Heterogeneous hydrogen distribution in ...

Lithos 302–303 (2018) 298–311

Contents lists available at ScienceDirect

Lithos journal homepage: www.elsevier.com/locate/lithos

Heterogeneous hydrogen distribution in orthopyroxene from veined mantle peridotite (San Carlos, Arizona): Impact of melt-rock interactions Carole M.M. Denis a, Sylvie Demouchy a,⁎, Olivier Alard a,b a b

Géosciences Montpellier, Université Montpellier & CNRS, 34095 Montpellier, France GEMOC National Key Centre, Department of Earth and Planetary Sciences, Macquarie University, Sydney, NSW 2109, Australia

a r t i c l e

i n f o

Article history: Received 7 October 2017 Accepted 11 January 2018 Available online 16 January 2018 Keywords: Hydrogen Peridotite Mantle Ionic diffusion Orthopyroxene

a b s t r a c t Experimental studies have shown that hydrogen embedded as a trace element in mantle mineral structures affects the physical properties of mantle minerals and rocks. Nevertheless, hydrogen concentrations in mantle minerals are much lower than predicted by hydrogen solubilities obtained experimentally at high pressure and temperature. Here, we report textural analyses and major and trace element concentrations (including hydrogen) in upper mantle minerals from a spinel-bearing composite xenolith (dunite and pyroxenite) transported by silica-undersaturated mafic alkaline lavas from the San Carlos volcanic field (Arizona, USA). Our results suggest that the composite xenolith results from the percolation-reaction of a basaltic liquid within dunite channels, and is equilibrated with respect to trace elements. Equilibrium temperatures range between 1011 and 1023 °C. Hydrogen concentrations (expressed in ppm H2O by weight) obtained from unpolarized and polarized Fourier transform infrared spectroscopy are low, with average values b2 ppm H2O, 24 ppm H2O, and 53 ppm H2O for olivine, orthopyroxene, and clinopyroxene, respectively; hydrogen concentrations in olivine are below the detection limit. These low hydrogen concentrations are consistent with depletion by high melt-rock ratio interactions. Clinopyroxene hydrogen concentrations are homogeneous, whereas polarized infrared profile measurements across single-crystals of orthopyroxene reveal hydrogen-depleted rims, which are interpreted as the result of dehydration by ionic diffusion, possibly triggered by melt-rock interactions. We conclude that pyroxenes, like olivine, are unreliable hydrogen proxies, and that the remaining hydrogen concentrations observed in peridotites might only represent the ‘tip of the iceberg’ of the water stored in the Earth's upper mantle. © 2018 Elsevier B.V. All rights reserved.

1. Introduction Experimental studies have shown that a significant part of Earth's water can be stored as hydrogen dissolved as point defects in nominally anhydrous minerals (NAMs) in the mantle, such as olivine, pyroxene, and garnet. Measurements of hydrogen concentrations have indeed confirmed that NAMs from mantle-derived rocks such as peridotite transported by lavas or tectonically exhumed (e.g., Bell and Rossman, 1992; Demouchy and Bolfan-Casanova, 2016; Ingrin and Skogby, 2000; Peslier, 2010), contain a non-negligible (ppm H2O wt level) amount of hydrogen trapped in their atomic structures. Recent data compilations (e.g., Demouchy and Bolfan-Casanova, 2016; Peslier et al., 2017; Xia et al., 2017) report concentration ranges of 0–220 ppm H2O wt for olivine (Ol), 10–300 ppm H2O wt for orthopyroxene (Opx), and 20–640 ppm H2O wt for clinopyroxene (Cpx) using mineral-dependent IR calibrations by Bell et al. (1995) and Withers et al. (2012). Recent studies on diamond mineral inclusions have also reported the occurrence of hydrogen in NAMs, confirming that the deep Earth is indeed a ⁎ Corresponding author. E-mail address: [email protected] (S. Demouchy).

https://doi.org/10.1016/j.lithos.2018.01.007 0024-4937/© 2018 Elsevier B.V. All rights reserved.

non-negligible part of the water cycle (e.g., Jean et al., 2016; Novella et al., 2015; Pearson et al., 2015; Taylor et al., 2016). From mantle xenoliths, the maximum hydrogen concentration in olivine increases with increasing depth (almost linearly down to 250 km), but a similar trend is not observed for pyroxenes. Using average NAM concentrations and averaged mineral modes, the resulting bulk water content of the peridotitic upper mantle is around 74 ppm H2O wt in the spinel stability field (32–80 km depth) and 43 ppm H2O wt in the garnet stability field (80–230 km depth; dataset available in Demouchy and Bolfan-Casanova, 2016). On average, this yields an upper mantle concentration of 60 ppm H2O wt, which equals 3.6 × 1016 tons of water or 1/40 of the current ocean mass. To comprehend if the water observed in upper mantle rocks and minerals is representative of the mantle or only the ‘tip of the iceberg’, we must consider the mobility of hydrogen and the mantle processes impacting hydrogen distribution. Therefore, we must first consider hydrogen ionic diffusivity in mantle minerals at pressure and temperature conditions relevant to the upper mantle. To date, laboratory measurements on NAMs with chemical compositions close to mantle mineral compositions indicate a rapid mobility of hydrogen in NAMs, with diffusivities on the order of 6 × 10−12 to 4 × 10−10 m2/s at 1100 °C in iron-

C.M.M. Denis et al. / Lithos 302–303 (2018) 298–311

bearing olivine, and 10−13 m2/s at 1100 °C in iron-bearing pyroxenes (Demouchy and Mackwell, 2003, 2006; Du Frane and Tyburczy, 2012; Ferriss et al., 2016; Mackwell and Kohlstedt, 1990; Novella et al., 2017; Stalder et al., 2007; Stalder and Behrens, 2006; Stalder and Skogby, 2003). For grains with diameters ranging from 200 μm (such as Cpx) to 2 mm (as for Ol) in mantle-derived peridotites, the diffusivities listed above predict that a peridotite should be hydrogen-poor (b10 ppm H2O wt) after 5 days at 1100 °C (typical of basalts; see discussion in Thoraval and Demouchy, 2014) in a hydrogen under-saturated system. Since clinopyroxenes accurately record trace element concentration modifications during metasomatism and partial melting, pyroxenes have also been considered a reliable proxy of mantle hydrogen concentrations (e.g., Demouchy and Bolfan-Casanova, 2016; Peslier, 2010; Xia et al., 2010). Hydrogen in olivine, on the other hand, is known to be easily affected by mantle processes such as melt percolation (e.g., Satsukawa et al., 2017), magma ascent (e.g., Demouchy et al., 2006; Denis et al., 2013; Lloyd et al., 2012, 2016; Peslier and Luhr, 2006), and metasomatism (Demouchy et al., 2015; Denis et al., 2015; Hao et al., 2016; Jean et al., 2016; Novella et al., 2015; Peslier et al., 2012; Satsukawa et al., 2017; Taylor et al., 2016; Yu et al., 2010). A recent study by Tian et al. (2016) has questioned the utility of Opx as a proxy for mantle hydrogen concentrations as they reported, for the first time, dehydration profiles in mantle Opx from Tianchang volcano (Eastern China). This observation is corroborated by a recent experimental study, inferring very rapid hydrogen concentration modifications (intake 3 times more than the starting concentration in only 10 h) in co-existing Opx and Cpx at subsolidus conditions (i.e., 3 GPa and 1100 °C; Demouchy et al., 2017). Both studies have pointed out that the mechanisms of hydrogen preservation in NAMs from mantle peridotites remain poorly understood and need further consideration and quantification. To decipher the reality of dehydration in natural mantle specimens, we need not only to quantify hydrogen concentrations, but also to fully establish the petro-geochemical history of the mantle rock in question. Here, we report major and trace element concentrations, textural evidence, and hydrogen concentrations in upper mantle minerals from a composite spinel-bearing peridotite that was transported by silica undersaturated mafic alkaline magmas of the San Carlos Volcanic Field (Arizona, USA). Our aim is to further infer the mobility of hydrogen in typical upper mantle rocks that have experienced two essential mantle processes: melt-rock interactions and transport in an ascending magma. 2. Analytical methods Textural characterizations and crystallographic orientations were obtained from scanning electron microscopy (SEM) and electron backscatter diffraction (EBSD). The concentration and distribution of hydrogen was quantified using Fourier transform infrared spectroscopy (FTIR). The major element composition of each mineral was analyzed by electron probe micro-analyses (EPMA) and their trace element compositions by laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS). SEM-EBSD, EPMA, and LA-ICP-MS analyses were performed on a first sample section, while FTIR analyses were performed on a second mirrored thick section. Analyses were performed in a specific order: (i) carbon-coating and EPMA; (ii) coating-removal, final polishing, and SEM-EBSD; (iii) reshaping of the sample section and LA-ICP-MS; and (iv) glue removal and FTIR on the mirrored thick section. 2.1. Sample description and preparation Sample 14SC49 is from the San Carlos Volcanic Field (Arizona, USA), a Neogene monogenetic field of about 50 km2 (e.g., Frey and Prinz, 1978). The locality of San Carlos is particularly known for its abundant large coarse-grained peridotite xenoliths, often containing centrimetric

299

gem-quality peridots with very homogeneous major element compositions (Frey and Prinz, 1978). The specimen was obtained from the Géosciences Montpellier (University of Montpellier, France) peridotite collection. Two sections of the peridotite, away from weathered surfaces or enclosing basalt, were cut for analyses. The first section was singly polished to 100 μm thickness for mineral major and trace element compositional analyses. After EPMA and carbon-coating removal by grinding the surface with 0.5 μm diamond paste, the same section was used for SEM imaging and EBSD analyses. For EBSD, a final polish with colloidal silica was necessary to remove surface damage from the mechanical polishing to allow higher quality phase identification by electron diffraction (i.e., Kikuchi line resolution; Kikuchi, 1928). The second section was double-polished to 500 μm thickness for FTIR analysis. 2.2. Major element analysis by EPMA Mineral major element contents were analyzed using a Cameca SX100 EPMA at the Microsonde Sud facility (University of Montpellier, France). After carbon-coating, compositional analyses were performed with an accelerating voltage of 20 kV, 1 μm spot size, and on peak and background counting times of 20 s for Fe, Mn, and Ni, and 30 s for other elements. Standards used for calibration were forsterite (Mg), Fe2O3 (Fe), Ni metal (Ni), chromite (Cr), Al2O3 (Al), wollastonite (Si, Ca), albite (Na), TiO2 (Ti), orthoclase (K), and pure Mn metal (Mn). Equilibrium temperatures were calculated using the three geothermometers from Brey and Köhler (1990): TB&K, based on Na partitioning between Opx and Cpx; TCa-in-Opx, based on the Ca content of Opx; and TBKN, based on Fe-Mg exchange between Opx and Cpx. 2.3. Imaging and textural analysis by SEM & EBSD The EBSD system is coupled to the JEOL JSM 5600 SEM at the Géosciences Montpellier microscopy facility (University of Montpellier, France). The sample section was tilted at 70° to the horizontal and surrounded by double-faced C-Cu conductive tape to reduce electrical overcharge on the sample. Analyses were performed with an accelerating voltage of 17 kV and a working distance of 24 mm to further avoid sample charging. We obtained crystallographic orientation maps covering almost the entire section (31 mm long and 17 mm wide) with sampling steps of 30 μm (i.e., sampling steps must be at least 1/4 of the average grain size). Data were processed and indexed by the Oxford Instruments HKL CHANNEL5 software, and statistics were obtained using the MTEX MATLAB toolbox (Hielscher and Schaeben, 2008; Mainprice et al., 2014). Indexation rates were lower for pyroxenes than for olivine (see Soustelle et al., 2010 for an example). Due to the irregular sample shape and cracks present in the section, the total indexation rate was around 76%. Post-acquisition data treatments allowed us to further increase the indexation rate by (i) filling the non-indexed pixels, which have up to 8 identical neighbors with this orientation, (ii) repeating this operation using 7 and 6 identical neighbors, (iii) identifying grains as continuous domains characterized by an internal misorientation b10°, and (iv) within each olivine crystal, searching and correcting for systematic indexation errors due to the olivine hexagonal pseudosymmetry, which results in similar diffraction patterns for orientations differing by a rotation of 60° around [100]. Crystal preferred orientation (CPO) data are displayed as pole figures, with lower hemisphere stereographic projections. All CPOs were reoriented to have the maximum concentrations of olivine [100] and [010] axes parallel to the E–W and N–S directions, respectively. This allows a straightforward comparison of CPOs between the different samples and with previously published textures for mantle peridotites. Data are plotted as one point per grain to prevent over-representation by large or small grains. The strength of the fabric is quantified using the dimensionless J-index (i.e., the volume-averaged integral of the squared orientation densities; Bunge, 1982; Mainprice and Humbert, 1994; Mainprice and Silver,

300

C.M.M. Denis et al. / Lithos 302–303 (2018) 298–311

1993): the J-index has a value of 1 for a perfectly random CPO and tends toward infinity for a single crystal.

