H2O storage capacity of olivine at 5–8 GPa and consequences for dehydration partial melting of the upper mantle

H2O storage capacity of olivine at 5–8 GPa and consequences for dehydration partial melting of the upper mantle

Earth and Planetary Science Letters 345-348 (2012) 104–116 Contents lists available at SciVerse ScienceDirect Earth and Planetary Science Letters jo...

576KB Sizes 3 Downloads 85 Views

Earth and Planetary Science Letters 345-348 (2012) 104–116

Contents lists available at SciVerse ScienceDirect

Earth and Planetary Science Letters journal homepage: www.elsevier.com/locate/epsl

Letters

H2O storage capacity of olivine at 5–8 GPa and consequences for dehydration partial melting of the upper mantle Ardia P.a,n, Hirschmann M.M.a, Withers A.C.a, Tenner T.J.a,b a b

Department of Earth Sciences, University of Minnesota, Minneapolis, MN 55455, USA University Wisconsin-Madison, Madison, WI 53706, USA

a r t i c l e i n f o

a b s t r a c t

Article history: Received 2 January 2012 Received in revised form 21 May 2012 Accepted 27 May 2012 Editor: T. Elliot Available online 20 July 2012

The H2O storage capacities of peridotitic minerals place crucial constraints on the onset of hydrous partial melting in the mantle. The storage capacities of minerals in equilibrium with a peridotite mineral assemblage (‘‘peridotite-saturated’’ minerals) are lower than when the minerals coexist only with fluid because hydrous partial melt is stabilized at a lower activity of H2O. Here, we determine peridotite-saturated olivine H2O storage capacities from 5 to 8 GPa and 1400–1500 1C in layered experiments designed to grow large (  100–150 mm) olivine crystals in equilibrium with the full hydrous peridotite assemblage (melt þ ol þopx þ gar þcpx). The peridotite-saturated H2O storage capacity of olivine at 1450 1C rises from 57 7 26 ppm (by wt.) at 5 GPa to 254 7 60 ppm at 8 GPa. Combining these with results of a parallel study at 10–13 GPa (Tenner et al., 2011, CMP) yields a linear relation applicable from 5 to 13 GPa for peridotite-saturated H2O storage capacity of olivine at 1450 1C, C olivine H2 O ðppmÞ ¼ 57:6ð 7 16Þ  PðGPaÞ169ð 7 18Þ. Storage capacity diminishes with increasing temperature, but is unaffected by variable total H2O concentration between 0.47 and 1.0 wt%. Both of these are as predicted for the condition in which the water activity in the melt is governed principally by the cryoscopic requirement of melt stability for a given temperature below the dry solidus. Measured olivine storage capacities are in agreement or slightly greater than those predicted by a model that combines data from experimental freezing point depression and olivine/melt partition coefficients of H2O (Hirschmann et al., 2009). Considering the temperature along the mantle geotherm, as well as gar=ol px=ol available constraints on garnet/olivine and pyroxene/olivine partitioning of H2O (DH2 O ,DH2 O ), we estimate the peridotite H2O storage capacity in the low velocity zone. The C H2 O required to initiate melting between 150 and 250 km depth is between 270 and 855 ppm. We conclude that hydrous partial melting does not occur at these depths for H2O concentrations (50–200 ppm) typical of the convecting upper mantle sampled by mid-ocean ridge basalts. & 2012 Elsevier B.V. All rights reserved.

Keywords: olivine water storage capacity hydrous partial melting hydrous peridotite upper mantle geotherm

1. Introduction There is now a large body of literature documenting the incorporation of hydrous components in olivine and other minerals at conditions relevant to the upper mantle (e.g. Bai and Kohlstedt, 1992; Kohlstedt et al., 1996; Lu and Keppler, 1997; Withers et al., 1998; Rauch and Keppler, 2002; Lemaire et al., 2004; Mosenfelder et al., 2006; Smyth et al., 2006; Mierdel et al., 2007; Withers and Hirschmann, 2007, 2008, Litasov et al., 2009). Although these studies establish broadly the importance of hydrous components in olivine to the geophysical and petrologic properties of the mantle (Hirth and Kohlstedt, 2003;Mei and

n Correspondence to: IGP ETH Zurich, Clausiusstrasse 25, 8092 Zurich, Switzerland. Tel.: þ 41 76 332 35 37; fax: þ41 44 632 35 37. E-mail address: [email protected] (P. Ardia).

