Fast assimilation of serpentinized mantle by basaltic magma

The most abundant terrestrial lavas, mid-ocean ridge and ocean island basalt (MORB and OIB), are commonly considered to be derived from a depleted MORB-mantle component and more specific, variably enriched mantle plume sources. However, findings of oceanic lavas and mafic cumulates issued from melts, enriched in chlorine and having a radiogenic Sr ratio, can be attributed to an interaction between the asthenosphere-derived melts and lithospheric peridotite variably hydrated due to penetration of hydrothermal water down to and below Moho level. To constrain mechanisms and rates responsible for the interaction, we report results of experiments of reaction between serpentinite and tholeiitic basaltic melt. Results show that the reaction proceeds via a multi-stage mechanism: (i) transformation of serpentinite into Cr-rich spinel-bearing harzburgite containing pore fluid, (ii) partial melting and dissolution of the harzburgite assemblage with formation of interstitial hydrous melts, and (iii) final assimilation of the Cr-rich spinel-bearing harzburgite/dunite and formation of hybrid basaltic melts with high MgO and elevated Cr and Ni contents. Assimilation of serpentinite by basaltic melt may occur under elevated melt/rock ratios and may lead to chromitite formation. Our experiments provide evidence that MORB and high-Mg-Cr orthopyroxene-rich cumulates depleted in incompatible elements can be produced from common mid-ocean ridge basaltic melts modified by reaction with hydrated lithospheric peridotite. We established that the rate of assimilation of serpentinized peridotite is controlled by silica diffusion in the reacting hydrous basaltic melt.


Introduction
Mid-ocean ridge basalt (MORB) and ocean island basalts (OIB) are considered as products of decompression melting of several asthenospheric sources characterized by contrasted isotopic signatures (e.g., DMM, HIMU, EM-1 and EM-2; e.g., Zindler and Hart, 1986). This paradigm implies that the oceanic basalt composition is some kind of "carbon copy" of their deep mantle sources. However, the occurrence of basaltic glasses and high-Mg cumulates with radiogenic 87 Sr/ 86 Sr ratios along present-day mid-ocean ridges (e.g. Ross andElthon, 1993, Nonnotte et al, 2005;van der Zwan et al., 2017) and in ophiolites, (Amri et al., 1996;Benoit et al., 1999;Clenet et al., 2010;Lange et al., 2013) suggest more complex mechanisms for the generation of oceanic magmas involving the assimilation of altered lithospheric material. Experimental and melt inclusion studies highlight the important role of basaltic melt-lithospheric rocks reactions on the chemical composition of both mantle rocks and silicate melts (e.g., Morgan & Liang, 2003;Kvassnes and Grove, 2008;France et al., 2010;Van den Bleeken et al., 2010;2011;Borisova et al., 2012a;. For example, Sr isotope diversity of basalts and mafic cumulates possibly related to assimilation of seawater-altered rocks is common in spreading environment (Michael and Cornell, 1998;Lange et al., 2013;van der Zwan et al., 2017). Magmas resulting from the mixing between "asthenospheric" tholeiites and hydrated "lithospheric" melts of depleted andesitic affinity were proposed as parental for some puzzling occurrences of orthopyroxene-rich primitive cumulates along mid-ocean ridges and in ophiolites (Benoit et al, 1999;Nonnotte et al., 2005). The formation of such hybrid or/and depleted magmas due to assimilation of serpentinized mantle peridotite leads to the formation of the Moho transition zone composed by dunites and chromitites (Borisova et al., 2012a;Zagrtdenov et al., 2018;Rospabé et al., 2019a,b).
Hydrothermal circulation in the oceanic lithosphere near and below the mantle/crust boundary ("petrologic Moho") is recorded in such hydrothermal rocks as mantle diopsidites hosted by peridotites in the shallow mantle section of ophiolites (Python et al., 2007;Akizawa & Arai, 2014) as well as in multiphase inclusions representing fluid phase, not a magma, in chromitite orebodies (Borisova et al., 2012a;Johan et al., 2017). Assimilation of seawaterderived brine or aqueous fluid by mafic rocks, magmas and melts, is recorded also at ultraslow-, slow-, and fast-spreading mid-ocean centers and preshield-stage Loihi Seamount lavas (Hawaii) (Michael & Schilling, 1989;Jambon et al., 1995;Michael & Cornell, 1998;Kent et al., 1999;Dixon & Clague, 2001;Dixon et al., 2008;Klein et al., 2019). Seawater-derived component assimilation, possibly through interaction with serpentine, happens at slowspreading centers or at small oceanic islands (Simons et al., 2002;Dixon et al., 2008). It is also reported that basaltic glasses, melt inclusions and the serpentinite-hosted gabbroic veins in the environment of the slow-spreading ridges are affected by assimilation of hydrothermally altered lithosphere enriched in such volatile elements as Cl, H2O, and atmospheric Ne, Ar isotopes (Stroncik & Niedermann, 2016;van der Zwan et al., 2017;Ciazela et al., 2017;. The shallow depths of the melt-or magma-rock interactions extend to 10 to 13 km, suggesting not only crustal but also upper mantle source of such contamination (van der Zwan et al., 2017).