(e.g., Kohlstedt et al., 1996); and the normalized integrated area (absorption) are also reported for application to future IR mineraldependent calibrations.

2.4. Trace element analyses by LA-ICP-MS 3. Results The same section used for the above analyses was cut into a square slab to fit the analytical cell of the spectrometer. Trace elements in olivine, orthopyroxene, and clinopyroxene were analyzed by in situ LA-ICPMS at Géosciences Montpellier (France), using a ThermoFinnigan ELEMENT XR (eXtended Range) high resolution (HR) ICP-MS system. The ICP-MS is coupled with a Geolas (Microlas) automated platform, housing a LambdaPhysik ArF 193 nm Compex 102 laser. Measurements were conducted in a He atmosphere, which enhances measurement sensitivity and reduces inter-element fractionations. The He gas stream and particles from the sample were mixed with Ar before entering the plasma. Signals were measured in time resolved acquisition mode, devoting 3 min for the blank and 1 min for mineral analysis. The laser was fired using an energy density of 15 J/cm2 at a frequency of 10 Hz and using a spot size of 77 μm. Oxide levels, measured using the UO/U ratio, were below 0.7%. Standard BIR-1G was included during analytical runs and our measurements are in good agreement with working values for this international standard (Jochum et al., 2005). Concentrations were calibrated against standard glass NIST 612 using the values given by Pearce et al. (1997). Data were subsequently reduced using the GLITTER software (Griffin et al., 2008) and time-resolved analyses were carefully inspected to check for homogeneity in the analyzed mineral volume. Several trace element concentrations in olivine were below detection limits (i.e., Rb, Sr, Cs, Ba, U, La, Nb, Eu, and Dy). 2.5. Fourier transform infrared spectroscopy Hydroxyl groups in the minerals were detected using transmission FTIR spectroscopy at the Laboratoire Charles Coulomb (University of Montpellier, France). Unpolarized and polarized infrared spectra were acquired using a Bruker IFS66v, equipped with a liquid nitrogencooled MCT detector (Mercatel alloy, HgCdTe), a KBr/Ge beam splitter, and coupled to a Bruker HYPERION microscope. The infrared beam were polarized with a KRS-5 polarizer (Tl-bromoiodiode) placed between the mid-IR source and the sample. Unpolarized IR measurements were performed following the protocol of Denis et al. (2015): optically clean areas were chosen for analyses, and over 200 scans were accumulated at a resolution of 4 cm−1. A square aperture (50 × 50 μm or 100 × 100 μm) was used. Each spectrum was baseline corrected and the absorbance was normalized to 1 cm thickness. Since sample 14SC49 is a spinel-bearing peridotite, we have chosen to cut a thick slab (~500 μm) to maximize detection limits for olivine, while keeping the pyroxenes transparent enough for transmission FTIR analyses. We used two types of IR calibrations to calculate hydrogen concentrations: the frequency-dependent calibration of Paterson (1982) and the mineral-dependent calibrations of Bell et al. (1995) for pyroxenes and Withers et al. (2012) for olivine. For the frequency-dependent calibration, the following mineral specific factors χi (Paterson, 1982) were used: χol = 2678 ppm H2O wt; χOpx = 2425 ppm H2O wt; and χCpx = 2731 ppm H2O wt. Hydrogen concentrations were calculated by integrating each spectrum over the wavenumber ranges 3620– 3050 cm−1 for Ol, 3670–3050 cm−1 for Opx, and 3770–3050 cm−1 for Cpx. This calibration allows a detection limit of about 1 ppm H2O wt for a 1-mm-thick olivine sample. The estimated error from the empirical calibration in the resulting H concentration is around 30% (Paterson, 1982; Rauch, 2000). Hydrogen concentrations were obtained from averaging at least 10 unpolarized spectra from each grain. The maximum linear absorption of non-normalized spectra did not exceed 0.3, in agreement with the recommendations of Withers et al. (2012) for unpolarized FTIR measurements of olivine. Finally, conversion factors to change concentrations from ppm H2O by weight to atomic H per 106 atoms of Si (at. H/106 Si = at. ppm H/Si), as historically used

3.1. Petrography Sample 14SC49 is a composite spinel-bearing xenolith comprising a central dunite section between two pyroxene-rich units as seen in Fig. 1. The dunite has a coarse-granular texture (Mercier and Nicolas, 1975) and is composed almost exclusively of olivine (99%, Table 1). In the macroscale sample, olivine crystals are up to 15 mm long. Under optical microscopy, olivines contain sub-grain boundaries and triple junctions. Both sides of the dunite show relatively plan-parallel contacts with the pyroxenites (Fig. 1). The pyroxenite units contain Ol (36%), Cpx (34%), Opx (28%), and spinel (2%) (Table 1) and are classified as olivine websterite (Streckeisen, 1976); neither amphibole nor plagioclase were observed. Average grain sizes (i.e., equivalent radii) are 420, 223, and 141 μm for Ol, Opx, and Cpx, respectively. The smallest grain of each mineral phase is around 100 μm, whereas the largest grains are 14.20, 4.436, and 2.679 mm for Ol, Opx, and Cpx, respectively. Olivine grains in the pyroxenites are smaller (≪1 mm) than in the dunite (often ≫1 mm). Several studies have reported the occurrence of thin melt/glass films or glass pockets along grain boundaries or at triple junctions in San Carlos peridotites (Frey and Prinz, 1978; Wirth, 1996). Here, despite careful investigation via optical microscopy and SEM imaging, we were unable to identify such glass occurrences in either lithology. Grain boundaries are mostly straight, but curved inter-mineral interfaces also occur. Mineral maps from EBSD reveal interpenetration of pyroxenes along the dunite portion (corrosion and tortuous boundaries), advocating the secondary character of the pyroxenites compared to the dunite as shown in Fig. 1b. EBSD maps allowed us to calculate the CPO of each phase, which are displayed in Fig. 2. Olivine shows a clear CPO, with well-developed axial-[100] patterns characterized by an important point concentration of the [100] axes. [010] axes are distributed in a girdle distribution normal to the [100] maximum, whereas [001] axes are disseminated. This olivine texture is typical of lithospheric mantle domains (e.g., Ben Ismail and Mainprice, 1998). Pyroxenes display more dispersed CPOs, even when N100 grains were measured, and it is difficult to distinguish a clear Opx CPO pattern since the pole figures display multiple weak maxima (Fig. 2). The diopside CPO is more defined than for Opx, with a stronger axis concentration normal to the (010) planes. The J-indexes (Bunge, 1982; Mainprice and Humbert, 1994; Mainprice and Silver, 1993) are 2.62, 2.58, and 3.26 for Ol, Opx, and Cpx, respectively; these values are relatively low for mantle-derived rocks, as J-indexes for typical peridotitic olivines usually range from 2 to 14 with an average around 6 (e.g., Ben Ismail and Mainprice, 1998; Tommasi et al., 2000). If olivine and pyroxene were deformed together, the [100]olivine and [001]pyroxene CPO distributions should be similar, which is not the case here. 3.2. Mineral compositions Selected mineral major element concentrations are reported in Table 2. To test for any potential spatial distributions, the major element compositions of the constituent minerals in both units, dunite and pyroxenites, are presented in Fig. 3 as a function of the distance to the contact between the two lithologies. Olivine major element compositions are extremely homogeneous in the dunite (e.g., 90.5 ± 0.1 Fo% and 0.39 ± 0.02 wt% NiO). Olivines in the pyroxenite unit have the same chemical characteristics as those in the dunite (e.g., 90.5 ± 0.1 Fo% and 0.40 ± 0.02 wt% NiO). Overall, pyroxene compositions are quite homogeneous at the sample scale, i.e., across both the dunite and websterite units. However, we note that we were only able to analyze

C.M.M. Denis et al. / Lithos 302–303 (2018) 298–311

(a)

(b)

5mm

Opx

Cpx

301

0.5mm Olivine Opx Cpx C

Ol

Fig. 1. (a) Thin section microphotograph of a doubly polished section of 14SC49. (b) EBSD map showing the different mineral phases present in the composite xenolith, and highlighting corrosion of the dunite by the websterite.

one grain of Opx and one grain of Cpx in the dunite section due to their scarcity in that lithology. Orthopyroxene is enstatite with 3.78 ± 0.07 wt% Al2O3 and 0.87 ± 0.03 wt% CaO. Core-to-rim zonations are extremely weak to non-existent: rims have average Al2O3 and CaO contents of 3.82 ± 0.06 and 0.86 ± 0.02 wt%, respectively, compared to core compositions of 3.73 ± 0.05 and 0.87 ± 0.03 wt%, respectively. Clinopyroxene is Cr-diopside (Cr2O3 = 1.32 ± 0.09 wt%) clearly indicating that this sample is a Group I xenolith (Cr-diopside of mantle origin) and not a Group II xenolith, which are characterized by Alaugite and are more akin to a magmatic cumulate origin (Frey and Prinz, 1978; Wilshire and Shervais, 1975). Core-to-rim zoning in Cpx is very subtle, with rim CaO, Al2O3, and NaO contents of 20.8 ± 0.2, 5.07 ± 0.08, and 1.22 ± 0.05 wt%, respectively, within error of the core compositions (CaO = 20.9 ± 0.2 wt%, Al2O3 = 5.2 ± 0.1 wt%, NaO = 1.21 ± 0.05 wt%). Spinels in the websterite section are extremely homogeneous with, on average, Mg# = 74 ± 2 and Cr# = 27 ± 1 (Table 2; Mg# = 100 × Mg/(Mg + Fetot); Cr# = 100 × Cr/(Cr + Al)). Equilibrium temperatures calculated using the formulations of Brey and Köhler (1990) are reported in Table 1, and yield values between 1011 and 1023 °C. No systematic variations are observed between cores and rims or as a function of the position of the Opx-Cpx pair