0012-821X/$ - see front matter & 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.epsl.2012.05.038

Kohlstedt, 2000; Poe et al., 2010; Hirschmann et al., 2005, 2009; Ni et al., 2011), there is as yet less agreement regarding the possible concentrations of H2O in olivine and in peridotite under mantle conditions (Mierdel et al., 2007; Hirschmann et al., 2009; Green et al., 2010). When parcels of mantle contain H2O beyond that which can be incorporated in stable minerals at a particular temperature and pressure, they exceed their H2O storage capacity and the excess H2O resides in a melt or fluid. It has been argued that this condition applies to seismic low velocity zones (LVZ) at intermediate depths beneath the oceans (Mierdel et al., 2007; Green et al., 2010) and continents (Till et al., 2012). On the other hand, Hirschmann et al. (2009) argued that peridotite with normal oceanic mantle concentrations of H2O does not exceed its storage capacity in the oceanic LVZ. Alternative explanations for the LVZ include formation of carbonate-rich melts (Presnall et al., 2002; Hirschmann, 2010) or simply the effects of temperature, pressure,

P. Ardia et al. / Earth and Planetary Science Letters 345-348 (2012) 104–116

and dissolved H2O on solid peridotite (Karato and Jung, 1998; Faul and Jackson, 2005; Stixrude and Lithgow-Bertelloni, 2007; Behn et al., 2009). Experiments equilibrating olivine with H2O-rich fluid establish considerable concentrations of hydrous component, structurally bound as an OH- anion but generally discussed in terms of its equivalent weight as H2O (e.g. Kohlstedt et al., 1996). These simple-system experiments differ from conditions in the mantle in a number of respects. Most importantly, the phase assemblage in the experiments consists of olivine, fluid and, in some cases, an additional mineral such as orthopyroxene. At a given temperature and pressure, these simple-system experiments contain a fluid with a comparatively high activity of H2O; such a high H2O activity could not be sustained in the presence of peridotite at the same conditions, as additional partial melting in peridotite would occur, producing a fluid that is more enriched in silicate. The resulting lower H2O activity would reduce the concentrations of H2O in coexisting olivine and other minerals. Thus, recent studies have emphasized the importance of studying the H2O storage capacity of olivine under peridotite-saturated conditions, in which the olivine coexists with the full peridotite mineral assemblage as well as a hydrous fluid or melt (Green et al., 2010; Tenner et al., 2011). In the absence of direct determinations of peridotite-storage capacity, alternative estimates can be made by considering partitioning of H2O between olivine and coexisting silicate melts. Given an estimate of the H2O content in a partial melt of peridotite at a given temperature and pressure, the H2O content of the coexisting olivine can be calculated (Hirschmann et al., 2009; Hirschmann, 2010; O’Leary et al., 2010). Recently, estimates of the H2O storage capacity of olivine and of peridotite at intermediate upper mantle depths similar to those in the asthenosphere and intracratonic LVZ (Green et al., 2010; Ni et al., 2011) have been formulated from different approaches. Mierdel et al. (2007) used simple system solubility experiments to estimate the aggregate storage capacity of peridotite. They inferred that the sum of simple system storage capacities of peridotite minerals goes through a minimum of approximately 800 ppm at a depth of 150 km and speculated that peridotitesaturated storage capacities, commensurately diminished by the reduced H2O activity of hydrous partial melt, could produce hydrous partial melting in the asthenosphere. This estimate was based on the assumption that near-solidus partial melts have approximately 5 wt% H2O. Green et al. (2010) conducted experiments determining directly peridotite-saturated storage capacities at 1.5–6 GPa and 970–1350 1C. In agreement with Mierdel et al. (2007), Green et al. (2010) concluded that mantle with reasonable concentrations of H2O ( 180 ppm) contains a small amount of hydrous partial melt at conditions in the asthenosphere. In contrast to Mierdel et al. (2007), they inferred near-solidus partial melts with approximately 30 wt% H2O, although this inference was not based directly on their experiments. Hirschmann et al. (2009) employed mineral/melt partition coefficients to argue, in contrast to Mierdel et al. (2007) and Green et al. (2010), that normal upper mantle compositions do not form hydrous partial melts in the asthenosphere. Hirschmann et al. (2009) used experimental data on the influence of H2O on peridotite partial melting temperatures to infer that near-solidus partial melts of peridotite along the asthenosphere geotherm should have 5 wt% H2O, in agreement with Mierdel et al. (2007). However, the comparatively large value of peridotite/melt partition coefficients inferred from experiments (Aubaud et al., 2004, 2008; Hauri et al., 2006; Tenner et al., 2009; O’Leary et al., 2010) lead Hirschmann et al. (2009) to conclude that along the mantle geotherm, near-solidus partial melts approaching 5 wt.%

