Behavior of Fe and S during serpentinite dehydration
Subduction zones are main global element cycling systems. The main element-transport agent in subduction zones is widely assumed to be an aqueous fluid released due to dehydration of oceanic lithosphere. The mobility and compatibility of fluid mobile elements strongly depends on P-T conditions and oxygen fugacity (fO2) of the zone where H2O is released. While pressure and temperature conditions of various depths in the slab and along the slab are well established, information about fO2 of subduction zones is hardly provided. The fO2 of subduction zones and upper mantle in general may vary by 6 orders of magnitude (Frost and McCammon, 2008). The understanding of fO2 conditions in mantle and in subduction zones in particular includes investigation of slab lithologies and their relations between each other and a mantle wedge. The fact that mantle wedge peridotites are more oxidized than abyssal peridotites brings compelling evidence that oxidized subducted slab influences the redox state of the overlying mantle. Composition of the slab and speciation of elements capable for electron exchange, Fe and S in particular, are the principal parameters which are controlled by fO2 in subduction zones. The evolution of Fe3+/Fetotal ratio and S speciation in serpentinites may provide invaluable information about fO2 in subduction zones.
In this chapter the behavior of Fe and S, main redox sensitive elements in serpentinites, and the redox capacity of different in composition serpentinites are investigated. Accurate X-ray absorption spectroscopy measurements, applied in this study, allows to quantify Fe3+/Fetotal in all ex-situ and in-situ experimental products. The discussion about oxidizing properties of released fluids is provided. In addition, the influence of neighboring lithologies on reduction/oxidation of serpentinites and the role of serpentinites in mantle wedge oxidation is discussed.
Sulfur may be not only oxidized/reduced during serpentinite dehydration, but also can be volatilized. The speciation of volatile sulfur and understanding the mechanism of S release from subducting slab is important regarding the ability of sulfur complexes transport economically important elements to the mantle wedge and continental crust. Experimental results of this chapter demonstrate the speciation and loss of S during serpentinite dehydration and allow to speculate about analogous processes in natural settings.
Dehydration of serpentinites produces a large amount of H2O at elevated P-T conditions relevant for subduction settings. Our results identified that this aqueous fluid has a different redox capacity at intermediate and high P-T conditions due to different degree of Fe and S reduction.
The progress of the redox reactions during serpentinite dehydration was constrained by experimental approach. Experiments were performed in piston-cylinder apparatus. Three serpentinites compositions varied in terms of bulk Fe, Fe3+ and S contents were used. Intrinsic oxygen fugacity (fO2) of the experimental setup was stable at fO2 ~QFM-2, close to conditions of subduction settings. Experimental results demonstrated decrease of an initial high bulk Fe3+/Fetotal ratios (0.9) of serpentinites down to ~0.2 in anhydrous high temperature assemblages through magnetite and Fe3+-bearing antigorite breakdown. Serpentinites without magnetite are shown to drastically reduce their Fe3+/Fetotal ratio at 650°C. Magnetite is occurred to be crucial for production of highly oxidized fluids and volatile sulfur species which can be transported from the subducting slab to the mantle wedge. The presence of pyrite, which transforms to pyrrhotite below 450°C, imposes a release of ¼ of initial sulfur, most-likely in H2S form.
The experimental approach and detailed characterization of mineral assemblages and bulk Fe and S speciation allowed to estimate quantitatively the amount of released H2O and redox potential of the released fluid at different steps of dehydration process. The extrapolation of experimental results to natural settings was accomplished with the help of thermodynamic modeling. We suggested that the presence of reduced species (e.g., sulfur and carbon) in subducting slab may drive elemental reduction in serpentinites. The demonstrated great oxidizing power of serpentinites could explain why mantle wedge peridotites and arc magmas are more oxidized compared to the upper mantle in other geological settings.
Subduction of sediments, metabasalts and serpentinites comprising oceanic lithosphere is accompanied by a release of aqueous fluid, which is widely known to be a cause of partial melting in the mantle wedge, arc magma formation and generation of earthquakes. Dehydration of oceanic subducting slab is also an important process regarding global geochemical cycling of elements. It is well established that fore-arc mantle rocks and arc magmas inherit part of their geochemical composition from oceanic slab via the release of aqueous fluids from dehydrating subducting lithologies (e.g., Plank and Langmuir, 1998; Kerrick and Connoly, 2001a; Kessel et al., 2005). Each of the slab lithological components, i.e. sediments, metabasalts or serpentinites, releases fluids to a different extent and in different ranges of subduction depths. Sediments and metabasalts which compose the upper layers of the slab, dehydrate at depths of about 50-80 km (e.g., Syracuse et al., 2010) and mainly contribute to the hydration of fore-arc mantle, before the front of arc-magmas generation (Kerrick and Connolly, 2001a; Evans, 2012). Serpentinites, in their turn, dehydrate at greater depths, due to the successive breakdown of antigorite, the main high-pressure and high-temperature serpentine variety (Ulmer and Trommsdorff, 1995), and clinochlore (Bromiley and Pawley, 2003). Contribution of serpentinites to the slab water budget is important since serpentinites contain 8 wt.% of H2O in average (Hacker, 2003) whereas metabasalts and sediments host about 3 wt.% (Kerrick and Connolly, 2001b) and 5 wt.% (Kerrick and Connolly, 2001a), respectively.