Table 1 Mineral modes (ol, olivine; opx, orthopyroxene; cpx, clinopyroxene; sp., spinel) in the dunite and websterite lithologies obtained by EBSD. Equilibrium temperatures of the websterite were calculated using the geothermometers of Brey and Köhler (1990): ‘B&K’, based on Na partitioning between Opx and Cpx; ‘BKN’, based on Fe-Mg exchange between Opx and Cpx; and ‘Ca-in-Opx’ based on calcium in orthopyroxenes (see main text for details). Mineral modal (%)

Dunite Websterite

ol

opx

cpx

sp

99 36

b0.4 28

b0.4 34

b0.2 2

BKN 1023 4

Ca-in-Opx 1022 6

Equilibrium temperature Websterite B&K (°C) 1011 ±1sd 4

relative to the dunite-websterite contact. Such equilibrium temperatures are common for all San Carlos mantle xenoliths, irrespective of lithology (1038 ± 50 °C, Frey and Prinz, 1978; 1027 ± 46 °C, Galer and O'Nions, 1989; 1044 ± 27 °C, Macris et al., 2015). Trace element concentrations in Cpx, Opx, and Ol are reported in Table 3. Each mineral shows homogeneous traces element concentrations across both lithologies, and no core-to-rim zonations were observed. Rare earth element (REE, normalized to chondrite, N; McDonough and Sun, 1995) and extended trace element (normalized to primitive mantle, PM) patterns are shown in Fig. 4. Clinopyroxenes contain 1.17 ± 0.07 ppm Yb (i.e., (Yb)N ≈ 7.0 ± 0.4). The Cpx REE patterns are strongly depleted in Light-REE (LREE) relative to MiddleREE (MREE) and Heavy-REE (HREE) (e.g., (La/Sm)N = 0.016 ± 0.002) and are relatively flat from MREE to HREE ((Sm/Lu)N = 0.79 ± 0.04; Fig. 4a). Orthopyroxenes contain about 0.21 ± 0.01 ppm Yb (i.e., (Yb)N ≈ 1.2 ± 0.1), and their chondrite-normalized REE patterns are characterized by a steep and continuous depletion from HREE to LREE ((Ce/Sm)N = 0.012 ± 0.007; (Sm/Lu)N = 0.18 ± 0.04). Significant deviations are only observed for La (±Ce), and La shows occasional small enrichments relative to Ce. Calculated REE equilibrium temperatures between Cpx and Opx following Liang et al. (2012) are 1053 ± 15 °C, similar within errors to the major element-based geothermometer values (Table 1). Olivine REE concentrations are very low and are often close to detection limits (Fig. 4a), hence the relative standard deviation associated with those values is large, 20–50%; for example, Yb concentrations are about 0.013 ± 0.003 ppm (i.e., (Yb)N = 0.08 ± 0.02). Olivine REE patterns are characterized by a sharp decrease from HREE to MREE ((Sm/Lu)N = 0.07 ± 0.02) and from MREE to LREE ((Ce/Sm)N = 0.23 ± 0.09). As in Opx, Ol shows La enrichment relative to Ce. Within the relative scatter of the data, no difference is observed between dunite and websterite Ol patterns (Fig. 4). To assess if the websterite is at equilibrium, we calculate the HREE and MREE ratios between Ol, Opx, and Cpx. If the uncertainties of (i) the REE concentration in Ol and (ii) the mineral/melt partition coefficient DOl/melt for the REE are taken into account (e.g., Bédard, 2005; Zanetti et al., 2004), the resulting ratios are consistent with experimental and calculated partition coefficients (e.g., Adam and Green, 2006; Liang et al., 2012), indicating that REE concentrations in Ol are in

C.M.M. Denis et al. / Lithos 302–303 (2018) 298–311

[100]

[010]

[001] 4 3.5 3 2.5 2 1.5 1 0.5

Olivine N=520 J=2.63

M.U.D.

302

3

Enstatite N=195 J=2.58

2 1.5

M.U.D.

2.5

1 0.5 3

2

Diopside N=302 J=3.26

1.5

M.U.D.

2.5

1 0.5

Fig. 2. Crystallogrpahic preferred orientations for (a) olivine, (b) orthopyroxene, and (c) clinorpyroxene in peridotite 14SC49. Pole figures are in lower hemisphere equal-area projections. N is the number of analyzed olivine grains. J-indexes are provided, indicating the strength of the rock texture (see main text for details). M.U.D: multiples of a uniform density.

equilibrium with their respective concentrations in Opx and Cpx. However, LREE concentrations do not reflect equilibrium; for instance, the mineral/mineral distribution coefficients KdOl/Opx(La, Ce) converge toward a value of 1, which is not in agreement with experimental partition coefficients. This apparent disequilibrium is common in mantle peridotites and is ascribed to the incorporation of sub-percent amounts of highly incompatible element-enriched fluids by solid micro- to nanophases (e.g., Garrido et al., 2000). Clinopyroxene extended trace element patterns (Fig. 4b) show pronounced U and Th enrichments relative to LREEs (e.g., (U/Ce)PM = 9.2 ± 1), and U is fractionated relative to Th (2.9 ≤ (U/Th)PM ≤ 7.3). Compared to Cpx, Opx and Ol show even more pronounced U enrichments relative to LREEs ((U/Ce)PM = 38 ± 18 and 26 ± 7, respectively) and stronger U/Th fractionations ((U/Th)PM = 9.3 ± 4.7 and 5.9 ± 1.6, respectively). Such strong U/Th fractionations are relatively common in alkali basalt-embedded xenoliths, and U enrichment is rather ascribed to the percolation of oxidized volatile-rich fluids (Alard et al., 1996, 2011). In Cpx, high field strength elements (HFSE: Nb, Ta, Zr, Hf, and Ti), except Nb (see below), show negative anomalies relative to their neighboring REEs (e.g., (Hf/Sm)PM = 0.56 ± 0.06 and (Ti/Gd)PM = 0.67 ± 0.02). Conversely, HFSE concentrations in Opx show positive anomalies relative to the same REEs ((Hf/Sm)PM = 2.9 ± 0.8 and (Ti/Gd)PM = 7.7 ± 1.1). Such inverse HFSE behaviour between Cpx and Opx is well-documented and is due to the higher compatibility of HFSE elements relative to their neighboring REEs in Opx, whereas HFSE partition coefficients for Cpx are lower than for the REEs (Kelemen et al., 1993; Rampone et al., 1991). Olivine shows positive HFSE anomalies with average (Hf/Sm)PM and (Ti/Gd)PM values of 1.7 ± 0.9 (similar within error

to the Opx fractionation) and 3.4 ± 0.9 (i.e., half of the Opx fractionation), respectively. However, Nb displays a rather unusual and unexpected behaviour relative to LREEs and other HFSEs, especially relative to its geochemical sibling Ta. Indeed, while Ta shows a negative anomaly relative to LREEs, Nb is enriched in Cpx ((Nb/Ce)PM = 3.1 ± 0.4). This leads to an unusual and variable Nb-Ta fractionation in Cpx, with (Nb/Ta)PM ranging between 3.4 and 12.7. This relative enrichment is enhanced in Opx (i.e., (Nb/Ce)PM = 39 ± 10) due to the stronger affinity of HFSEs for the Opx structure. Nevertheless, the Nb/Ta ratio in Opx is closer to unity (i.e., (Nb/Ta)PM = 1.9 ± 0.4). In Ol, (Nb/Ce)PM = 3.9 ± 1.3, similar to Cpx, but here Ta is also enriched ((Ta/Ce)PM = 27 ± 10) leading to a low (Nb/Ta)PM value of 0.15 ± 0.07. Such behaviour has not been observed in other olivine-rich peridotites showing high degrees of re-equilibration. We cannot offer a straightforward explanation to account for such complex, phase specific, but obviously interrelated behaviour of HFSEs between the three co-existing mantle minerals; although intriguing, this unusual HFSE behaviour is beyond the scope of the present study. 3.3. Mineral hydrogen concentrations FTIR absorption spectra are typical of spinel lherzolite minerals (Fig. 5). Our measurements do no reveal significant OH bands in olivine. The main hydroxyl absorption bands in Cpx are located at 3635, 3517, and 3455 cm−1, and those in Opx at 3570, 3520, and 3408 cm−1. For pyroxenes, spectra with maximum and minimum hydroxyl absorbances are shown in Fig. 5. Hydrogen concentrations in Ol, Opx, and Cpx are reported in Table 4. Olivines are considered dry, and pyroxenes contain very small amounts

C.M.M. Denis et al. / Lithos 302–303 (2018) 298–311

303

Table 2 Selected major and minor element compositions (wt%) in the dunite and websterite lithologies determined by EPMA. Mg# = 100 × Mg/(Mg + Fetot); Cr# = 100 × Cr/(Cr + Al). Abbreviations: Ol, olivine; Opx, orthopyroxene; Cpx, clinopyroxene. Standard deviation is provided in italic. Dunite

Websterite

Wt%

Ol av.

±1sd

Ol core

±1sd

Ol rim

±1sd

Cpx av.

±1sd

Opx av.

±1sd

Ol av.

±1sd

Ol core

±1sd

Ol rim

±1sd

SiO2 TiO2 Al2O3 Cr2O3 FeO MnO MgO CaO Na2O K2O NiO Total mg# cr#

41.25 0.01 0.02 0.02 9.05 0.13 48.80 0.07 0.01 0.00 0.40 99.75 90.48 14.17

0.13 0.01 0.01 0.01 0.08 0.02 0.30 0.01 0.01 0.00 0.02 0.31 0.09 0.33

41.28 0.01 0.02 0.02 9.08 0.13 48.74 0.07 0.01 0.00 0.40 99.75 90.46 10.36

0.14 0.01 0.01 0.01 0.07 0.02 0.24 0.01 0.01 0.00 0.02 0.27 0.08 0.09

41.22 0.01 0.02 0.02 9.03 0.12 48.86 0.07 0.01 0.00 0.40 99.75 90.51

0.10 0.01 0.01 0.01 0.07 0.02 0.35 0.01 0.01 0.00 0.02 0.35 0.11

52.55 0.30 5.12 1.26 2.63 0.09 16.13 20.98 1.22 0.00 0.06 100.32 91.62

0.01 0.00 0.06 0.05 0.00 0.00 0.05 0.15 0.03 0.00 0.01 0.25 0.03

56.28 0.09 3.77 0.65 5.79 0.16 33.01 0.87 0.09 0.00 0.11 100.81 91.05

0.45 0.04 0.00 0.01 0.09 0.02 0.06 0.02 0.00 0.00 0.00 0.62 0.12

41.12 0.01 0.02 0.03 9.04 0.13 49.09 0.08 0.01 0.00 0.39 99.91 90.56

0.43 0.01 0.01 0.01 0.04 0.02 0.22 0.01 0.01 0.00 0.02 0.39 0.06

41.16 0.01 0.01 0.02 9.06 0.13 49.14 0.08 0.01 0.00 0.40 100.02 90.56

0.17 0.01 0.01 0.01 0.04 0.02 0.17 0.01 0.01 0.00 0.02 0.20 0.06

41.04 0.01 0.02 0.03 9.01 0.13 48.98 0.09 0.01 0.00 0.38 99.71 90.56

0.73 0.01 0.01 0.01 0.03 0.02 0.27 0.01 0.01 0.00 0.02 0.57 0.07

Websterite Wt%

Opx av.