105

H2O would be in equilibrium with peridotitic residues having 4300 ppm H2O, meaning that normal oceanic mantle with o200 ppm H2O could not produce such melts (unless melts are stabilized in part by other volatile components such as CO2; Hirschmann, 2010). The arguments and calculations of Hirschmann et al. (2009) conflict with the experimental results of Green et al. (2010), who inferred that highly H2O-rich ( 30 wt%) partial melts of peridotite can be in equilibrium with comparatively dry (180 ppm) peridotitic residua. This disagreement can be quantified in terms of effective peridotite/melt partition coefficients. Whereas Hirschmann et al. (2009) used experimental mineral/melt partiper=melt tion coefficients to infer DH2 o between 0.004 and 0.009 at depths of 60–200 km, the experiments of Green et al. (2010) per=melt imply DH2 o equal to 0.0006 under these conditions. The origin of this disagreement is not understood. Here we present new experimental determinations of the H2O storage capacity of olivine in equilibrium with peridotite and hydrous melt at 5–8 GPa. The experiments employ a layered design similar to that used by Green et al. (2010) and, at higher pressures, Tenner et al. (2011). The layered geometry facilitates growth of large olivine crystals, amenable to accurate microbeam analysis of H, that are in equilibrium with a hydrous melt for which the composition and H2O activity is fixed by coexistence with the full peridotite assemblage at the temperature and pressure of interest. Hydrogen concentrations in olivine are measured using SIMS using established low-blank techniques (Koga et al., 2003; Aubaud et al., 2004; Hauri et al., 2006; Tenner et al., 2009). We apply these measurements to constrain the H2O storage capacity of olivine and of peridotite at conditions prevailing in the asthenosphere.

2. Experimental and analytical methods Measurement of peridotite-saturated water storage capacities of nominally anhydrous minerals (NAM) requires experimental synthesis of mineral grains equilibrated with peridotite residual minerals and melts and of sufficient size to be analyzed by microbeam methods. To achieve this we employed a layered geometry, in which 2/3 of the experimental capsule consists of a hydrous peridotite assemblage and 1/3 is a layer that will produce a monomineralic layer of olivine that produces large crystals over the course of the experiment (Fig. 1). This method has been employed previously by Green et al. (2010) and Tenner et al. (2011). Initial experiments were conducted with natural dry San Carlos olivine powder with grain size of 1 mm together with synthetic oxide/hydroxide mixture whose composition is similar to hydrous peridotite with 1% H2O. Both materials had an Mg# of 89.6 (Table 1). The synthetic hydrous peridotite was itself prepared by mixing a synthetic hydrous olivine from SiO2, MgO, FeO, and Mg(OH)2 with 10% H2O and synthetic anhydrous peridotite (KLB1 oxide, Davis et al., 2009) (Table 1). A flaw of this approach is that the average H2O concentration of the experimental charge varied according to the proportions of hydrous peridotite and dry olivine powder and these proportions are not easily controlled. Accordingly, for most of the experiments we adopted a different strategy, employing hydrous peridotite with a layer of hydrous olivine, each of which contained the same average H2O concentration, in which case the average H2O content of the charge does not depend on with the proportions of olivine and peridotite layers. The hydrous olivine was prepared by mixing oxides (SiO2, MgO, FeO) and hydroxide (Mg(OH)2) in exact proportion to obtain the desired Mg# (100  molar Mg/ (MgOþFeO)) of 89.6, and water contents of 0.5, 0.75 and 1.0 wt% (Table 1). The hydrous peridotite compositions were prepared by mixing natural anhydrous KLB1 peridotite and synthetic peridotite

106

P. Ardia et al. / Earth and Planetary Science Letters 345-348 (2012) 104–116

M462 o. i.

