PRODUCTION D’H2 IN SITU
The high concentration of Ca measured in fluid vent at the Rainbow hydrothermal field (66 mM) is consistent with the alteration of a lherzolite-type peridotite (Charlou et al., 2002). For that reason, this experimental study was conducted using a non-altered spinel-lherzolite composition as starting material. The peridotite was sampled at Borée (Massif Central, France), where it occurs as xenoliths in a basaltic flow. This granular peridotite, the so-called G-type peridotite, was investigated in several previous studies (e.g., Leggo and Hutchison, 1968; Mercier and Nicolas, 1975; Xu et al., 1998) and is composed of about 65 vol% olivine, 30 vol% pyroxene and 5 vol% spinel. The mineral phases were individually analyzed by XRF spectroscopy. The inferred compositions are in agreement with the data published by Xu et al. (1998). The bulk analysis of the peridotite was performed by ICP-AES. The chemical analyses of the bulk and the individual minerals are both referenced in Supplementary Table. Peridotite fragments were ground in agate mortars to fine particles (~1 μm in size). The experiments were run in a 0.25 L hastelloy PARR autoclave with a w/r of 3/2, i.e. 180 mL distilled water and 120 g of peridotite. The experimental pressure and temperature conditions were defined according to the thermodynamic simulations of Klein et al. (2009) and McCollom and Bach (2009), which predict the most efficient production of hydrogen at 30 MPa and around 300 °C. The setting pressure was regulated at the beginning of the run and after each gas sampling during the run by pumping neutral gas (Argon) into the autoclave, and it was measured with a pressure gauge. The temperature was measured by two thermocouples located at the top and 4 cm above the bottom of the autoclave. The operating temperature was controlled by the thermocouple to ±1 °C. The cumulative uncertainties on temperature and pressure are about 3 °C and 5 bar, respectively. Four experimental runs, of 7, 18, 34 and 70 days duration were performed. At the end of each run, the sample was quenching from 300 °C to 30 °C in 10 min with a flow of compressed air.
Fluid sampling and analysis
Gas phases were sampled in-situ using a regulator system coupled with a 5 cm 141 3 hastelloy tube. The hastelloy tube was first quenched in a 15 °C water for 3 min before the gas was injected in the chromatograph (Varian) at 3 bar. Then, the composition was measured over 10 min using two thermal conductivity detectors (TCD). The absolute uncertainty of these detectors for the measurement of H2 gas concentration is about 1.25%. For each run, four analyses were performed to check reproducibility. The ―Galaxie Chromatograph‖ software provides peak processing and gas fraction calculations.
The liquid phases were diluted in 2% HNO3 and then analyzed by inductively-coupled plasma atomic emission spectroscopy (ICP-AES) at the Laboratoire de Géodynamique des Chaînes Alpines (LGCA; Grenoble; France). Results are listed in Table A1.2.3. Mineralogical characterization The solid phases recovered after the experiments were dehydrated at 50 °C for 3 h. They were ground in an agate mortar to obtain micrometric grain size and compressed to 4 mm diameter pellets for mineralogical characterization. Each pellet was then characterized by X-ray diffraction (XRD), Raman and Fe K-edge XANES spectroscopy.
XRD analyses have been performed with a Siemens D501 diffractometer at the LGCA. A large X-ray beam was emitted by a cobalt source with a 40 kV accelerating voltage and a 37.5 mA current.
Raman measurements were performed with an InVia Renishaw Raman spectrometer using a green laser light at 514 cm−1 with a beam size of 20 μm on the sample in Laboratoire d’Electrochimie et de Physicochimie des Matériaux et des Interfaces (LEPMI; Grenoble; France). The spectra were recorded between 3600 and 3750 cm−1 ; area that corresponds to the structural OH group vibrations. At minimum, 3 acquisitions per sample were done at EXPÉRIENCES ET RÉSULTATS 91 different locations in order to check reproducibility and sample homogeneity. The spectra were then averaged to obtain better statistics. 2.3.3. X-ray absorption spectroscopy : XANES spectra were collected at the Swiss Light Source (SLS), on the ―super bending magnet‖ (i.e., 2.9 T) beamline, SuperXAS. The storage ring operating conditions were 2.4 GeV electron energy and 400 mA electron current in the ―Top-up‖ injection mode. A Si(111) double monochromator was used to provide a flux on the sample of 1012 ph/s, with a beam-size of 100×100 μm and an energy resolution of 1.4 eV at the Fe K-edge (i.e., 7712 eV). Energy calibration of the monochromator was based on the acquisition of XANES spectra in the transmission mode for an iron metallic foil. The incident beam intensity was measured using a ―lownoise‖ ion chamber filled with He-gas. XANES spectra were collected at the Fe K-edge, between 7100 and 7230 eV, in the fluorescence mode using a ―Ketek‖ silicon-drift detector positioned at 90° to the incident beam. For each XANES spectra, three measurements were made in the step-by-step mode with appropriate energy step and dwell times for the different regions of the spectra: 0.3 eV and 4 s for the pre-peak region; 0.5 eV and 3 s for thewhite line region; 2 eV and 2 s for the rest of the spectra. Data normalization (i.e., edge-jump and slope correction) was performed using the appropriate function of the XasMap program (Muñoz et al., 2006; Muñoz et al., 2008).
Between 0 and 7 days, the fluid composition is characterized by the strong increase of the Mg and Ca concentrations, from 0 to 1.32.10−3 and 3.25.10−4 mol.L−1 , respectively. Then the concentrations of these two elements decrease to reach, respectively, 4.12.10−5 and 1.68.10−4 mol L−1 after 34 days, and are then stabilized until 70 days. The concentrations obtained for Si are in the same order of magnitude compared to those of Ca. However, the maximum Si content is observed at 18 days (2.62 10−4 mol L−1 ) whereas it is located at 7 days for Ca. Note that the Fe concentrations were not measured since they remain below the detection limit. 3.2. Mineralogical evolution
The diffraction patterns for each altered sample are shown in Fig. A1. The mineral assemblage in the starting sample (i.e., 0 days) comprises forsterite, enstatite and clinopyroxene (diopside). After 7 days of hydrothermal alteration, the intensities of the diffraction peaks for these three minerals decrease significantly, suggesting degradation of the crystal structure. No other mineral species is observed at this stage. From then on, the diffractions peaks for the three departure minerals decrease progressively until 70 days, while the peaks for lizardite appear at 18 days and increase until the end of the experiment. Even after 70 days of exposure time, residual forsterite is still observed in the mineral assemblage, but the diffraction pattern is largely dominated by lizardite. As previously observed by Seyfried et al. (2007), magnetite seems to crystallize only in the latest stages of alteration since the related diffraction peaks only appear in the 70 days diffraction pattern. However, even small amounts of magnetite could potentially drive to the release of significant amounts of di-hydrogen, and detection limits using such ―traditional‖ laboratory XRD apparatus (in contrast with X-ray synchrotron sources for instance) may not allow the detection of magnetite in the early serpentinization stages