Additionally, serpentinites play an important role during melt-rock interactions of upper mantle, which was thereafter exposed in the detachment faults (e.g., Bach et al. 2004;Sauter et al., 2013;Ciazela et al., 2017), where upper mantle serpentinites may be abundant in the crust of slow and (mostly) ultraslow-spreading ridges. Indeed, hydrated peridotites, in particular, upper mantle serpentinites are high-Mg rocks variously enriched in refractory elements (Cr, Ni) and fluid-mobile elements (e.g., halogens, H, B, O, He, Ar, As, S, Sb, Sr and Pb) due to interaction between peridotite and seawater-derived low-to-moderate-temperature hydrothermal fluids (e.g., Guillot et al., 2001;Früh-Green et al., 2004;Bonifacie et al., 2008;Deschamps et al., 2010;Evans et al., 2013;Guillot & Hattori, 2013;Kendrick et al., 2013). These rocks crop out preferentially below the mantle-crust transition zone (Evans et al., 2013) which is locally exposed to the seafloor at mid ocean ridges (e.g., Bach et al., 2004) and may be sampled as upper mantle xenoliths by lavas of such oceanic islands as Canary Islands (Neumann et al., 2015) at possible reaction depths of up to 1.0 GPa pressure.
Although the interaction between dry tholeiitic basalt and anhydrous harzburgite has been investigated experimentally by Fisk (1986), Kelemen et al. (1990), Morgan and Liang (2003), and Van den Bleeken et al. (2010;2011), the rates and mechanisms of assimilation of the hydrated mantle lithosphere (or serpentinite rocks) by the basaltic magmas and the chemical impact of the hydrated mantle on the composition of oceanic basalts are still unconstrained.
Such constraints are important to understand the evolution of the oceanic lithosphere and to reconsider some inferences of geochemists on the nature and composition of mantle sources (Nonnotte et al., 2005;Kendrick et al., 2017). Our experiments were designed to study the reaction processes between moderately differentiated basaltic melt and serpentinite with high melt/rock ratio at pressures of 0.2 to 1.0 GPa. Our new data contribute to constrain the rate, mechanism and compositional impact of the assimilation of the serpentinized lithospheric mantle by the basaltic magma.

Starting Materials and Analytical Methods
The mid-ocean ridge basalt used in the experiments is a typical moderately differentiated ( (Sushchevskaya et al., 2000). The serpentinite used as starting material is a homogeneous rock composed by antigorite with accessory Fe-rich oxides, devoid of relics of primary mantle silicates, sampled in Zildat, Ladakh, northwest Himalaya (e.g., Deschamps et al., 2010). The composition of the starting materials is given in Table  For the hybrid runs, the serpentinite has been prepared as doubly polished ~1000 µmthick section, thereafter cut to 2.7 mm-diameter cylinders by a core drill machine. The MORB glass has been crushed to powder (<100 µm glass size). Additionally, for the mixed runs at 0.2 GPa pressure, the serpentinite sample has been crushed to powder (<100 µm glass size).

Experimental Strategy and Method
The experiments in the system containing basalt and serpentinite were performed at 0.2 to 1.0 GPa and 1250 to 1300°C. Although in modern oceanic settings, temperatures of 1050°C are sufficient to initiate reaction of hydrated peridotite with basaltic magma at 0.2 GPa (Borisova et al., 2012a), we chose higher temperatures for the experiments as higher temperatures substantially increase reaction rates of the serpentinite-basaltic melt interactions and ensure conditions corresponding to complete melting of the basalt, consistent with the majority of existing models of basaltic melt extraction from the mantle (Fisk, 1986;Hirschmann et al., 1998;Ulmer, 2001;Morgan and Liang, 2003).