1/3

after

before

2/3

peridotite-

in wt

ol- layer

M473 big ol (px, gar +melt) peridotite (ol, opx, cpx, gar, melt)

melt pool

Ol 400 µm

quench melt

AuPd capsule

Ol

P, T, t 200 µm

400 µm

M462 BSE Fig. 1. (a) Schematic illustration of the experimental capsules. On the left, prior to experiments, capsules of 2.0 mm and between 1.4 and 2.0 mm in length are filled with 1/3 (by weight) fine-grained olivine and 2/3 peridotite starting material (Table 1). The sealed capsule is placed with the olivine layer close to the thermocouple. On the right, after the experiment, a coarse-grained near-monomineralic olivine layer and fine grained peridotite layer result. (b) Representative back-scattered electron (BSE) image of experiment M473 with 0.47 wt% bulk H2O (8 GPa, 1450 1C) showing olivine and peridotite layers with distinct grain sizes. Quenched melt pools are evident on the left and right margins of the charge. Image (c) and (d) (M462: 5 GPa, 1500 1C) are, respectively, reflected light optical and BSE images taken in the same position, highlighting the large olivine crystals in the near-monomineralic layer and the interstitial hydrous melt surrounding the crystals.

prepared from oxides and hydroxides (Tenner et al., 2011) with 1 wt% water, producing batches with 0.47 and 0.75 wt% of water. Starting materials were packed into AuPd capsules using a small press fashioned from a steel die, steel filler rods and a vice, creating well-defined hard-pressed layers of peridotite and olivine. The weights of each layer were determined with a semimicrobalance, with target proportions of olivine and peridotite amounting to 1/3 and 2/3 by weight, respectively (Fig. 1). Filled capsules were crimped or coned and then welded shut. High Mg#s of residual phases suggested that initial experiments lost appreciable Fe, and so subsequent experiments were conducted with Fe-doped capsules. These AuPd capsules were prepared by melting of a MORB basalt under reduced condition at 1600 1C, and the residual silicate was removed with an HF bath (details are given in Stanley et al., 2011). Experiments in Fe-doped capsules experienced reduced Fe-loss (Table 2). The experiments were conducted at 5–8 GPa and 1400– 1500 1C for 24 h in a 1000 t Walker-type multi-anvil press, using cast MgO  Al2O3  SiO2  Cr2O3 octahedra with integrated gasket fins and WC anvils with 12 mm truncations. Straight-walled graphite furnaces were used for all experiments and temperature was controlled with type D (W97Re3/W75Re25) thermocouples that were located directly above the capsule on the olivine layer side. The P–T uncertainties of experiments are believed to be 70.3 GPa and 715 1C, respectively, based on the calibration described in Dasgupta et al. (2004). Oxygen fugacity is not controlled in these unbuffered experiments, but use of starting materials in which Fe is initially principally Fe2 þ and AuPd capsules for which modest alloying of Fe metal occurs suggests conditions in which iron in the silicate melt remains chiefly ferrous (Barr and Grove, 2010). Experiments at 8 GPa by Withers and Hirschmann (2008) show that the H2O storage capacity of olivine is insensitive to fO2 at high pressures, and so oxygen fugacity is not believed to be a critical parameter influencing the H content of olivine in the present experiments.