The dehydration of serpentinites causes the mobilization of FME (fluid-mobile elements) (Hattori and Guillot, 2007; Evans, 2012) potentially in their oxidized form (e.g., Evans, 2012; Debret et al., 2016). Release and migration of slab-derived fluids is potentially a way to modify the redox conditions occurring in the mantle wedge. Indeed, arc magmas were shown to be more oxidized than MORB and OIB (Parkinson and Arculus, 1999; Kelley and Cottrell, 2008). In addition, serpentinites forming in a mantle wedge were shown to be oxidized and enriched in FME (e.g., Hattori and Guillot, 2007) which could have been transported by aqueous fluids from the slab serpentinites to the mantle wedge.
The oxidizing/reducing capacity of a rock is determined by the presence of elements capable for electron exchange, such as Fe and S (Evans, 2012). Oceanic serpentinites are mainly produced by the hydrothermal alteration of the oceanic lithosphere which involves redox reactions. Globally, ferrous iron from oceanic mantle peridotites oxidizes while the alteration fluid is reduced (Evans, 2013). At slow spreading centers for example, olivine and pyroxene from abyssal peridotites are transformed into ferric serpentine (lizardite) and magnetite while H2-bearing fluids are produced (Marcaillou et al., 2011). It can be anticipated that the process of dehydration of serpentinized peridotites will conversely produce oxidized fluids while ferric iron in serpentine and magnetite reduces back to ferrous in olivine-pyroxene assemblages. Actually, the Fe3+/Fetot ratio of natural serpentinites is shown to decrease during serpentinite dehydration in a fossilized subduction setting from western Alps (Debret et al., 2014, 2015). As it was demonstrated by Debret et al. (2014), modal abundance of magnetite, together with bulk ferric iron, decreases progressively during serpentine breakdown. Consequently, we must consider iron reduction (Fe3+ Fe2+) as a half-reaction of oxidation with the potential of oxidizing the serpentinite environment.
Sulfur is another element capable for electron exchange in subducted rocks. A number of natural and thermodynamic modeling studies demonstrated a possible sulfur loss from deep parts of the slab. Release of sulfur is often proposed to be associated with redox reactions producing volatile (e.g., H2S) or soluble sulfur-bearing species (e.g., SO42-), as the reduction of sulfates to sulfides (Evans, 2014), the reduction of pyrite to pyrrhotite (Hall, 1986; Evans et al., 2015) or the oxidation of sulfides to sulfates (Debret et al., 2016). In serpentinites, sulfur is mostly hosted by sulfides (reference). The stability of sulfides at P-T-fO2 conditions relevant to subduction zones is however mostly unknown and, hence, the observed transfer of sulfur from sulfide to S-bearing volatile species is still poorly understood (see Frost, 1985 for the sulfur behavior at shallow depths in subduction zones).
Serpentinites composition with respect to these two elements with variable oxidation states, iron and sulfur, is somehow variable. Total iron content in serpentinites can range from ~4 wt.% (Guillot et al., 2001) to ~12 wt.% (Li et al., 2004). A range of Fe3+/Fetotal in serpentinites is found to be about 0.7-0.8 (Evans, 2008; Debret et al. 2014; 2015). Sulfur content in serpentinites changes from ~0.03 to ~1 wt.% (Alt et al., 2013). Consequently, different oceanic serpentinites may produce different types of fluids with potentially different redox potentials.
Based on our recent experimental results, reduction of iron in serpentinites begins at low temperatures, when magnetite destabilizes (Merkulova et al., 2016). Iron reduction proceeds up to 700°C, when Fe3+-antigorite breakdown is completed. Sulfur reduction is likely to occur in narrower and lower temperature range compared to Fe (Frost, 1985; Hall, 1986).