±1sd

Opx core

±1sd

Opx rim

±1sd

Cpx av.

±1sd

Cpx core

±1sd

Cpx rim

±1sd

Spl core

±1sd

Spl rim

±1sd

SiO2 TiO2 Al2O3 Cr2O3 FeO MnO MgO CaO Na2O K2O NiO Total mg# cr#

55.67 0.09 3.78 0.63 5.71 0.13 33.03 0.87 0.08 0.00 0.11 100.10 91.16 10.04

0.27 0.02 0.07 0.04 0.05 0.02 0.20 0.03 0.02 0.00 0.02 0.31 0.09 0.48

55.68 0.09 3.82 0.64 5.72 0.13 33.03 0.87 0.08 0.00 0.11 100.17 91.15 10.07

0.23 0.02 0.06 0.03 0.04 0.01 0.17 0.03 0.02 0.00 0.01 0.21 0.08 0.35

55.66 0.09 3.73 0.62 5.69 0.13 33.04 0.86 0.08 0.00 0.10 100.01 91.19 9.99

0.33 0.01 0.05 0.04 0.06 0.02 0.25 0.01 0.02 0.00 0.02 0.40 0.09 0.63

52.22 0.30 5.16 1.32 2.60 0.08 16.06 20.85 1.21 0.00 0.05 99.84 91.67 14.61

0.32 0.02 0.11 0.09 0.06 0.01 0.32 0.17 0.05 0.00 0.02 0.34 0.19 0.64

52.25 0.29 5.21 1.33 2.58 0.08 15.94 20.87 1.21 0.00 0.05 99.82 91.67 14.65

0.20 0.02 0.10 0.09 0.06 0.01 0.11 0.16 0.05 0.00 0.02 0.34 0.20 0.68

52.16 0.30 5.07 1.29 2.63 0.08 16.24 20.83 1.22 0.00 0.05 99.88 91.66 14.55

0.46 0.02 0.08 0.07 0.03 0.01 0.45 0.19 0.05 0.00 0.02 0.35 0.19 0.59

0.16 0.18 45.35 24.40 11.61 0.14 18.34 0.01 0.01 0.00 0.29 100.50 74.05 26.52

0.01 0.01 0.79 0.19 0.35 0.02 0.44 0.03 0.01 0.00 0.01 0.77 1.26 0.47

0.19 0.18 45.88 24.26 11.34 0.14 18.40 0.03 0.00 0.00 0.29 100.70 73.97 26.19

0.02 0.02 1.19 0.23 0.46 0.01 0.77 0.04 0.00 0.00 0.02 1.34 2.19 0.66

We have performed a total of 69 points of analyzes for olivine, 40 points for cpx, 26 points for opx and 8 points for spinel. We reported here a representative selection, not averages. The full data set is available upon request.

(a)

(b)

Websterite 58

Opx

Websterite

Dunite 54

Cpx

44

2.8 2.6 2.4 2.2 5.4 5.2 5.0

FeO

4.0 3.8 3.6 3.4

Al2O3

0.8 0.6 0.4

Cr2O3

Olivine

40 9.5

FeO

Dunite SiO2

42

52

54 6.5 6.0 5.5

Websterite

Dunite

SiO2

53

SiO2

56

(c)

FeO

9.0 8.5

Al2O3

1.6 1.4 1.2

Cr2O3

0.08 0.06 0.04 0.02

Cr2O3

0.10

CaO

1.0 0.9 0.8 0.3 0.2 0.1 0.0 -1.0

CaO

0.05

21.0

CaO

20.5 0.4

TiO2 -0.5

0.0 0.5 Distance [cm]

TiO2

0.2 1.0

0.0 -1.0

0.06 0.04 0.02 0.00 -1.0

-0.5

0.0 0.5 Distance [cm]

1.0

EMPA

TiO2 -0.5

0.0 0.5 Distance [cm]

core

1.0

rim LA-ICMP

Fig. 3. Mineral major and minor element contents (in wt%) and distributions in the composite xenolith (peridotite 14SC49) as a function of the distance to the contact between the two lithologies (displayed as the vectical dashed line), for (a) orthopyroxene (Opx), (b) clinopyroxene (Cpx), and (c) olivine. The dashed horizontal lines indicate the EPMA detection limits for Cr and Ti in olivine.

304

C.M.M. Denis et al. / Lithos 302–303 (2018) 298–311

Table 3 Trace element concentrations measured by LA-ICP-MS in olivine, orthopyroxene, and clinopyroxene in ppm (n is the number of the measurements, sd the standard deviation, and TDL-Ol the typical detection limit for measurements in olivine). Standard deviation is provided in italic. Olivine Dunite

Websterite

Core & Rim

Dunite & Websterite

Core & Rim

av.

Orthopyroxene

Clinopyroxene

Websterite

Websterite

av.

n

6

±1sd

5

±1sd

11

±1sd

TDL-ol

12

ppm Sc Ti Ni Rb Sr Y Zr Nb Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta Pb Th U

3.12 17.73 2884 b0.006 b0.0040 0.0274 0.0278 0.00144 0.0069 0.00086 0.00105 0.00026 0.00192 0.00108 0.00043 0.00197 0.00040 0.00296 0.00112 0.0052 0.00152 0.0140 0.00360 0.00166 0.00068 0.0230 0.000250 0.000330

0.11 0.68 25

3.18 17.94 2876 b0.004 b0.0040 0.0275 0.0220 0.00160 0.0064 0.00071 0.00105 0.00030 0.00223 0.00121 0.00050 0.00235 0.00046 0.00295 0.00128 0.0057 0.00137 0.0134 0.00310 0.00139 0.00047 0.0222 0.000170 0.000308

0.10 0.96 36

3.16 17.92 2889 b0.006 b0.0040 0.0279 0.0249 0.00151 0.0066 0.00079 0.00105 0.00028 0.00208 0.00116 0.00046 0.00219 0.00043 0.00296 0.00120 0.0054 0.00145 0.0137 0.00338 0.00154 0.00059 0.0227 0.000218 0.000316

0.11 0.80 42

0.014 0.027 0.052 0.0052 0.0056 0.0260 0.0285 0.00089 0.0052 0.00085 0.00126 0.00031 0.00209 0.00098 0.00045 0.00288 0.00038 0.0019 0.00059 0.0045 0.00144 0.0138 0.00278 0.00095 0.00063 0.014 0.00023 0.00026

n.a. n.a. 750.270 b0.005 0.017 0.854 0.360 0.0203 b0.006 0.00098 0.00121 0.00082 0.0080 0.0106 0.00704 0.0350 0.0099 0.1059 0.854 0.0306 0.119 0.0253 0.0367 0.0204 0.00072 0.044 0.00023 0.00064

0.0030 0.0090 0.00034 0.0030 0.00025 0.00021 0.00007 0.00050 0.00024 0.00011 0.00017 0.00004 0.00094 0.00018 0.0013 0.00020 0.0031 0.00095 0.00052 0.00019 0.0109 0.000044 0.000062

0.0032 0.0025 0.00039 0.0021 0.00009 0.00005 0.00005 0.00056 0.00035 0.00010 0.00060 0.00008 0.00086 0.00021 0.0015 0.00038 0.0027 0.00082 0.00038 0.00005 0.0054 0.000029 0.000075

of H. Using the mineral-dependent IR calibrations, Ol, Opx, and Cpx contain 0.7–1.4, 24, and 53 ppm H2O wt, respectively. At first, olivines in the websterite seem to contain more hydrogen (1.4 ppm H2O wt) compared to those in the dunite (0.7 ppm H2O wt), but we reiterate that the IR spectra show no detectable OH bands (Fig. 5, Table 4); thus, these concentrations are effectively below the detection limit (∼1 ppm H2O wt). In both lithologies, Cpx hydrogen concentrations are homogeneous, but Opx in the websterite display notable core-to-rim variations with depleted rims (Fig. 6a,b). Unpolarized and polarized FTIR profile measurements are presented in Fig. 7, which show that inter-grain Ol and Cpx hydrogen concentrations are homogeneous (Fig. 7a,b), whereas concentrations in Opx are heterogeneous along some profiles, but not every path (Figs. 7c, 8, see also Fig. S1). Further investigation by polarized FTIR, with various orientations of the polarizer, are compiled in Fig. 8, and evidence that Opx have heterogeneous rim-depleted hydrogen distributions (Figs. 6, 8) at higher concentrations, but again some paths reveal almost homogeneous concentrations when the concentration is low (Fig. 8a). This attests to the necessity of investigating FTIR profiles in different directions (three orthogonal directions when possible) and at different orientations of the polarizer to accurately probe the hydrogen distribution in each grain. Fig. 8a–c shows that hydrogen depletion in Opx is marked in the major OH bands (3570–3520 cm−1), while the minor OH band (3408 cm−1) seems significantly less affected by absorbance diminution (Figs. 5, 8). 4. Discussion 4.1. Compositional variations and petrogenesis San Carlos dunites and websterites have been described in previous studies, notably Frey and Prinz (1978), Galer and O'Nions (1989),

0.0031 0.0070 0.00036 0.0025 0.00019 0.00015 0.00006 0.00052 0.00029 0.00011 0.00048 0.00007 0.00086 0.00020 0.0013 0.00029 0.0028 0.00088 0.00046 0.00018 0.0087 0.000053 0.000062

av. ±1sd

17.727 0.007 0.061 0.028 0.0010 0.00013 0.00027 0.00016 0.0024 0.0025 0.00082 0.0059 0.0012 0.0089 0.061 0.0018 0.012 0.0036 0.0035 0.0061 0.00015 0.011 0.00009 0.00031

14

±1sd

69.72 2039.42 368.45 b0.01 9.09 11.43 5.82 0.276 0.0198 0.0187 0.2177 0.0991 1.037 0.774 0.360 1.401 0.277 2.024 0.441 1.258 0.180 1.173 0.1624 0.299 0.00187 0.0572 0.0270 0.0238

2.56 43.87 5.02 0.10 0.50 0.22 0.039 0.0078 0.0020 0.0079 0.0045 0.043 0.034 0.009 0.052 0.017 0.077 0.017 0.054 0.013 0.071 0.0078 0.028 0.00061 0.0057 0.0059 0.0024

Wilshire and Shervais (1975), and Zindler and Jagoutz (1988). Composite xenoliths are common in San Carlos; for instance, Frey and Prinz (1978) reported olivine-rich lenses completely surrounded by pyroxene-rich units with gradational contacts. PA-10, the dunite studied by Frey and Prinz (1978) is broadly similar to that in our xenolith (theirs noticeably differing only by a 2 wt% enrichment in Al2O3 and TiO2 in pyroxenes). However, their websterite (PA-65B, 80% Opx) is markedly different, and the websterite in our sample is closer in modal composition to their pyroxene-rich spinel lherzolith (PA-15A, 35% Opx). Note that samples PA-10 and PA-15A were further studied by Zindler and Jagoutz (1988). The lack of correlation between the [100]olivine and [001]pyroxene CPOs (see examples of correlation in Frets et al., 2014; Zaffarana et al., 2014), and the corrosion by pyroxenes and tortuosity of the dunite edges, indicate that the dunite was likely infiltrated by the pyroxenites, with the latter resulting from crystallization of a basalt-like (sensu lato) percolating melt. The olivine CPO, although slightly weaker than usual, remains characteristic of deformation within the lithospheric mantle domain (axial-[100]; cf., Ben Ismail and Mainprice, 1998). Therefore, the petrographic and microtextural information lead us to consider a scenario involving at least two stages; stage-1 is the formation and subsequent lithospheric deformation of the dunite, and stage-2 is involving the formation of the Opx-Cpx ± Ol assemblage. The lack of correlation between olivine and pyroxene CPOs in the two lithologies suggests that the websterite was formed by the addition of pyroxene at the expense of the dunite-like assemblage. The composite nature and microstructure (indicating at least a twostage petrogenesis) of the peridotite seem to be in agreement with previous isotopic studies on San Carlos peridotites, which have revealed a number of disequilibria mainly between olivine and pyroxenes (δ18O, Gregory and Taylor, 1986; δ25,26Mg, Young et al., 2009; and δ57Fe, Macris et al., 2015). Despite these observations, the studied composite