Fifteen experimental runs were performed at 0.2 to 1.0 GPa with similar serpentinite to basalt ratios (with 12 to 28 wt% of serpentinite in the mixture), the similar initial bulk water content and different run duration (Fig. 1, Table 2). Two series of experiments were performed: (a) hybrid series with serpentinite cylinder and basaltic powder at 0.2 and 1.0 GPa and (b) mixed series composed of intimately and well-mixed serpentinite and basaltic powders uniquely at 0.2 GPa pressure. The experimental design of the hybrid runs included a serpentinite cylinder in the upper part and dry or wet MORB glass powder in the lower part of the Au80Pd20 (exceptionally, in one run of the Pt) capsule. The mixed runs were performed with mixtures of basaltic powders and 20 wt% of serpentinite powder in the bulk mixture ( Fig. 1, Table 2). The starting components were weighed before runs. The distilled water MQ was mixed with the MORB powder in the experiments with additional water uniquely at 0.5 GPa pressure. The experimental runs were performed at temperature of 1250 -1300°C. Because the run duration was shorter than 48 hours which is necessary to reach oxygen fugacity equilibrium with a piston-cylinder double-capsule techniques of mineral buffer (Matjuschkin et al., 2015), the redox conditions in our kinetic experiments were controlled by the initial Fe 2+ /Fe 3+ ratios in the MORB glass and serpentinite, although more oxidized conditions were established during the runs due to the presence of water and partial H2 loss to Al2O3 pressure media. The Fe 2+ /Fe 3+ ratios in several quenched products were estimated using XANES (X-ray absorption near edge structure) at European Synchrotron Radiation Facility (ESRF) in Grenoble (France). Oxygen fugacity relative to the quartz-fayalite-magnetite buffer (QFM) was calculated based on the obtained ferric iron (Fe III ) mole fraction (XFeIII) in the analyzed samples (Table 2) using of the FeO total contents in the predominant basaltic melt and model of Borisov et al. (2018). The redox conditions in the shortest runs (<4 h) were estimated from the olivine-chromite assemblages using equations of Ballhaus et al. (1991) and from mole fraction of ferric iron Fe III obtained by XANES at corresponding temperatures to range from QFM(+1.5) (for P10) to QFM(+3.6) (for P1) (Fig. 1, Table 2).
Two piston-cylinder systems were used in our experiments. Experiments (P1 -P10) used the end-loaded Boyd-England piston-cylinder apparatus at the Korzhinskii Institute of Experimental Mineralogy, Chernogolovka, Russia. Standard talc-Pyrex cells 3/4 inch in diameter, equipped with tube graphite heaters and inserts made of MgO ceramics were used as pressure-transmitting medium. The pressure at elevated temperatures was calibrated against two reactions of brucite = periclase + H2O and albite = jadeite + quartz equilibria. A pressure correction (12%) was introduced for the friction between the cell and hard-alloy vessel. To minimize the friction, a Pb foil and molybdenum disulfide (MoS2) lubricant were used.
Au80Pd20 capsules with starting mixtures were mounted in the central parts of the cells. The temperature in the upper part of the capsules was controlled to be accurate to ±1°С using a MINITHERM controller via a W95Re5/W80Re20 thermocouple insulated by mullite and Al2O3 without pressure correction. For the Pyrex-bearing assemblies the sample was heated to 550 -600°C at low confining pressure (0.15 -0.2 GPa) for a few minutes in order to soften the Pyrex glass, subsequently both temperature and pressure were increased almost simultaneously up to the desired run conditions. The samples were maintained at run conditions during desired durations ( Table 2). The experiments were quenched by switching off electricity. The quench rate was 100-300°С/min.
The experiments (P15 -P26) used the "Max Voggenreiter" end-loaded Boyd-England piston-cylinder apparatus at the Bavarian Research Institute of Experimental Geochemistry and Geophysics (BGI), Bayreuth, Germany. Talc cells 3/4 inch in diameter with Pyrex sleeves were used. A tapered graphite furnace was inserted in each cell. Alumina (Al2O3) spacers were used as pressure-transmitting medium. An Au80Pd20 capsule loaded with starting materials was set in the central part of the assembly. A 20% pressure correction was applied for the friction between the talc cell and pressure vessel. A molybdenum disulfide (MoS2) lubricant was introduced to minimize the friction. The temperature in the upper part of the capsules was controlled by a EUROTHERM (2404) controller via either W3Re97/W25Re75 (type D) or Pt6Rh94/Pt30Rh70 (type B) thermocouple accurate to ±0.5°С. The sample was compressed to 0.5 GPa during a period of 20 minutes and then heated up to the run temperature (1300 °C) at a rate of 100 °C/min. The samples were maintained at run conditions during desired durations ( Table   2). The experiments were quenched by switching off electricity. We have applied decompression during periods from 20 minutes to 2 hours. The rate of quenching to the ambient temperature was ~ 300°С/min. Five experiments (P36, P37, SB1, SBbis3, and SBter1) were carried out at pressure of 0.2 to 0.5 GPa and a temperature of 1250 °C using an internally heated gas pressure vessel at the Korzhinskii Institute of Experimental Mineralogy, Chernogolovka, Russia. The pressure in the system was created by pure Ar gas. The system was heated by a furnace with two windings (to minimize the thermal gradient). The temperature was set and measured by a TRM-101 OVEN controller through two S-type (Pt90Rh10 vs Pt100) thermocouples. The thermocouples were mounted at the top and close to the bottom of the run hot spot to monitor the temperature gradient. The duration of experiment was from 0.5 to 48 h ( Table 2). The experiments were quenched by switching off the furnace. The pressure during the quench was maintained constant down to 550 °C, and then slowly released. The cooling rate from 1250 to 1000 °C was 167 °C/min, and then 90 °C/min down to 550 °C. After the runs, the capsule was mounted in epoxy, cut in two parts using a diamond saw, and then polished using SiC sand papers and diamond pastes.
