▪ Redox dynamics of subduction revealed by arsenic in serpentinite G.S. Pokrovski 1* , C. Sanchez-Valle 2 , S. Guillot 3 , A.Y. Borisova 1,8 , M. Muñoz 4 , A.-L. Auzende 3 , O. Proux 5 , J. Roux 6 , J.-L. Hazemann 7 , D. Testemale 7 , Y.V. Shvarov 8 Abstract https://doi.org/10.7185/geochemlet.2225 Redox dynamics of subduction processes remain poorly constrained owing to the lack of direct geochemical tracers. We studied, using X-ray absorption spectroscopy, the chemical and redox state of arsenic in the Tso Morari serpentinites that are witnesses of the Himalayan subduction. Our measurements reveal remarkably con- trasting redox speciation, from arsenide (As –III ) to arsenite (As III ) and arsenate (As V ). Combined with physical-chemical constraints, these data enable reconstruction of the ‘redox travel’ of arsenic in the subduction process. Upon early serpentinisation of mantle peridotite, arsenic was scavenged from the fluid and dragged down as insoluble nickel arsenide. Partial deserpentinisation close to the peak metamorphism (550–650 °C) resulted in oxidative dissolution of arsenide to aqueous As III and As V and their non-specific intake by antigorite. The As V /As III ratios (∼0.1–10) analysed in the mineral are ∼10 4 times higher on aver- age than predicted assuming bulk system thermodynamic equilibrium. These findings reflect a transient out-of-equilibrium release of highly oxidised fluids, with f O 2 reaching ∼10 log units above the fayalite-magnetite-quartz buffer (FMQþ10). Arsenic in serpentinite is thus a sensitive record of subduction redox dynamics inaccessible when using traditional equilibrium approaches applied to bulk fluid-mineral systems. Received 4 February 2022 | Accepted 19 May 2022 | Published 14 July 2022 Introduction Serpentinite formation and breakdown are major phenomena occurring in subduction zones. Knowledge of the redox condi- tions (f O 2 ) in these processes is necessary to interpret the trans- fers of many major and trace elements existing in multiple oxidation states. Most natural, experimental, and modelling studies have tackled redox evolution during subduction using major redox sensitive elements such as C, S, and Fe, but little attention has been devoted to trace elements. Among them, arsenic may be a promising redox indicator because it exhibits a wide range of formal oxidation states, from −III to þV, yielding a variety of minerals from (sulf)arsenides to arsenates, and oxy- hydroxide As III and As V species in fluids that may be scavenged or released by major minerals and silicate melts depending on f O 2 (e.g., Noll et al., 1996; O’Day, 2006; Perfetti et al., 2008; Borisova et al., 2010; Testemale et al., 2011; Scambelluri et al., 2019). The present study thus examines the potential of arsenic for tracing subduction zone redox conditions through measure- ment of arsenic oxidation state and speciation in the Tso Morari serpentinites (NW Himalaya). These rocks were formed by hydration of forearc mantle peridotites by slab-derived fluids and subducted to depth of ∼100 km and temperatures of ∼650 °C before having been exhumed during the Himalayan orogenesis (Hattori and Guillot, 2007 and references therein). The serpentinites, constituted mostly of antigorite and magnetite (Fig. 1), are highly enriched in As V and As III (up to ∼100 ppm As, which is ∼1000 times more than in mantle-derived rocks; Hattori et al., 2005; Witt-Eickschen et al., 2009). Combined with the well constrained geodynamic history of the Himalayan subduction (Supplementary Information), these serpentinites represent an excellent natural case to examine the redox cycle of arsenic in a palaeo-subduction zone across a wide range of temperatures (T) and pressures (P). We used synchrotron X-ray absorption spectroscopy (XAS), which is the most direct method to probe a trace element redox state, chemical bonding and coordination at the atomic scale. Arsenic K-edge X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS) spectra were acquired on a thoroughly characterised set of serpentinite rock samples as well as their antigorite- and magnetite-enriched mineral fractions (Supplementary Information; Table S-1). Combined with simulations of 1. Géosciences Environnement Toulouse, Université Toulouse III, GET-CNRS-OMP-UPS-IRD-CNES, 31400 Toulouse, France 2. Institute of Mineralogy, University of Münster, 48149 Münster, Germany 3. Université Grenoble Alpes, Université de Savoie Mont Blanc, CNRS, IRD, Univ. Gustave Eifel, ISTerre, 38000 Grenoble, France 4. Géosciences Montpellier, Université de Montpellier, CNRS, 34095 Montpellier, France 5. Observatoire des Sciences de l’Univers de Grenoble, CNRS-Université Grenoble Alpes, 38400 Saint Martin d’Hères, France 6. Institut de Physique du Globe de Paris, CNRS, 75005 Paris, France 7. Université Grenoble Alpes, CNRS, Grenoble INP, Institut Néel, 38000 Grenoble, France 8. Lomonosov Moscow State University, 119991 Moscow, Russia * Corresponding author (email: gleb.pokrovski@get.omp.eu; glebounet@gmail.com) © 2022 The Authors Published by the European Association of Geochemistry Geochem. Persp. Let. (2022) 22, 36–41 | https://doi.org/10.7185/geochemlet.2225 36