C.M.M. Denis et al. / Lithos 302–303 (2018) 298–311

10

0

cpx (n=14)

Concentration / Chondrite

10 1

opx (n=12)

0.1

olivine

1

01

00

1

La Ce Pr Nd

Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Below detection limit

Concentration / Primitive mantle

10

1 0.1 0.0

1

0.0

0.0

01

00

1

BaTh U NbTa La Ce Pr Sr Nd

SmZr Hf Eu Ti GdTbDyHo Y ErTmYbLu

Fig. 4. (a) Rare earth (normalized to chondrite) and (b) extended trace element (normalized to primitive mantle; McDonough and Sun, 1995) compositions of olivines, orthopyroxenes (Opx), and clinopyroxenes (Cpx) from peridotite 14SC49. The hollow star symbol denotes values for olivine in the dunite. Arrows indicate concentrations below the detection limit (see Table 3 for details).

Fig. 5. Representative unpolarized infrared spectra of each mineral in the peridotite 14SC49. The spectra with minimum and maximum absorbance are given for orthopyroxene (Opx) and clinopyroxene (Cpx). All spectra are normalized to 1 cm thickness.

Ol

Opx

Cpx

Dunite

Websterite

Websterite

Websterite

n

15

4

28

26

Norm. Int. Abs. min P82 max P82 COH av. P82 ±1sd COH av. W12 ±1sd COH av. B95 ±1sd Factor*

2.1 0.3 0.5 0.4 0.3 0.7 0.1 – – 16.29

3.8 0.3 0.7 0.6 0.1 1.4 0.1 – – 16.29

125.5 23 35 26 4.4 – – 24 4.7 11.53

125.0 18 39 30 7.3 – – 53 15.6 12.1

(n=12)

0.0 0.0

Table 4 Average integrated absorbances normalized to sample thickness (Norm. Int. Abs.) and hydrogen concentrations (ppm H2O wt) in olivine (Ol), orthopyroxene (Opx), and clinopyroxene (Cpx) based on the frequency-dependent calibration of Paterson (1982; ‘P82’) and based on the mineral-dependent calibrations of Withers et al. (2012; ‘W12’) and Bell et al. (1995; ‘B95’). Hydrogen concentrations are reported as the average (av.) of n grains; ±1sd represents the absolute uncertainty calculated by the algebraic method; ‘Factor’ denotes the value used to convert water contents in ppm H2O wt to H/106 Si. Standard deviation is provided in italic. 14SC49

0.0

305

peridotite shows striking and robust features of chemical equilibrium between Cpx, Opx, and Ol, both in terms of major and trace elements. We emphasize: (i) the Mg# of Opx and Cpx are consistent with Ol Fo %, despite the drastic differences in modal abundances between the websterite and dunite sections; (ii) there is no compositional differences in olivines from the websterite and the olivines from the pyroxenite (e.g., in contrary to formation of Mg-depleted secondary olivine previously reported in other studies, e.g., Rampone et al., 2004; Rampone and Borghini, 2008). (iii) the excellent consistency between major element and REE geothermometry, which is rarely seen in mantle peridotites, is indicative of a ‘high degree’ of equilibrium (i.e., even the slowest ionic species are equilibrated; Liang et al., 2012); (iv) within compositional uncertainties, the olivine and pyroxenes were in equilibrium with the same melt; Nevertheless, the origin of dunite within the uppermost mantle remains a controversial issue. Three end member origins are classically discussed: (i) a cumulate origin; (ii) an extremely refractory residue of melting; and (iii) a rock obtained through melt-rock reaction between a basaltic liquid (sensu lato) and a fertile/moderately depleted peridotite (e.g., lherzolite-harzburgite). Previous studies (e.g., Frey and Prinz, 1978) have suggested a cumulate origin for several San Carlos dunite specimens, especially those associated with websterite/ clinopyroxenite since their layering is reminiscent of that commonly observed in layered intrusions of unambiguous cumulate origin. In this study, there are several petrophysical and geochemical characteristics of the sample that are difficult to reconcile with a cumulate origin. First and foremost, the olivine has an axial-[100] texture indicative of ductile deformation, whereas olivine in cumulate rocks show axial[010] textures (e.g., Tommasi et al., 2004). Here, the high Fo% (≈91.5%), high Ni content (≈2887 ppm), and low LREE/MREE/HREE ratios of Ol in the dunite are also not in agreement with an exclusive derivation from a small amount of melting, but are indicative of a melting residue with complete removal of both Cpx and Opx by an extremely high degree of melting (e.g., stage-1: melting at 30–40%). Estimating the degree of melting based only on the REE abundances using the method of Norman (1998) yields a much more moderate degree of melting of 6 ± 1% for our sample. A fractional melting model is then favored by the steep negative LREEMREE trend shown in Fig. 4. This mismatch between REEs and the modal mineralogy (±mineral major element compositions) suggests that melting alone cannot account for all the geochemical characteristics of the studied peridotite. Complete dunitification through meltrock percolation-reaction occurs only at low pressure (b2 GPa), when the melt becomes olivine-saturated, and implies high melt/rock ratios.

306

C.M.M. Denis et al. / Lithos 302–303 (2018) 298–311

1:1

-1

Unpo. Norm. Int. Abs. (rim) [cm ]

250

200

150

100

50

0 0

50

100

150

200

250

-1

Unpo. Norm. Int. Abs. (core) [cm ] Fig. 6. The distribution of hydrogen in olivine, orthopyroxene (Opx), and clinopyroxne (Cpx) within the composite xenolith (peridotite 14SC49). (a) Natural light photograph of the analytical area in the xenolith. (b) Unpolarized integrated absorbances normalized to sample thickness (Unpo. Norm. Int. Abs.) for cores and rims as a function of the distance to the contact (dashed vertical line) between the two lithologies. The dashed horizontal line indicates the FTIR detection limit for H in olivine. (c) Unpolarized integrated absorbances normalized to sample thickness for pyroxene cores versus rims.

Therefore the chemistry of such a reactive rock tends to be buffered by the percolating-reacting melt, lowering the Fo% and Ni content of the olivine. Such percolation of large portion of reacting melt will enhance olivine modal content, recrystallization leading to polygonal (equilibrated) textures (i.e., straight grain boundaries, triple junctions, and weakening of the olivine CPO), associated with a subsequent (i.e., stage-2) melt crystallization of locally produced pyroxene-rich rock types (e.g., Tommasi et al., 2004). Within this two-stage framework, a very attractive scenario for the formation of such a dunitepyroxenite composite xenolith was already proposed by Frey and Prinz (1978) and Wilshire and Shervais (1975). They considered that precipitation (or flow differentiation) of pyroxenes along the edges of a feeder dike (i.e., a dunite channel) could account for most of the petrographic and geochemical characteristics of such composite peridotite (see also Galer and O'Nions, 1989). However, pyroxenites at the dike/ dunite channel wall do not show LREE and MREE depletions relative

to HREEs (e.g., Godard et al., 1995; Kelemen et al., 1995), but typically show high LREE/MREE/HREE ratios. Frey and Prinz (1978) argued for a second melting stage of the dikes and/or segregation processes within dikes, which would imply that the geochemical imprint of the percolating melt was not retained. Finally, trace element patterns of all minerals from both the dunite and websterite sections (Fig. 4) show a pronounced enrichment in U-Th-La ± Ce, some of the most incompatible large ion lithophile elements, and U is selectively more enriched than Th (i.e., (U/Th)PM ≫ 1). These enrichments and U/Th fractionation imply that both sections of the composite xenolith were infiltrated by a small volume of percolating melts or fluids enriched in those elements. Such U/Th fractionations have been tentatively linked to a change in U speciation (U4+ to U6+), allowing a significantly different partitioning behaviour between U and Th (Alard et al., 2011). The percolating fluids must be enriched in volatile and incompatible elements and able to lower the peridotite solidus temperature to achieve efficient percolation within the lithosphere. These fluids may be the product of an extremely small degree of melting (McKenzie, 1989) or result from a previous melt-rock reaction that consumed most of the melt (Bedini and Bodinier, 1999). The hypothesized partial melting or melt-rock interaction is in good agreement with the low hydrogen contents of olivine and the pyroxenes, which is typical of spinel-bearing peridotites resulting from high melt-rock ratio interactions (e.g., Satsukawa et al., 2017; reviews by Demouchy and Bolfan-Casanova, 2016; Peslier et al., 2017). Indeed, we note that our dunite section is as hydrogen-poor as previously reported for San Carlos olivine: 2.6 ppm H2O wt in sample SC99–2 from Li et al. (2008), and b0.5 ppm H2O wt in sample SC8803B from Grant et al. (2007) (olivines in both studies being 89–90 Fo%). The pyroxenes from this study, however, contain less hydrogen than previously reported for San Carlos pyroxenes (171 and 82 ppm H2O wt for Cpx and Opx, respectively, in sample SC99–1 from Li et al., 2008). In both olivine and pyroxene, hydrogen concentrations are at the lower end of the ranges reported for upper mantle peridotites (see Fig. 14 in Demouchy and Bolfan-Casanova, 2016). Thus, in theory, since hydrogen is an incompatible element, a reacting melt/fluid agents, should impact mineral hydrogen distributions if it is water-saturated, but here it did not notably modify hydrogen concentrations in the co-existing NAMs (b50 ppm H2O in pyroxenes). This could be due to the combined effect of low hydrogen solubility at low temperature (b1100 °C) and pressure (b2 GPa) (see Férot and Bolfan-Casanova, 2012) and/or a different incompatibility level of hydrogen if the reactive agent is a volatile (hydrous)rich fluid or a water under-saturated melt, as previously proposed (e.g., Demouchy and Bolfan-Casanova, 2016; Denis et al., 2015; Satsukawa et al., 2017).

4.2. Spatial distribution of hydrogen The studied sample could, in theory, allow discussion of the hydrogen distribution at both the cm (section) and mm (grain) scale. However, the lack of analyzable pyroxenes within the dunite unit precludes comparison at the cm scale, and we cannot rely on olivine because the hydrogen concentrations are below the detection limit. Thus, we cannot draw conclusions on the re-hydrating potential of the percolating melt responsible for the pyroxenites. Indeed, co-existing NAMs in centimetric amphibole-rich veins in peridotites (Denis et al., 2015; Schmädicke et al., 2013) were not specifically hydrated compared to amphibole-free peridotites of the same xenolith suite. At the grain scale, only Opx display heterogeneous hydrogen concentrations, but the rim-depletions are not observed in each Opx. Only Opx grains larger than ~1 mm and with higher hydrogen concentrations reveal pronounced depletion profiles. The maximum depletion (by a factor of 5; Fig. 8d) is only observed for specific directions of measurement and directions of the di-electric vector (Figs. 7c and 8d).