To calculate phase proportions, we have used the PYTHON language code. At the first step, the initial MORB glass was introduced instead of Lbas and Lint. On the second step, the basaltic and interstitial glasses (Lbas and Lint, respectively) were distinguished as two phases.

Scanning Electron Microscope (SEM) and Electron Microprobe analysis (EPMA)
Major and minor element analyses of minerals and glasses and the experimental sample imaging were performed at the Géosciences Environnement Toulouse (GET, Toulouse, France) laboratory using a scanning electron microscope (SEM) JEOL JSM-6360 LV with energydispersive X-ray spectroscopy (EDS), coupled with the automatic analyzer of particles by the program "Esprit". The main experimental phases (oxides, silicates and glasses) in the samples have been identified by EDS microprobe technique at GET (Toulouse, France) (Borisova et al., 2012b). Major, and minor element compositions of the crystals and glasses were analyzed using CAMECA SX-Five microprobe at the Centre de Microcaractérisation Raimond Castaing (Toulouse, France). Electron beam of 15 kV accelerating voltage, and of 20 nA current was focused or defocused on the sample to analyze minerals or glasses, respectively. The following synthetic and natural standards were used for calibration: albite (Na), corundum (Al), wollastonite (Si, Ca), sanidine (K), pyrophanite (Mn, Ti), hematite (Fe), periclase (Mg), Ni metal (Ni), and Cr2O3 (Cr). Element and background counting times for most analyzed elements were 10 and 5 s, respectively, whereas, peak counting times were 120 s for Cr and 80 to 100 s for Ni. Detection limits for Cr and Ni were 70 ppm and 100 ppm, respectively. The silicate reference materials of Jarosewich et al. (1980) as well as MPI-DING glasses of ultramafic to mafic composition (GOR132-G, GOR128-G, KL2-G and ML3B-G of Jochum et al., 2006) were analyzed as unknown samples to additionally monitor the analysis accuracy. The silicate reference material analysis allowed to control precision for the major and minor (e.g., Cr, Ni in glasses) element analyses to be in the limit of the analytical uncertainty (related to the count statistics). The accuracy estimated on the reference glasses ranges from 0.5 to 3 % (1σ RSD = relative standard deviation), depending on the element contents in the reference glasses.
H2O contents of glasses were estimated based on the in situ electron probe analyses of the major element oxides. The 0.5 -1.0 GPa glasses were analyzed following simplified "by difference method" without bracketing. The uncertainty of the water contents generally varies between 10 and 50 %. The 0.2 GPa glasses were analyzed by using bracketing mode (Borisova et al., 2020).

X-ray Absorption Near Edge Structure (XANES) Spectroscopy
Iron redox state in selected quenched glasses was determined from Fe K-edge (~7.1 keV) XANES spectra acquired at the FAME beamline (Proux et al., 2005) of the European Synchrotron Radiation Facility (ESRF). The beamline optics incorporates a Si (220) monochromator with sagittal focusing allowing an energy resolution of ~0.5 eV at Fe K-edge and yielding a flux of >1012 photons/s and a beam spot of about 300×200 µm. XANES spectra were acquired in fluorescence mode in the right-angle geometry using a 30-element solid-state germanium detector (Canberra). Energy calibration was achieved using a Fe metal foil whose K-edge energy was set to 7.112 keV as the maximum of the spectrum first derivative. Ironbearing oxides and silicates with different Fe redox and coordination environment, diluted by mixing with boron nitride to obtain Fe concentrations of a few wt%, were measured similarly to the glasses to serve as reference compounds.
Iron redox state in the glasses was determined by fitting the XANES pre-edge region (7.108-7.120 keV) according to the protocols developed in Muñoz et al. (2013) and using a background polynomial and two pseudo-Voight functions to determine the energy position of the pre-edge peak centroid, which is a direct function of the Fe III /Fe II ratio both in crystalline and glass silicate samples (Wilke et al. 2001;2005). Ferric iron (Fe III ) mole fraction (XFeIII) in the starting basalt, and sample P3 was determined using the calibration established for basaltic glasses (Wilke et al., 2005), while sample P1 that showed a mixture of glass and crystals was processed using the calibration established for Fe minerals (Wilke et al., 2001). The results are reported in Table 2. The uncertainties of XFeIII determination for dominantly glassy samples are 0.05 in absolute value, while those for glass-crystal mixtures are typically 0.07 of the value.