2.2. Major element composition: EMPA Major element concentrations of all phases were analyzed by electron microprobe using a JEOL JXA8900R (EMPA) (Table 2). Analyses were performed with a 15 kV accelerating voltage, and counting times of 10 s for the peak and 5 s for the background for all elements. Standards consisted of natural olivine, enstatite, diopside, omphacite, hornblende, pyrope, quartz, basaltic glasses, chromite, Mn-hortonolite, and potassium–feldspar (Jarosewich et al., 1980), used in various, but consistent, configurations to quantify the oxide contents in olivine, pyroxenes, garnet, and melt residue. Crystalline phases and quenched melt were analyzed with a 10 nA beam current and a fully focused beam. Na and K were always counted first on their respective spectrometers to minimize the effects of alkali migration. 2.3. Secondary ion mass spectrometry (SIMS) Water concentrations in olivine were determined by low-H background SIMS (Koga et al., 2003) using the Cameca 6f ion microprobe at Arizona State University. After electron microprobe analyses, C coats were removed from sample mounts by polishing and surfaces were cleaned using successive baths of acetone and ethanol. This process also removed residual Crystal Bond (Aubaud et al., 2004). Samples were mounted in an indium holder (Koga et al., 2003; Aubaud et al., 2004; Hauri et al., 2006; Tenner et al., 2009) and Au coated. Analyses were performed following the method described by Tenner et al. (2009), using olivine standards that were synthesized and characterized by Withers et al. (2011) and whose H2O concentrations were established by ERDA (Withers et al., submitted for publication). Based on an average of 13 measurements on synthetic forsterite over the course of the session, the H background was the equivalent of 1372 ppm H2O in olivine.

3. Results 2.1. Analytical techniques

3.1. Textures

Experimental charges were sectioned longitudinally using a 50 mm tungsten wire saw and polished using 30–0.5 mm diamond lapping films. To minimize possible contamination effects for SIMS analyses, impregnation with Crystal Bond (Armeco Corp.) was avoided when possible, but was employed in two experiments.

The exposed experimental charges revealed a nearly monomineralic olivine layer occupying ca. 1/3 of the upper side of the capsule and a peridotite layer (Fig. 1). The olivine grain sizes varied from 30 mm in the peridotite layer to 200 mm in the monomineralic layer. Phases in the peridotite layer included

Table 1 Composition of the starting material; peridotite and olivine compositions and resulting composition for the experiments. Olivine starting composition San Carlos

Peridotite starting composition Synthetic Olivine 0.50

0.75

1.00

Nat.

Synthetic peridotite

KLB-1

0.45

0.75

1.00

SiO2 TiO2 Al2O3 Cr2O3 FeO MnO MgO CaO Na2O K2O H2O

40.86

40.60

40.50

40.39

9.53

10.09

10.07

10.05

50.07 0.06

48.81

48.68

48.56

Total

ratio in wt. Ol S.Carlo Per 1.00n

M435 0.456 0.544

M439 0.348 0.652

M442 0.337 0.663

0.50

0.75

1.00

100

100

100

ratio in wt. Ol 0.5n Per 0.45n

M446 0.302 0.698

M453 0.335 0.665

M454 0.322 0.678

M456 0.318 0.682

44.84 0.11 3.51 0.32 8.20 0.12 39.52 3.07 0.30 0.02

44.46 0.13 3.52 0.29 8.18 0.11 39.35 3.22 0.27 0.02 0.45

44.20 0.14 3.53 0.27 8.16 0.11 39.24 3.33 0.24 0.03 0.75

43.99 0.14 3.53 0.25 8.15 0.11 39.15 3.41 0.23 0.03 1.00

100.01

100.00

100.00

100.00

M460 0.334 0.666

M462 0.324 0.676

M473 0.332 0.668

ratio in wt. Ol 0.75n Per 0.75n

M464 0.254 0.746

Ol 1.00n Per 1.00n

M466 0.335 0.665

(B) Bulk composition in wt% and weight-ratios of the two reagent-derived components used for the experiments. SiO2 TiO2 Al2O3 Cr2O3 FeO MnO MgO CaO Na2O K2O H2O