Considering the slab being an open or closed system it may have major implications regarding processes of slab dehydration. According to several authors (e.g., Marchesi et al., 2013; Evans et al., 2014), the open system should be favored, meaning that the formed fluids leave the system, and do not impact the forthcoming mineral reactions, especially in terms of redox conditions. Based on this assumption, released fluids potentially escape to upper parts of the slab or ascent to the mantle wedge.
Hence, possible liberation of oxygen, due to potential Fe and S reduction in the open-system conditions of subduction, leads to the oxidation of upper layers of the slab and the mantle wedge, provides oxidized conditions for arc magma formation, and affects mobility of FME. However, no experimental work has been done in order to estimate quantitatively the effect of transition elements reduction on the release of oxidizing fluids.
The goal of the present work is to examine the redox behavior of iron and sulfur during the dehydration of simplified serpentinite compositions at the P-T conditions of subduction zones and fO2 ~QFM. Our approach is to combine experimental results and thermochemical modeling in order to evaluate the reducing/oxidizing potential of the aqueous fluid released coincidentally. Three different starting compositions were chosen for experiments in the piston-cylinder apparatus (PC): 1) Fe3+-bearing antigorite; 2) mixture of Fe3+-antigorite and magnetite; 3) mixture of Fe3+-antigorite and pyrite. Such a choice of starting materials allows to investigate the effect of magnetite and sulfide on the phase relationships of antigorite as a function of temperature at 2 GPa. We show that fO2 conditions in our experiments are controlled by the PC cell assembly and, thus, thermodynamic modeling is used to extrapolate the experimental results to a range of naturally relevant redox conditions. Experimental products were characterized by X-ray diffraction, electron microprobe and X-ray absorption near edge structure (XANES) spectroscopy at both the Fe and S K-edges.
Three starting materials containing antigorite as major component were used: (1) 100 wt.% natural antigorite; (2) 95 wt.% natural antigorite + 5 wt.% magnetite (Magnox powder, 99% purity); (3) 97 wt.% natural antigorite + 3 wt.% natural pyrite. Experiments performed with these starting materials will be referred to as “set#1”, “set#2” or “set#3”, respectively. The same natural antigorite was used in all three starting mixtures, antigorite was collected at the Mont-Cenis massif, French Alps (Muñoz et al., 2013). Muñoz et al. (2013) determined a Fe3+/Fetotal of 0.97 ratio for this antigorite sample which is relevant to the Fe3+/Fetotal ratio encountered for lizardite in general and for lizardite from serpentinized peridotite in particular (e.g., Evans et al., 2012; Debret et al., 2014). Furthermore, this antigorite sample was selected because of its homogeneity and availability in sufficient quantity for experimental petrology. The relatively high amount of pyrite in set#3, which is 3 times higher than in natural serpentinites (Delacour et al., 2008), is aimed at performing sulfur elemental analysis and XANES measurements at S K-edge on the run products with sufficient resolution.
The major-element composition of starting materials was measured by X-ray fluorescence using (XRF) EDAX Eagle III spectrometer (ISTerre, Grenoble) and mineral abundance was determined using X-ray powder diffraction (XRPD). Minor amounts of andradite and clinochlore were observed in all starting mixtures, which account for the presence of CaO and Al2O3 (2.43 wt.% and 1.92 wt.%, respectively) in the starting antigorite. Bulk oxide compositions of the three starting materials are shown in Table 3.2-2 together with Fe3+/Fetotal ratios obtained from XANES spectroscopy measurements at Fe K-edge.
Experiments were carried out in an end-loaded piston-cylinder apparatus at pressures from 1.5 to 2 GPa and temperatures from 450 to 900°C, with experimental durations ranging from 4.5 to 9 days. The detailed experimental procedure is described in Merkulova et al. (2016). The oxygen fugacity (fO2) was imposed by the cell assembly and no fO2 mineral buffer was added. The permeability of gold capsule to hydrogen, produced by the graphite furnace oxidation (Chou., 1986), imposes, at least transiently, a fO2 between QFM and QFM-2 (e.g., Truckenbrodt et al., 1997). The capacity of our cell assembly to produce H2 has been tested (see hereafter).
Table 3.2-1 summarizes the P-T conditions, durations, starting material and mineral assemblages observed in run products for three experimental sets.
After each run, the recovered capsule was pierced to check whether the water produced during mineral reactions, if any, had not escaped in the course of the experiment. “Rotten egg” smell was sensed after capsules of set#3 experiments were pierced, characterizing the presence of H2S gas.
One third of the recovered powder sample was crushed in an agate mortar for XRPD characterization. Another third of the sample was mounted in epoxy and polished for electron probe micro-analysis (EPMA) and scanning electron microscope (SEM) observations. The remaining part of the sample was finely crushed and prepared as a 5 mm diameter pellet for XANES spectroscopy measurements. Samples containing sulfide (set#2) were crushed in a glove bag filled with He/Ar gas to prevent oxidation of sulfides.