Laser Ablation Inductively Coupled Plasma Mass Spectrometry
Major and trace element concentrations were determined by LA-ICP-MS at the Max Planck Institute for Chemistry, Mainz, using a New Wave 213 nm Nd:YAG laser UP 213, which was combined with a sector-field ICP-MS Element2 (Thermo Scientific) (Jochum et al., 2007;. Ablation took place in the New Wave Large Format Cell under He atmosphere. Spot analyses were performed in low mass resolution mode using crater sizes of 30 µm (lines of spots) and 8 µm (single spots), and a pulse repetition rate of 10 Hz at a fluence of about 6.5 J/cm 2 (lines of spots) and 8.3 J/cm 2 (single spots). Isotopes used for analysis are the following:  175 Lu, 178 Hf, 232 Th and 238 U. Data reduction was performed by calculating the ion intensities of each isotope relative to the intensity of 43 Ca. The NIST SRM 610 silicate glass (for trace and major elements except Mg, K, Fe) and the basaltic glass GSE-1G (Mg, K, Fe) were used for calibration. Major element concentrations were calculated to a total oxide content of 99 wt%.

Results of Experiments on Basaltic Melt-Serpentinite Interaction Experimental Sample Description
The first group of experiments was conducted at 0.5 GPa and 1300 °C with duration up to 8 hours (Tables 2 -4, Table A1). Products of the shortest run P1 (1 minute at the run temperature) which was considered as a zero-time experiment, the quenched basaltic glass zone and a zone replacing serpentinite (former serpentinite zone) are present. The former serpentinite zone contains fine-grained (5 to 10 µm in size) aggregates of olivine Fo95, enstatite (Mg# = 95), chromite (Cr# = 89) and interstitial glass of basaltic andesite composition ( Fig. 1). Chromite crystals (a few micrometers in size) are disseminated within this zone.
Samples P15 and P10 were kept for 0.5 and 2.6 hours, respectively, at run conditions. Distinct quenched melt zone and the former serpentinite zone are also present in this sample.
Sample P18 provides an information about 5 hours-lasting basaltic melt-serpentinite interaction. The sample shows a hydrous basaltic zone and former serpentinite zone. The former serpentinite zone consists of two areas. The outer area of nearly 200 μm width contains olivine, interstitial basaltic glass and disseminated chromite (grain size of a few micrometers). The inner area consists of forsteritic olivine Fo93, enstatite (Mg# = 97) and contains magnetite aggregate (~20 × 40 μm) (Fig. 1).
P36 is the longest (8 hours) experiment of the series at 0.5 GPa. Serpentinite is completely dissolved in the basaltic melt, and the run products are represented by homogeneous basaltic glass with 11.5 wt% MgO. The second series of experiments has been performed at 1.0 GPa and 1300 °C. Experiment P3 with duration 2.5 hours shows basaltic glass zone and the former serpentinite zone where forsteritic olivine Fo92, enstatite (Mg# = 96) and Cr-bearing magnetite are associated with interstitial glass of basaltic composition (Fig. 1). The 9 h-long run without additional water (P7) and 3 h-long run with additional water (P12) contain uniquely hydrous basaltic glasses with 13.0 wt% MgO.
Thus, at 0.5 -1.0 GPa pressure range, the initial stage of basaltic melt-serpentinite reaction generates two contrasting zones: an olivine-rich zone composed of mostly harzburgite (forsteritic olivine Fo93 -95 and enstatite Mg# = 93 -95) with an outer dunite portion, and a reacting basaltic zone. These zones are similar to those produced in the anhydrous peridotitebasalt systems at 0.1 MPa -0.8 GPa (Fisk, 1986;Morgan & Liang, 2003). An addition of water at 1.0 GPa at the conditions of the basaltic melt saturation with water fluid phase likely decreases the timescale required for the serpentinite assimilation from 9h in the P7 run to 3h in the P12 run (see Table 2).
Additionally, the hybrid run P37 sample is represented by predominant basaltic glass of Lbas (72.8 wt% in the sample) with assemblage of interstitial glass (Lint, 0.4 wt%), forsteritic olivine (17.0), residual orthopyroxene (6.8), and accessory clinopyroxene and chromiferous magnetite with pores of fluid (3.0) ( Table 2). The mixed sample number SB1 obtained at 0.2 GPa is represented by polyhedral olivine phenocrystals in matrix. In the matrix, this sample contains assemblage of clinopyroxene microphenocrysts, rims of the olivine phenocrysts and interstitial felsic glasses. Oxide minerals are represented by chromite microphenocrysts. The sample SBter1 is represented by homogeneous basaltic glass formed by complete hybridation of the starting basaltic liquid with serpentinite, whereas the sample SBbis3 contains residual crystallized aggregate of olivine. It is worth noting that the current experiments with predominant proportion of basaltic melt (72 -88 wt%) longer than 5 -8 h at 0.2 -1.0 GPa produce the total assimilation of the serpentinite zone by the basaltic melt ( Fig. 1), resulting in homogeneous Mg-rich basaltic glasses.