C.M.M. Denis et al. / Lithos 302–303 (2018) 298–311

(a)

307

Olivine AA Pro1 unpol Pro1 pol. // NS Pro1 pol. // EW Pro2 pol. // NS Pro2 pol. // EW

-1

Norm. Int. Abs [cm ]

3

2

1

0 0.0 (b)

0.5

1.0 1.5 Distance [mm]

2.0

2.5

-1

Norm. Int. Abs [cm ]

15

10

CPX 10 Pro1 unpol. Pro1 pol. // NS Pro1 pol. // EW Pro2 pol. // NS Pro2 pol. // EW

5

0 0.0

0.5

(c)

1.0 1.5 Distance [mm]

2.0

-1

Norm. Int. Abs [cm ]

10 8 6 4

OPX 1 Pro1 unpol. Pro1 pol. // NS Pro1 pol. // EW Pro2 pol. // NS Pro2 pol. // EW

2 0 0.0

0.4

0.8 Distance [mm]

1.2

1.6

Fig. 7. Mineral hydrogen concentration variations in peridotite 14SC49: (a) integrated absorbances normalized to sample thickness (Norm. Int. Abs.) from unpolarized and polarized FTIR measurements of the olivine grain shown to the left; (b) unpolarized and polarized FTIR measurements of the clinopyroxene (Cpx) grain shown to the left; and (c) unpolarized and polarized FTIR measurements for the orthopyroxene (Opx) grain shown to the left.

One important issue is whether these depletion profiles are correlated with major element variations in the atomic structure, such as Al, Fe, Ca, or Cr, which have been shown to modify hydrogen incorporation mechanisms (and the resulting concentrations) in Opx (e.g., Férot and Bolfan-Casanova, 2012; Mierdel et al., 2007; Rauch and Keppler, 2002; Stalder, 2004; Stalder et al., 2005, 2015). Our core-to-rim EPMA analyses have revealed no Fe, Mg, Cr, or Ca heterogeneities (Fig. 3a), but minor variations are observed for Al. Opx cores are indeed slightly

enriched in Al2O3, although the variations are narrow with average values of 3.73 wt% Al2O3 for Opx rims (range is 3.68–3.82 wt%) and 3.81 wt% Al2O3 for Opx cores (range is 3.74–3.95 wt%; Fig. 3a). Unfortunately, experimental data for this range of Al2O3 contents at the appropriate pressure and temperature conditions (1–1.5 GPa and 1000 °C) are not yet available (e.g., Férot and Bolfan-Casanova, 2012; Mierdel et al., 2007; Rauch, 2000). Nonetheless, if we follow the correlation reported by Stalder (2004) at 2.5 GPa for Al-H coupled incorporation,

C.M.M. Denis et al. / Lithos 302–303 (2018) 298–311

Abs. Coeff. [cm-1]

Abs. Coeff. [cm-1]

Abs. Coeff. [cm-1]

308

the relative variations in Opx Al2O3 contents (Δ0.08 wt% Al2O3) yield a theoretical relative hydrogen concentration variation (Δ16 ppm H2O wt) similar to the variation reported here (Δ19 ppm H2O wt; Fig. 8c). However, the absolute hydrogen concentration predicted by Stalder (2004) yields 1032 ppm H2O wt for 3.77 wt% Al2O3, which is difficult to reconcile with the very low hydrogen concentrations observed in our Opx if Al and H are indeed coupled during hydration/ dehydration. Therefore, at present, we exclude Al2O3 as a control on hydrogen mobility after hydration in our Opx. The non-negligible pleochroism of the OH bands in Opx could also be responsible for the apparent hydroxyl depletion profiles. Previous studies that have reported polarized FTIR analyses of mantle-derived Opx are surprisingly rare for San Carlos locality (Carpenter Woods, 2001), but are available for other localities (Kilbourne hole, Bell et al., 1995; Mexico and Simcoe, Peslier et al., 2002; Kaapvaal, Peslier et al., 2012; see also Mosenfelder and Rossman, 2013). These studies reported polarized spectra with main OH bands positioned at 3596, 3565, 3523, 3404, 3308, and 3068 cm−1, each having maximum absorbance with E//[001] (Bell et al., 1995; Carpenter Woods, 2001; Peslier et al., 2012). The average absorbance distribution, based on integrated spectra between 3700 and 2800 cm−1 is: 16% for E//[100], 23% for E//[010], and 60% for E//[001] (N.B. the optical directions α, β, and γ correspond to the [010], [100], and [001] crystallographic axes, respectively). In our sample, based on the polarized IR spectrum with the maximum absorbance (Fig. 8d, profile 1, E// East–West +32°), the OH bands have very similar positions at 3568, 3521, 3405, and 3307 cm−1, plus a weak (shoulder) band at 3580 cm−1, and the same band intensity distribution (dominant bands at 3596 and 3565 cm−1). This strongly advocates that the orientation E//EW + 32° is near E//[001]. The Si-O overtones in that orientation are also similar to [001] (Supplementary Fig. S2); thus we conclude that the profile with E//EW + 32° is close to E//[001]. This orientation, and the symmetry of the profile, reinforces our interpretation that the hydrogen depletion observed in this Opx grain is indeed the result of dehydration by ionic diffusion, and is not simply an apparent variation due to pleochroism of the OH bands in a randomly crystallographically oriented Opx.

4.3. Time scale of dehydration Experimental studies have quantified hydrogen diffusivities in NAMs at various high temperature (600–1110 °C) and high pressure conditions (1 atm to 3 GPa) and in several chemical systems (see Ingrin and Blanchard, 2006 for a review; Du Frane and Tyburczy, 2012; Novella et al., 2017; Poe et al., 2010). While mantle-derived olivines, such as San Carlos peridots, are rather popular for deformation and hydration experiments at high pressure, mantle compositions are rarely used for hydrogen diffusion experiments in pyroxenes (e.g., Carpenter Woods, 2001; Carpenter Woods et al., 2000; Hercule and Ingrin, 1999; Ingrin and Blanchard, 2006; Stalder et al., 2007). Hydrogen diffusion coefficients at 1100 °C in San Carlos olivine range from 6 × 10−12 (protonvacancy mechanism) to 4 × 10−10 cm2/s (proton-polaron mechanism) (Demouchy and Mackwell, 2006; Mackwell and Kohlstedt, 1990), in the same range as diffusivities for San Carlos enstatite (4 × 10−12 to 1.3 × 10−11 cm2/s; Carpenter Woods, 2001). Although no H diffusion coefficients for San Carlos diopside have been reported to date, a recent experimental study reported diffusion coefficients at 1100 °C ranging from 10−13 cm2/s for Kunlun diopside to 10−9 cm2/s for Jaipur diopside and Fuego Cpx (see Fig. 11 in et al., 2016). For a composition as close as possible to mantle-derived diopside (megacryst in basanite from Nüshan; Xia et al., 2000), the hydrogen diffusion coefficient is around 10−11 cm2/s. If similar vacancy-controlled diffusion mechanisms are Fig. 8. Hydrogen concentration variations from polarized FTIR spectra in orthopyroxene 1 from peridotite 14SC49. (a) 3D profile for E//EW −53°; (b) 3D profile for E//EW; (c) 3D profile for E//EW + 32°; and (d) corresponding normalized integrated absorbances for profiles (a)–(c). For E//EW + 32° (c), the profile is very close to E//[001] (see main text for details).

C.M.M. Denis et al. / Lithos 302–303 (2018) 298–311

considered at 1100 °C, hydrogen diffusion in Cpx is slightly faster than in Opx, and diffusion in olivine is the slowest. Nonetheless the differences in diffusivities are very small (diopside, 3.1 × 10−11 cm2/s; enstatite, 1.3 × 10−11 cm2/s; olivine, 6 × 10−12 cm2/s), and the absolute diffusivities remain very rapid and predict effective dehydration in b5 days at 1100 °C for the average grain sizes reported here, i.e., 420, 223, and 141 μm for Ol, Opx, and Cpx, respectively. Although, several studies have reported isotropic diffusion in Opx (e.g., Ingrin and Blanchard, 2006), Carpenter Woods (2001) reported that hydrogen diffusion is anisotropic in San Carlos Opx, with diffusion coefficients DH[001] N DH[100] N DH[010] at 1100 °C. The concentration observed in our composite xenolith from San Carlos (i.e., dry olivine, dehydrating enstatite, and Cpx not yet dehydrating) is then at odds with the reported H diffusivities from experimental studies. Although the Ol and Opx crystals used during diffusion experiments were of mantle composition, this was not the case for diopside. Therefore, we can only propose several hypotheses to explain the mineral specific hydrogen concentrations in the composite xenolith: (i) the proton-polaron mechanism of hydrogen diffusion plays a non-negligible role during dehydration in olivine, as previously proposed by Thoraval and Demouchy (2014), leading to hydrogen diffusivity in olivine significantly faster than in pyroxenes; (ii) hydrogen diffusivity in diopside is extremely sensitive to chemical composition, and experimental diffusion data for mantle compositions are not yet available (see compilation by Ferriss et al., 2016) to accurately constrain the relative hydrogen diffusion in peridotitic NAMs; and (iii) the hydrogen diffusion coefficient in pyroxenes could be concentration-dependent or affected by local hydrogen chemical potentials within grain boundaries. 5. Conclusion Mantle processes lead to lithological heterogeneities and variable trace and REE distributions. Major and trace elements (excluding H) show robust features of chemical equilibrium. Hydrogen, as a light trace element, was expected to follow other incompatible elements such as La and Ce. Here, our study of San Carlos composite peridotite sample 14SC49 allows us to decipher if melt-rock interactions would lead to depleted or enriched hydrogen concentrations. The hydrogen concentrations obtained from FTIR measurements are negligible for Ol (b2 ppm H2O wt) and low for Opx and Cpx (24 and 53 ppm H2O wt, respectively) and ascribed as the result of high melt-rock ratio interactions. No significant differences in chemical composition are observed between the host dunite and the surrounding pyroxenites (olivine websterite) due to long annealing at depth and thus complete chemical reequilibration. Nevertheless, we have observed heterogeneous (rimdepleted) hydrogen concentrations in Opx and interpreted them as the result of a late dehydration by ionic diffusion during rapid uplift. Our results are the second (after Tian et al., 2016) to report Opx as a feeble recorder of hydrogen in the depth Earth. We stress the need for experimental diffusion coefficients in peridotitic diopside; however, considering available experimental hydrogen diffusivities as accurate (i.e., very rapid), hydrogen, even already presents in small amount, was lost during ascent toward the surface, and the remaining hydrogen found in peridotitic olivine and pyroxenes very likely represents only the ‘tip of the iceberg’ of the water stored in the Earth's mantle. Supplementary data to this article can be found online at https://doi. org/10.1016/j.lithos.2018.01.007. Acknowledgements CNRS supported this study through INSU 2011, 2012, and 2013 programs to S. Demouchy and O. Alard. The University of Montpellier supported this study through a Ph. D fellowship “Bourse Président” to O. Alard and C. Denis. The authors thank Prof. A. Nicolas and Prof. F. Boudier for use of the peridotite collection. S.D. thanks two anonymous reviewers for their useful comments on the manuscript, M. Scambelluri

