-related magmatism. Science 268 (5212), 858–861. Underwood, M.B., Pickering, K., Gieskes, J.M., Kastner, M., Orr, R., 1993. Sediment geochemistry, clay mineralogy, and diagenesis; a synthesis of data from Leg 131, Nankai Trough. Proc. Ocean Drill. Program Sci. Results 131, 343–363. Wei, W., Kastner, M., Deyhle, A., Spivack, A.J., 2005. Geochemical cycling of fluorine, chlorine, bromine, and boron and implications for fluid–rock reactions in Mariana forearc, South Chamorro Seamount. In: Shinohara, M., Salisbury, M.H., Richter, C. (Eds.), Proc. ODP, Sci. Results, vol. 195. Wei, W., Kastner, M., Spivack, A.J., 2003. Halogen concentrations and stable isotopes (O, Sr, and Cl) in the Nankai Muroto Transect and their implication for fluid-sediment interactions and fluid flow. Eos Trans. AGU, Fall Meeting Suppl. Abstract 84 (1459). Xiao, Y.K., Zhang, C.G., 1992. High-precision isotopic measurement of chlorine by thermal ionization mass-spectrometry of the Cs2Cl+ Ion. Int. J. Mass Spectrom. Ion Proces. 116 (3), 183–192. Xiao, Y.K., Zhou, Y.M., Liu, W.G., 1995. Precise measurement of chlorine isotopes based on Cs2Cl+ by thermal ionization mass spectrometry. Anal. Lett. 28 (7), 1295–1304. You, C.F., Gieskes, J.M., Chen, R.F., Spivack, A., Gamo, T., 1993. Iodide, bromide, manganese, boron, and dissolved organic carbon in interstitial waters of organic carbon-rich marine sediments; observations in the Nankai accretionary prism. Proc. Ocean Drill. Program Sci. Results 131, 165–174.