Summary on the Melt Composition
The olivine-rich zones host glass pockets of 10 to 200 µm in size. The composition of the interstitial glasses produced in the shortest hybrid runs at 0.5 to 1.0 GPa varies from basaltic, basaltic andesite to andesitic (Fig. 2). The interstitial dacitic melts even richer in silica (up to 66 to 71 wt%) were produced at 0.2 GPa pressure compared to those obtained in the 0.5 -1.0 GPa pressure (Fig. 2) and are discussed elsewhere (Borisova et al., 2020). In this work, we pay more attention to the mechanisms and to the rates of assimilation relevant to the magmas interacting with hydrated mantle lithosphere in oceanic setting. The interstitial melts produced at 0.2 to 1.0 GPa are formed close to equilibrium with olivines of the olivine-rich zones (Fig.   3). The major element composition of the basaltic melts produced due to the bulk serpentinite ratios compared to those of the starting basaltic melt (1.45 wt%, 1.1 and 1.6, respectively) due to the low content of these elements in the serpentine (Table 4). Thus, the chemical impact of serpentinite on the final basaltic melt as a result of the bulk assimilation is a dilution in incompatible elements (e.g., Ti) contents and an enrichment in compatible Cr, Ni and Mg elements, as well as a slight enrichment in Si contents.

Mechanism of the Basaltic Melt-Hydrated Peridotite Reaction
Hybrid and mixed experiments on identical materials with high basaltic melt to serpentinite rock ratio (>2) performed at different run durations allowed determination of the reaction mechanism and the assimilation rate. The transformation of serpentinite to dehydrated harzburgite with pore fluid is demonstrated in the zero-time experiment at 1300°C and 0.5 GPa (Fig. 1, Table 2). The serpentinite in the shortest runs produces forsteritic olivine (Fo91-95), enstatite (Mg# = 94 -97) and chromite and/or chromiferous magnetite with Cr# = 7 -89 and Mg# = 27 -54 (Fig. 1, Table 2), similarly to the results of Chepurov et al. (2016). The phase composition diagram and MgO-SiO2 plot (Figs. 1, 3) suggest that the basaltic melt-serpentinite reaction at 0.2 -1.0 GPa is controlled by the following stages: (1) transformation of serpentinite to chromite-bearing harzburgite (crystallization of forsteritic olivine and enstatite with accessory chromite and liberation of pore fluid), (2) incongruent partial melting of the harzburgite and formation of olivine-rich zones with hydrous interstitial melts. Subsequently, the mechanism involves (3) formation of external chromite-bearing dunite zone due to incongruent dissolution of orthopyroxene associated with diffusive exchange between the reacting basaltic melt and the interstitial melts, (4) partial dissolution of the chromite-bearing dunite in the initial basaltic melt and finally (5)  Fractional crystallization of the hybrid, depleted and chromite-saturated basaltic melts can produce chromitites. Indeed, according to the recently proposed model of the chromitite genesis, the bulk serpentinite assimilation by MORB basaltic melt is the main factor responsible for the massive chromite crystallization (Borisova et al., 2012a) at the oceanic Moho mantle-crust transition zone. Our experiments demonstrate that formation of the hybrid mid-ocean ridge basaltic melt saturated with chromite is possible at conditions of predominant proportion of basaltic melt (above 70 wt%) and, therefore, due to high melt/rock ratio (>2) at 0.2 to 1.0 GPa pressure. The main physico-chemical parameters controlling the chromite crystallization are the presence of aqueous fluid or/and hydrous basaltic melts into the reactive system (Borisova et al., 2012a;Johan et al., 2017;Zagrtdenov et al., 2018). The fluid presence at 0.2 GPa pressure is also necessary condition for the chromite concentrating by a physical, not a chemical process due to surface tension of the fluid, which is sufficient to maintain dispersed chromite crystals inside the fluid upon the chromite crystallization (Matveev & Balhaus, 2002).
Additionally, an initial stage of the basaltic melt reaction with dehydrated serpentinite may result in formation of chromite-bearing harzburgite and dunite. The first direct evidence that the initial stage really happens at the 0.2 GPa mantle-crust transition zone is occurrence of chromite-hosted silica-rich inclusions of the Oman ophiolite chromitite ore bodies (Rospabé et al., 2019b).

Assimilation Rate of the Hydrated Mantle Lithosphere
The calculated average rate of the serpentinite assimilation by the basaltic melt in the experiments without additional water is 4.3 × 10 -10 m 2 /s ( Table 5). It is surprisingly similar to the assimilation rate by basaltic melt in the runs with an additional water at conditions of the melt saturation with an aqueous fluid (~4.0 × 10 -10 m 2 /s). Figure 4 demonstrates that the serpentinite assimilation rate by dry basaltic melt measured at 0.5 GPa is progressively decreasing during the experimental run. This may be explained by approaching equilibrium, in accordance with principles of the chemical kinetics. The calculated rates are at least one order of magnitude higher compared to 10 -12 -10 -11 m 2 /s for the basalt interacting with anhydrous harzburgite at 0.6 GPa (Morgan and Liang, 2003). The assimilation rate estimated in our work is comparable to the silica diffusivity in a hydrous basaltic melt (Zhang et al. 2010). Similarly, the reaction rate established by Morgan and Liang (2003) in anhydrous system is mostly comparable to the silica diffusivity in dry basaltic melt. Since both reaction rates are controlled by the silica diffusion, the difference in the reaction rates is related to the well-known promoting effect of H2O on the silica diffusion in silicate melts (e.g., Zhang et al. 2010 and references therein). Thus, the serpentinite assimilation by basaltic magma may proceed at least 10 times faster than the formation of dunitic reaction margins (i.e. "dyke walls") in the oceanic lithosphere during transport of dry basaltic melt (Morgan and Liang, 2003) at 0.6 -0.8 GPa.