The methods of X-ray diffraction characterization of experimental products and electron probe micro-analysis (EPMA) of individual mineral phases performed at ISTerre (Grenoble, France) are identical to those described in Merkulova et al. (2016).
The phase abundances in each experimental product were estimated from mass balance calculations using the average individual mineral compositions measured by EPMA (Table 3.2-S1) and the bulk composition of starting materials (Table 3.2-2). Modal abundances were derived following the method described by Lamberg et al. (1997). Estimated deviations from average values are ± 1-20 %.
XANES analyses were performed at Phoenix beamline of SLS (Swiss Light Source) synchrotron and at LUCIA beamline of SOLEIL synchrotron (France). Experimental products from set#2 were analyzed at the Fe K-edge at the Phoenix beamline. Incident energy was scanned from 7050 eV to 7290 eV using Si(111) double-crystal monochromator. The spot size on the sample was 1.4 x 1.4 mm. XANES spectra were collected in fluorescence mode using a 1-element silicon-drifted diode (SDD) located at 90° from the incident X-ray beam direction.
XANES spectra of experimental products from set#1 and set#3 were collected at LUCIA beamline. Fe K-edge XANES spectra were acquired between 7050 and 7600 eV, using a Si(311) fixed-exit double-crystal monochromator, and the S K-edge XANES spectra were acquired between 2400 and 2600 eV using the same monochromator equipped with KTP(011) crystals. The beam size on the sample was ~1 mm in diameter. In order to prevent beam damage and possible changes in valence for iron and sulfur (i.e., photo-oxidation/reduction) during measurements, XANES spectra were collected in a primary vacuum chamber cooled at -70°C using a liquid nitrogen cryostat. Spectra were collected in fluorescence mode using 4-elements silicon drifted diode (SDD) located at 86° from the incident beam.
XANES spectra were normalized using the Athena software (Ravel and Newville, 2005). Quantitative bulk rock Fe3+/Fetot was obtained from Fe-K pre-edge analysis, which includes baseline subtraction and fitting the pre-peak region of the spectra using a Matlab routine from the XasMap package (Muñoz at al., 2006; 2008). The Fe3+/Fetot quantification was based on the pre-edge centroid energy calibration (see Wilke et al., 2001) using the following model phases: olivine, staurolite, andradite and sanidine (Muñoz et al., 2013).
The determination of sulfur concentration in starting material and 3 experimental products of the set#3 was performed at G.G. Hatch Lab (Ottawa, Canada). Detailed procedure of elemental analysis can be found in Operating Instructions vario EL (2005) and in supplementary material of this article.
Table of contents :
I.1 Composition of subducting slab H2O content
Composition of slab serpentinites
I.2 Serpentinite dehydration in subduction zones and stability of hydrous phases
I.3 Composition of subduction fluid
I.4 Redox state of subduction zones
II Open questions and strategy
III Outline of the thesis
Chapter 1 Experimental techniques and analytical methods
1.1. Experimental techniques
1.1.1. High pressure experimental techniques.
1.1.2. Piston-cylinder experiments
1.1.3. Diamond anvil cell experiments
1.2. Analytical methods
1.2.1 X-ray Powder Diffraction (XRPD)
1.2.2 Scanning electron microscope
1.2.3 Electron probe microanalysis
1.2.4 Elemental and isotope ratio analysis of sulfur
1.2.5 X-ray absorption spectroscopy (XAS)
Chapter 2 Role of iron content on serpentinite dehydration depth in subduction zones: experiment and thermodynamic modeling
Mineral reactions during dehydration
Antigorite upper thermal stability limit
Continuous water release
Implications for subduction zone seismicity
Chapter 3 Behavior of Fe and S during serpentinite dehydration
3.2 Consequences of Fe and S reduction during serpentinite dehydration: experimental study
3.3. Reduction of Fe during serpentine dehydration: in-situ observations
3.3.1 HDAC experiments
3.3.2 In-situ XANES and XRD experiments on lizardite-antigorite transition
3.4. Sulfur mobility during serpentinite dehydration
3.4.2 Results and discussion
Chapter 4 Thermodynamic modeling
4.3. Modeling CFMASH and CFMASH-Sulfur systems relative to experiments
CFMASH + Sulfur system
4.5 Application of thermodynamics to natural systems modeling
Mantle wedge peridotite metamorphism above subduction zone: hydration in lithospheric mantle.
Conclusions and perspectives