309

for excellent editorship, J.-A. Padròn-Navarta, A. Tommasi for constructive discussions, C. Nevado and D. Delmas for sample preparation, F. Barou for assistance with EBSD measurements, and D. Maurin for management of the FTIR lab at the University of Montpellier. Electron microprobe analyses were performed at ‘Microsconde SUD’ with the help of B. Boyer. References Adam, J., Green, T., 2006. Trace element partitioning between mica-and amphibolebearing garnet lherzolite and hydrous basanitic melt: 1. Experimental results and the investigation of controls on partitioning behaviour. Contributions to Mineralogy and Petrology 152 (1), 1–17. Alard, O., Dautria, J.-M., Bodinier, J.-L., 1996. Nature of the upper mantle and metasomatic processes on either side of the Sillon Houiller (French Massif central). Comptes Rendus de l' Academie des Sciences Serie 2: Sciences de la Terre et des Planètes 323, 763–770. Alard, O., Lorand, J.P., Reisberg, L., Bodinier, J.L., Dautria, J.M., O'reilly, S.Y., 2011. Volatilerich metasomatism in Montferrier xenoliths (southern France): implications for the abundances of chalcophile and highly siderophile elements in the subcontinental mantle. Journal of Petrology 52, 2009–2045. Bédard, J.H., 2005. Partitioning coefficients between olivine and silicate melts. Lithos 83, 394–419. Bedini, M.R., Bodinier, J.L., 1999. Distribution of incompatible trace elements between the constituents of spinel peridotite xenoliths: ICP-MS data from the east African rift. Geochimica et Cosmochimica Acta 63, 3883–3900. Bell, D.R., Rossman, G.R., 1992. Water in Earth's mantle: the role of nominally anhydrous minerals. Science 255, 1391–1397. Bell, D.R., Ihinger, P.D., Rossman, G.R., 1995. Quantitative and analysis of trace OH in garnet and pyroxenes. American Mineralogist 80, 465–474. Ben Ismail, W., Mainprice, D., 1998. An olivine fabric database: an overview of upper mantle fabrics and seismic anisotropy. Tectonophysics 296, 145–157. Brey, G.P., Köhler, T., 1990. Geothermobarometry in four-phase lherzolites II. New thermobarometers, and practical assessment of existing thremobarometers. Journal of Petrology 31, 1353–1378. Bunge, H.J., 1982. Texture Analysis in Materials Sciences. Buttleworth, London. Carpenter Woods, S., 2001. The Kinetics of Hydrogen Diffusion in Single Crystal Orthopyroxene. (PhD thesis). Dept. Geosciences. Pennsylvania State University. Carpenter Woods, S., Mackwell, S.J., Dyar, D., 2000. Hydrogen in diopside: diffusion profiles. American Mineralogist 85, 480–487. Demouchy, S., Bolfan-Casanova, N., 2016. Distribution and transport of hydrogen in the lithospheric mantle: a review. Lithos 240-243, 402–425. Demouchy, S., Mackwell, S., 2003. Water diffusion in synthetic iron-free forsterite. Physics and Chemistry of Minerals 30, 486–494. Demouchy, S., Mackwell, S., 2006. Mechanisms of hydrogen incorporation and diffusion in iron-bearing olivine. Physics and Chemistry of Minerals 33, 347–355. Demouchy, S., Jacobsen, S.D., Gaillard, F., Stern, C.R., 2006. Rapid magma ascent recorded by water diffusion profiles in mantle olivine. Geology 34, 429–432. Demouchy, S., Ishikawa, A., Tommasi, A., Alard, O., Keshav, S., 2015. Characterization of hydration in the mantle lithosphere: peridotite xenoliths from the Ontong java plateau as an example. Lithos 212-215, 189–201. Demouchy, S., Shcheka, S., Denis, C.M.M., Thoraval, C., 2017. Subsolidus hydrogen partitioning between nominally anhydrous minerals in garnet-bearing peridotite. American Mineralogist 102, 1822–1831. Denis, C.M.M., Demouchy, S., Shaw, C., 2013. Evidence of dehydration in peridotites from Eifel volcanic field and estimates of magma ascent rates. Journal of Volcanology and Geothermal Research 258, 85–99. Denis, C.M.M., Alard, O., Demouchy, S., 2015. Water content and hydrogen behaviour during metasomatism in the uppermost mantle beneath ray pic volcano (massif central, France). Lithos 236-237, 256–274. Du Frane, W.L., Tyburczy, J.A., 2012. Deuterium-hydrogen exchange in olivine: implications for point defects and electrical conductivity. Geochemistry, Geophysics, Geosystems 13. https://doi.org/10.1029/2011GC003895. Férot, A., Bolfan-Casanova, N., 2012. Water storage capacity in olivine and pyroxene to 14 GPa: implications for the water content of the Earth's upper mantle and nature of seismic discontinuities. Earth and Planetary Science Letters 349-350, 218–230. Ferriss, E., Plank, T., Walker, D., 2016. Site-specific hydrogen diffusion rates during clinopyroxene dehydration. Contributions to Mineralogy and Petrology. https://doi. org/10.1007/s00410-016-1262-8. Frets, E.C., Tommasi, A., Garrido, C.J., Vauchez, A., Mainprice, D., Targuisti, K., Amri, I., 2014. The Beni Bousera peridotite (Rif Belt, Morocco): an oblique-slip low-angle shear zone thinning the subcontinental mantle lithosphere. Journal of Petrology 55, 283–313. Frey, F.A., Prinz, M., 1978. Ultramafic inclusions from San Carlos, Arizona: petrologic and geochemical data bearing on their petrogenesis. Earth and Planetary Science Letters 38, 129–176. Galer, S.J.G., O'Nions, R.K., 1989. Chemical and isotopic studies of ultramafic inclusions from the San Carlos volcanic field, Arizona: a bearing on their petrogenesis. Journal of Petrology 30 (4), 1033–1064. Garrido, C.J., Bodinier, J.L., Alard, O., 2000. Incompatible trace element partitioning and residence in anhydrous spinel peridotites and websterites from the Ronda orogenic peridotite. Earth and Planetary Science Letters 181, 341–358. Godard, M., Bodinier, J.-L., Vasseur, G., 1995. Effects of mineralogical reactions on trace element redistributions in mantle rocks during percolation processes: a chromatographic approach. Earth and Planetary Science Letters 133, 449–461.

310

C.M.M. Denis et al. / Lithos 302–303 (2018) 298–311

Grant, K.J., Brooker, R.A., Kohn, S.C., Wood, B.J., 2007. The effect of oxygen fugacity on hydroxyl concentrations and speciation in olivine: implications for water solubility in the upper mantle. Earth and Planetary Science Letters 261, 217–229. Gregory, R.T., Taylor, H.P., 1986. Possible non-equilibrium oxygen isotope effects in mantle nodules, an alternative to the Kyser-O'Neil-Carmichael 18O/16O geothermometer. Contributions to Mineralogy and Petrology 93, 114–119. Griffin, W.L., Powell, W., Pearson, N.J., O'Reilly, S.Y., 2008. GLITTER: Data Reduction Software for Laser Ablation ICP-MS. Laser Ablation-ICP-MS in the Earth Sciences. Mineralogical Association of Canada Short Course Series vol. 40 pp. 204–207. Hao, Y.-T., Xia, Q.-K., Tian, Z.-Z., Liu, J., 2016. Mantle metasomatism did not modify the initial H2O content in peridotite xenoliths from the Tianchang basalts of eastern China. Lithos 260, 315–327. Hercule, S., Ingrin, J., 1999. Hydrogen in diopside: diffusion, kinetics of extractionincorporation, and solubility. American Mineralogist 84, 1577–1587. Hielscher, R., Schaeben, H., 2008. A novel pole figure inversion method: specification of the MTEX algorithm. Journal of Applied Crystallography 41, 1024–1037. Ingrin, J., Blanchard, M., 2006. Diffusion of Hydrogen on Minerals. vol. 62. American Mineralogical Society Geochemical Society, Chantilly (Vir.), pp. 291–320. Ingrin, J., Skogby, H., 2000. Hydrogen in nominally anhydrous upper-mantle minerals: concentration levels and implications. European Journal of Mineralogy 12, 543–570. Jean, M.M., Taylor, L.A., Howarth, G.H., Peslier, A.H., Fedele, L., Bodnar, R.J., Guan, Y., Doucet, L.S., Ionov, D.A., Logvinova, A.M., Golovin, A.V., Sobolev, N.V., 2016. Olivine inclusions in Siberian diamonds and mantle xenoliths: contrasting water and traceelement contents. Lithos 265, 31–41. Jochum, K.P., Nohl, U., Herwig, K., Lammel, E., Stoll, B., Hofmann, A.W., 2005. GeoReM: a new geochemical database for reference materials and isotopic standards. Geostandards and Geoanalytical Research 29, 333–338. Kelemen, P., Shimizu, N., Dunn, T., 1993. Relative depletion of niobium in some arc magmas and the continental crust: partitioning of K, Nb, La, and Ce during melt/ rock reactions in the upper mantle. Earth and Planetary Science Letters 120, 111–134. Kelemen, P.B., Shimizu, N., Salters, V.J.M., 1995. Extraction of mid-ocean-ridge basalt from the upwelling mantle by focused flow of melt in dunite channels. Nature 375, 747–753. Kikuchi, S., 1928. Diffraction of cathode rays by mica. Japanese Journal of Physics 5, 83–96. Kohlstedt, D.L., Keppler, H., Rubie, D.C., 1996. Solubility of water in the α, β and γ phases of (Mg,Fe)2SiO4. Contributions to Mineralogy and Petrology 123, 345–357. Li, Z.X.A., Lee, C.T.A., Peslier, A., Lenardic, A., Mackwell, S.J., 2008. Water contents in mantle xenoliths from the Colorado Plateau and vicinity: implications for the mantle rheology and hydration-induced thining of continental lithosphere. Journal of Geophysical Research 113 (doi: 10.1029-2007JB005540). Liang, Y., Sun, C., Yao, L., 2012. A REE-in-two-pyroxene thermometer for mafic and ultramafic rocks. Geochimica et Cosmochimica Acta 102, 246–260. Lloyd, A.S., Plank, T., Ruprecht, P., Hauri, E.H., Rose, W., 2012. Volatile loss from melt inclusions in pyroclasts of differing sizes. Contributions to Mineralogy and Petrology 165, 129–153. Lloyd, A.S., Ferriss, E., Ruprecht, P., Hauri, E.H., Jicha, B.R., Plank, T., 2016. An assessment of clinopyroxene as a recorder of magmatic water and magma ascent rate. Journal of Petrology 57, 1865–1886. Mackwell, S.J., Kohlstedt, D.L., 1990. Diffusion of hydrogen in olivine: implications for water in the mantle. Journal of Geophysical Research 95, 5079–5088. Macris, C.A., Manning, C.E., Young, E.D., 2015. Crystal chemical constraints on intermineral Fe isotope fractionation and implications for Fe isotope disequilibrium in San Carlos mantle xenoliths. Geochimica et Cosmochimica Acta 154, 168–185. Mainprice, D., Humbert, M., 1994. Methods of calculation petrophysical properties from lattice preferred orientation data. Surveys in Geophysics 15, 575–592. Mainprice, D., Silver, P.G., 1993. Interpretation of SKS-waves using samples from the subcontinental lithosphere. Physics of the Earth and Planetary Interiors 78, 257–280. Mainprice, D., Bachmann, F., Hielscher, R., Schaeben, H., 2014. Descriptive tools for the analysis of texture projects with large datasets using MTEX: strength, symmetry and components. Geological Society of London, Special Publication 409. https://doi. org/10.1144/SP1409.1148. McDonough, W.F., Sun, S.S., 1995. The composition of the earth. Chemical Geology 120, 223–253. McKenzie, D., 1989. Some remarks on the movement of small melt fractions in the mantle. Earth and Planetary Science Letters 95, 53–72. Mercier, J.-C., Nicolas, A., 1975. Textures and fabrics of upper mantle peridotites as illustrated by xenoliths from basalts. Journal of Petrology 16 (1), 454–487. Mierdel, K., Keppler, H., Smyth, J.R., Langerhorst, F., 2007. Water solubility in aluminous orthopyroxene and the origin of the asthenosphere. Science 315, 364–368. Mosenfelder, J.L., Rossman, G.R., 2013. Analysis of hydrogen and fluorine in pyroxenes: I. Orthopyroxene. American Mineralogist 98, 1026–1041. Norman, M.D., 1998. Melting and metasomatism in the continental lithosphere: laser ablation ICPMS analysis of minerals in spinel lherzolites from eastern Australia. Contributions to Mineralogy and Petrology 3-4, 240–255. Novella, D., Bolfan-Casanova, N., Nestola, F., Harris, J.W., 2015. H2O in olivine and garnet inclusions still trapped in diamonds from the Siberian craton: implications for the water content of cratonic lithosphere peridotites. Lithos 230, 180–183. Novella, D., Jacobsen, B., Weber, P.K., Tyburczy, J.A., Ryerson, F.J., Frane, Du, W.L., 2017. Hydrogen self-diffusion in single crystal olivine and electrical conductivity of the Earth's mantle. Nature Publishing Group 7, 629. Paterson, M.S., 1982. The determination of hydroxyl by infrared absorption in quartz, silicate glasses and similar materials. Bulletin de Minéralogie 105, 20–29. Pearce, N.J.G., Perkins, W.T., Westgate, J.A., Gorton, M.P., Jackson, S.E., Neal, C.R., Chenery, S.P., 1997. A compilation of new and published major and trace element data for NIST SRM 610 and NIST SRM 612 glass reference materials. Geostandards and Geoanalytical Research 21, 115–144.