Additionally, the newly produced hybrid basaltic melts which become highly saturated in olivine, chromite and, likely, in orthopyroxene would produce plagioclase-free chromitebearing harzburgite through reactive porous flow in a peridotite. This process may be expressed in nature by formation of chromite-bearing harzburgitic rather than dunitic channels of depleted hybrid magmas enriched in Si, Mg, Cr, and H2O. Similarly, the radiogenic 87 Sr/ 86 Sr coupled with an excess of H2O, SiO2, halogens, 4 He, 36 Ar in oceanic basalts, related cumulates and associated mantle peridotites is undisputable evidence for their origin due to serpentinization of the mantle lithosphere (Neumann et al., 2015) and the basaltic melt reactions with the serpentinized mantle rather than due to partial melting of deepseated serpentinized mantle (DMM, HIMU, e.g., Kendrick et al., 2017).

Comparison to Chemistry of Oceanic Magmas and Glasses
The effect of the altered lithosphere assimilation is likely widespread given the elevated chlorine contents (> 50 ppm) and the resulting high Cl/K ratios (> 0.09) in the MORB glasses Additionally, ultra-depleted MORB and related variously depleted melts are produced beneath mid-ocean ridges, likely at 0.5 -2.0 GPa pressure in assemblage with forsteritic olivine (Fo82-91 mol.%) according to Ross and Elthon (1993), Sobolev andShimizu (1993) andHunsen et al. (2016). Routinely, origin of such ultra-depleted and differently depleted MORB melts has been attributed to an effect of critical (continuous) melting with formation of lherzolite/harzburgite residue. Chemical similarity of our experimental Cr-Mg-H2O-rich melts and associated high-Mg silicates such as forsteritic olivine (Fo91-95 mol.%) to those associated to forsteritic olivine (Fo82-91 mol.%) (e.g., Sobolev and Shimizu, 1993;Husen et al., 2016) implies that origin of some part of these variously depleted oceanic basaltic melts may be attributed to the reaction between tholeiitic basaltic melt and the serpentinized mantle rather than to fractional "dynamic" melting of mantle peridotite.

Integrated Model of Basaltic Melt -Hydrated Peridotite Reaction
It is likely that the serpentinized lithosphere assimilation may be stronger along slow-and ultraslow-spreading ridges due to faults rooting deeper, providing pathway for hydrothermal fluids (e.g., Mével & Cannat, 1991;van der Zwan et al., 2017   2) The effect of the reaction of tholeiitic basaltic melt with hydrous peridotite contrasts strongly with that observed on anhydrous peridotite at 0.65 -0.80 GPa by Van den Bleeken et al. (2010;2011). The main difference is the total assimilation of hydrous harzburgite by the basaltic magma and a complete absence of plagioclase in the reacted harzburgite. That is likely due to the effect of elevated water contents into the hybrid systems suppressing crystallization of plagioclase. Tholeiitic basaltic melts reacting with and assimilating the hydrous peridotite become more saturated in olivine, chromite and orthopyroxene and may produce chromite-bearing harzburgite upon reactive melt transport in the upper mantle.
3) The rate of the bulk serpentinized peridotite assimilation by tholeiitic basaltic melt (4.0 -4.3 × 10 -10 m 2 /s) is controlled by the silica diffusion in hydrous basaltic melts. The calculated rates of the serpentinite assimilation are at least one order of magnitude higher than 10 -12 -10 -11 m/s 2 for dry basaltic melt interacting with anhydrous harzburgite and producing dunite channels due to reactive porous flow of the basaltic melt (Morgan and Liang, 2003). The bulk assimilation of serpentinite by basaltic melt may happen at conditions of high melt/rock ratio (>2) and the predominant proportion of basaltic melt above 70 wt% in the hybrid system.