Pearson, D.G., Brenker, F.E., Nestola, F., McNeill, J., Nasdala, L., Hutchison, M.T., Matveev, S., Mather, K., Silversmit, G., Schmitz, S., Vekemans, B., Vincze, L., 2015. Hydrous mantle transition zone indicated by ringwoodite included within diamond. Nature 507, 221–224. Peslier, A.H., 2010. A review of water contents of nominally anhydrous minerals in the mantles of Earth, Mars and the moon. Journal of Volcanology and Geothermal Research 197, 239–258. Peslier, A., Luhr, J.F., 2006. Hydrogen loss form olivines in mantle xenoliths from Simcoe (USA) and Mexico: mafic alkalic magma ascent rates and water budget of the subcontinental lithosphere. Earth and Planetary Science Letters 242, 302–314. Peslier, A., Luhr, J., Post, J., 2002. Low water contents in pyroxenes from spinel-peridotites of the oxidized, sub-arc mantle wedge. Earth and Planetary Science Letters 201, 69–86. Peslier, A.H., Woodland, A.B., Bell, D.R., Lazarote, M., Lapen, T.J., 2012. Metasomatic control of water contents in the Kaapvaal cratonic mantle. Geochimica et Cosmochimica Acta 97, 213–246. Peslier, A.H., Schönbächler, M., Busemann, H., Karato, S.-I., 2017. Water in the Earth's interior: distribution and origin. Space Science Reviews. https://doi.org/10.1007/ s11214-017-0387-z. Poe, B.T., Romano, C., Nestola, F., Smyth, J.R., 2010. Electrical conductivity anisotropy of dry and hydrous olivine at 8 GPa. Physics of the Earth and Planetary Interiors 181, 103–111. Rampone, E., Borghini, G., 2008. Melt migration and intrusion in the Erro-Tobbio peridotites (Ligurian Alps, Italy): insights on magmatic processes in extending lithospheric mantle. European Journal of Mineralogy 20, 573–585. Rampone, E., Bottazzi, P., Ottolini, L., 1991. Complementary Ti and Zr anomalies in orthopyroxene and clinopyroxene from mantle peridotites. Nature 354, 518–520. Rampone, E., Romairone, A., Hofmann, A.W., 2004. Contrasting bulk and mineral chemistry in depleted mantle peridotites: evidence for reactive porous flow. Earth and Planetary Science Letters 218, 491–506. Rauch, M., 2000. Der Einbau von Wasser in Pyroxene. (PhD thesis). Bayerisches Geoinstitut. Bayreuth Universität, Bayreuth. Rauch, M., Keppler, H., 2002. Water solubility in orthopyroxene. Contributions to Mineralogy and Petrology 143, 525–536. Satsukawa, T., Godard, M., Demouchy, S., Michibayashi, K., Ildefonse, B., 2017. Chemical interactions in the subduction factory: new insights from an in situ trace element and hydrogen study of the Ichinomegata and Oki-Dogo mantle xenoliths (Japan). Geochimica et Cosmochimica Acta 208, 234–267. Schmädicke, E., Gose, J., Witt-Eickschen, G., Brätz, H., 2013. Olivine from spinel peridotite xenoliths: hydroxyl incorporation and mineral composition. American Mineralogist 98, 1870–1880. Soustelle, V., Tommasi, A., Demouchy, S., Ionov, D., 2010. Deformation and fluid-rock interactions in supra-subduction mantle: microstructures and water contents in peridotite xenoliths from the Avacha volcano, Kamchatka. Journal of Petrology 51, 363–394. Stalder, R., 2004. Influence of Fe, Cr, and Al on hydrogen incorporation in orthopyroxene. European Journal of Mineralogy 16, 703–711. Stalder, R., Behrens, H., 2006. D/H exchange in pure and Cr-doped enstatite: implications for hydrogen diffusivity. Physics and Chemistry of Minerals 33, 601–611. Stalder, R., Skogby, H., 2003. Hydrogen diffusion in natural and synthetic orthopyroxene. Physics and Chemistry of Minerals 30, 12–19. Stalder, R., Klemme, S., Ludwig, T., Skogby, H., 2005. Hydrogen incorporation in orthopyroxene: interaction of different trivalent cations. Contributions to Mineralogy and Petrology 150, 473–485. Stalder, R., Purwin, H., Skogby, H., 2007. Influence on hydrogen diffusivity in orthopyroxene. European Journal of Mineralogy 19, 899–903. Stalder, R., Karimova, A., Konzett, J., 2015. OH-defects in multiple-doped orthoenstatite at 4-8 GPa: filling the gap between pure and natural systems. Contributions to Mineralogy and Petrology 1–10. Streckeisen, 1976. Classification of the common igneous rocks by means of their chemical composition: a provisional attempt. Neues Jahrbuch fur Mineralogie, Monatshefte 1976 (H.l), 1–15. Taylor, L.A., Logvinova, A.M., Howarth, G.H., Liu, Y., Peslier, A.H., Rossman, G.R., Guan, Y., Chen, Y., Sobolev, N.V., 2016. Low water contents in diamond mineral inclusions: proto-genetic origin in a dry cratonic lithosphere. Earth and Planetary Science Letters 433, 125–132. Thoraval, C., Demouchy, S., 2014. Numerical models of ionic diffusion in one and three dimensions: application to dehydration of mantle olivine. Physics and Chemistry of Minerals 41, 709–723. Tian, Z.-Z., Liu, J., Xia, Q.-K., Ingrin, J., Hao, Y.-T., Christophe, D., 2016. Water concentration profiles in natural mantle orthopyroxenes: a geochronometer for long annealing of xenoliths within magma. Geology 45 (1), 87–90. Tommasi, A., Mainprice, D., Canova, G., Chastel, Y., 2000. Viscoplastic self-consistent and equilibrium-based modeling of olivine lattice preferred orientations: implications for the upper mantle seismic anisotropy. Journal of Geophysical Research 105, 7893-1908. Tommasi, A., Godard, M., Coromina, G., Dautria, J.-M., Barsczus, H., 2004. Seismic anisotropy and compositionally induced velocity anomalies in the lithosphere above mantle plumes: a petrological and microstructural study of mantle xenoliths from French Polynesia. Earth and Planetary Science Letters 227, 539–556. Wilshire, H.G., Shervais, J.W., 1975. Al-augite and Cr-diopside ultramafic xenoliths in basaltic rocks from western United State. Physics and Chemistry of the Earth 9, 257–272. Wirth, R., 1996. Thin amorphous films (1-2 nm) at olivine grain boundaries in the mantle xenoliths from San Carlos, Arizona. Contributions to Mineralogy and Petrology 124, 44–54.

C.M.M. Denis et al. / Lithos 302–303 (2018) 298–311 Withers, A.C., Bureau, H., Raepsaet, C., Hirschmann, M.M., 2012. Calibration of infrared spectroscopy by elastic recoil detection analysis of H in synthetic olivine. Chemical Geology 334, 92–98. Xia, Q., Chen, D., Carpenter, S., Zhi, X., Wang, R., Cheng, H., 2000. Hydrogen diffusion in clinopyroxene: dehydration experiments. Science in China 43, 561–568. Xia, Q.K., Hao, Y., Deloule, E., Coltorti, M., Dallai, L., Yang, X., Feng, M., 2010. Low water content of the Cenozioc lithosphere mantle beneath the eastern part of North China Craton. Journal of Geophysical Research 115 (doi: 1029–2009JB006694). Xia, Q.-K., Liu, J., Kovács, I., Hao, Y.-T., Li, P., Yang, X.-Z., Chen, H., Sheng, Y.-M., 2017. Water in the upper mantle and deep crust of eastern China: concentration, distribution and implications. National Science Review. https://doi.org/10.1093/nsr/nwx016. Young, E.D., Tonui, E., Manning, C.E., Schauble, E., Macris, C.A., 2009. Spinel–olivine magnesium isotope thermometry in the mantle and implications for the Mg isotopic composition of Earth. Earth and Planetary Science Letters 288, 524–533.

311

Yu, Y., Xu, X.-S., Griffin, W.L., O'Reilly, S.Y., Xia, Q.-K., 2010. H2O contents and their modification in the Cenozoic subcontinental lithospheric mantle beneath the Cathaysia block, SE China. Lithos 126, 1–16. Zaffarana, C., Tommasi, A., Vauchez, A., Grégoire, M., 2014. Microstructures and seismic properties of south Patagonian mantle xenoliths (Gobernador Gregores and Pali Aike). Tectonophysics 621, 175–197. Zanetti, A., Tiepolo, M., Oberti, R., Vannucci, R., 2004. Trace-element partitioning in olivine: modelling of a complete data set from a synthetic hydrous basanite melt. Lithos 75, 39–54. Zindler, A., Jagoutz, E., 1988. Mantle cryptology. Geochimica et Cosmochimica Acta 52, 319–333.