4) Our study challenges the routine interpretation of variations in chemical and isotopic
composition of oceanic lavas (e.g., MORB and OIB) in terms of deep mantle plume source heterogeneities or/and mechanism of partial melting.  Soc. Spec. Publ., vol. 42, pp. 313-345. Sushchevskaya N.M., Tsekhonya T.I., Kononkova N.N., Tcherkashov G.A., Bogdanov Y.A., Belyatsky B.V., (2000).  Figure 1. Pressure (in GPa) versus experimental run duration (in hours) diagram in the experimental basaltic melt-serpentinite system (without additional water) showing mineral and glass assemblages produced in hybrid and mixed runs at 1250 -1300 °C. The diagram and the back-scattered electron images of the samples show the main melt and mineral phases observed after quenching: Lbaszone of hydrous basaltic glass; Lintinterstitial glass; Opxorthopyroxene; Olforsteritic olivine; Ffluid; Serpserpentine, mostly antigorite minerals issued from the starting serpentinite. The run numbers and corresponding run products are given in Table 2. Figure 2. Al2O3,TiO2,FeO,MgO,CaO and Na2O (in wt%), Cr (ppm) and H2O (in wt%) contents versus SiO2 contents (in wt%) in the experimental glasses. Composition of the experimental glasses produced from 0.2 to 1.0 GPa is compared to the starting basalt (MORB) and the serpentinite compositions as well as to the compositional fields of natural tholeiitic and oceanic boninitic glasses (from database of PetDB, Lehnert et al., 2000), ultra-depleted melts marked as UDM (Sobolev and Shimizu, 1993) and chromite-hosted melts (Husen et al., 2016) which are differently depleted MORB melts. The recalculated melts obtained experimentally at 0.2 GPa pressure are after Borisova et al. (2020). H2O contents in the experimental glasses are values calculated from the EPMA. All experimental data are available in Tables 3 and A1. Figure 3 (a,b). Mechanism of the basaltic melt interaction with serpentinite. Five steps of the interaction have been distinguished: (1) dehydration and transformation of serpentinite to Crrich spinel-bearing harzburgite (crystallization of olivine and orthopyroxene) with appearance of pore fluid, (2) incongruent melting of the harzburgite and formation of associated hydrous interstitial melts in harzburgite/dunite, (3) progressive formation of external dunite zone due to orthopyroxene dissolution and diffusive exchange between the external basaltic melt and interstitial melts and (4) dissolution of the dunite in the basaltic melt and finally (5) bulk dissolution of the dunite and formation of hybrid basaltic melts. Lint and Lbas are interstitial and basaltic melts, respectively (see Table 2). Equilibrium Kd are calculated as theoretical FeO-MgO partition coefficient between olivine (Fo90-95 mol%) and co-existing interstitial melt of 57 to 62 wt% SiO2 at 0.5 GPa pressure according to Toplis (2005). The recalculated melts obtained experimentally at 0.2 GPa pressure are after Borisova et al. (2020). All experimental data are available in Tables 3 and A1.  Table 5 demonstrates the applied calculation method for the "relative dissolution rate". formation of harzburgite wall rocks in assemblage with interstitial hydrous silica-enriched (basaltic to dacitic) melts and diffusive homogenization between the initial oceanic basaltic (e.g., MORB, OIB) melt and the produced hydrous silica-enriched melts. At this stage, formation of peridotite xenoliths containing silica-rich inclusions is highly probable. (c) The final production of variously depleted hybrid melts locally associated with the chromite-bearing dunite/harzburgite. The bulk assimilation of serpentinite by basaltic melt may happen at conditions of high melt/rock ratio and the predominant proportion of basaltic melt above 70 wt% into the hybrid system.  (1) Lbas + F 0.21 -a Weight percent of the serpentine in the system is calculated as mass of serpentine divided by total mass of the all components: (MSerp / (MSerp + MMORB) for P1, P15, P10, P18, P25, P3, P7 or MSerp / (MSerp + MMORB + MH2O) for P20, P21, P26, P12, where MSerp, MMORB and MH2O are mass of serpentine, basaltic glass and additional water, respectively. b "Lint" = interstitial glass; "Lbas" = hydrous basaltic glass; "Serp" = serpentine; "Ol" = olivine; "Opx" = orthopyroxene; "Cpx" = clinopyroxene; "Amph" = amphibole; "Chr" = chromite; "ChrMgt" = chromiferous magnetite; "F" = water bubble(s). $ "Basalt" = Mid Atlantic Ridge basaltic glass; "Water" -additional water added to the starting system. $$ QFM = oxygen fugacity expressed in log units compared to the quartz-fayalite magnetite (QFM) mineral redox buffer according to Ballhaus et al. (1991).    Table 2. Bracketed numbers correspond to the number of analyses. The average composition and the glass homogeneity is represented as 1 σ std. deviation. La/Sm and La/Yb are normalized to the composition of the primitive mantle (Lyubetskaya and Korenaga, 2007). * volume calculated as cylinder volume based on initial 3D measurements of the serpentinite disc shape and based on measurements of 2D dimensions of the residual serpentinite using SEM in the polished sections. The volume difference is calculated as (Vint -Vexp)/Vint, where Vexp is volume after experimentation and Vint is initial volume. Volumes have been calculated from the serpentinite mass and density for the initial volume (Vint) and the 2D size measured using SEM after experimentation (Vexp). The uncertainty on the assimilation rate (relative rate) is in the limit of 15 rel. %. Avr.average rate values.