42.47 0.08 1.92 0.14 8.76 0.06 44.01 1.88 0.12 0.01 0.54

42.83 0.09 2.30 0.16 8.61 0.07 42.86 2.25 0.15 0.02 0.65

42.86 0.10 2.34 0.17 8.60 0.07 42.74 2.28 0.15 0.02 0.66

43.29 0.09 2.46 0.20 8.76 0.08 42.21 2.25 0.19 0.02 0.47

43.16 0.08 2.34 0.19 8.82 0.08 42.52 2.14 0.18 0.02 0.47

43.21 0.09 2.39 0.20 8.79 0.08 42.40 2.19 0.18 0.02 0.47

43.23 0.09 2.40 0.20 8.79 0.08 42.36 2.20 0.18 0.02 0.47

43.17 0.08 2.34 0.19 8.82 0.08 42.51 2.15 0.18 0.02 0.47

43.20 0.08 2.38 0.20 8.80 0.08 42.42 2.18 0.18 0.02 0.47

43.17 0.08 2.35 0.19 8.81 0.08 42.49 2.15 0.18 0.02 0.47

43.26 0.10 2.63 0.20 8.65 0.08 41.65 2.48 0.18 0.02 0.75

42.79 0.10 2.35 0.17 8.79 0.07 42.30 2.27 0.15 0.02 1.00

Total

100.00

100.00

100.00

100.00

100.00

100.00

100.00

100.00

100.00

100.00

100.00

100.00

Mg # capsule

90.0 AuPd

89.9 AuPd

89.9 AuPd

89.6 AuPdFe

89.6 AuPdFe

89.6 AuPdFe

89.6 AuPdFe

89.6 AuPdFe

89.6 AuPdFe

89.6 AuPdFe

89.6 AuPdFe

89.6 AuPdFe

n

P. Ardia et al. / Earth and Planetary Science Letters 345-348 (2012) 104–116

(A) Olivine and peridotite composition used for the starting reagents.

Water concentration of the synthetic olivine (Ol) and peridotite (Per).

107

108

Table 2 Chemical composition of the phases present in the experiments. Exp.

TiO2

Al2O3

Cr2O3

FeO

MnO

MgO

41.81 54.23 n.a. 43.13

1.11 1.20

0.01 0.09

0.04 0.09

0.11 3.81

0.07 1.64

0.02 0.26

0.05 0.07

7.53 5.73

0.70 0.67

0.07 0.12

0.04 0.07

50.80 33.55

0.77 0.76

0.84

0.26

0.06

22.08

0.47

1.84

0.13

6.05

0.69

0.20

0.06

22.66

1.43

41.15 55.83 54.02 41.36

0.28 0.57 0.31 0.13

0.01 0.02 0.05 0.20

0.09 0.07 0.07 0.03

0.09 1.48 1.81 22.87

0.09 0.07 0.15 0.19

0.04 0.20 0.34 1.35

0.04 0.06 0.06 0.11

6.85 4.33 3.12 6.48

0.27 0.12 0.69 0.10

0.10 0.11 0.10 0.24

0.03 0.03 0.04 0.04

51.16 35.40 22.42 21.13

0.47 0.26 5.59 0.33

41.27 56.22 54.56 39.64

0.38 0.18 0.32 1.96

-0.03 0.03 0.03 0.26

0.06 0.09 0.05 0.08

0.10 1.03 1.59 19.26

0.08 0.04 0.07 0.20

0.03 0.17 0.31 1.46

0.05 0.06 0.07 0.08

6.98 4.11 3.06 5.93

0.41 0.24 0.25 0.44

0.10 0.08 0.11 0.19

0.05 0.05 0.04 0.05

51.10 35.82 21.70 20.17

0.69 0.23 0.15 1.23

41.39 56.68 n.a. 43.75

0.51 1.30

0.04 0.07

0.07 0.04

0.11 3.31

0.04 1.19

0.07 0.38

0.05 0.05

6.99 4.81

0.34 0.60

0.07 0.09

0.04 0.04

51.33 33.50

0.43 0.83

0.60

0.15

0.05

23.31

0.13

1.80

0.16

4.94

0.18

0.16

0.04

23.14

0.24

41.51 55.35 53.85 43.08

0.25 0.72 2.52 0.44

 0.02 0.04 0.09 0.16

0.06 0.12 0.14 0.09

0.12 2.01 3.36 22.78

0.02 0.07 2.22 0.22

0.07 0.26 0.51 1.80

0.05 0.07 0.23 0.09

5.70 3.48 2.10 4.44

0.21 0.11 0.42 0.20

0.07 0.07 0.08 0.13

0.04 0.03 0.04 0.04

51.92 36.30 21.73 24.40

0.39 0.31 5.34 0.27

41.94 56.39 n.a. 43.80

0.53 0.26

 0.02  0.01

0.08 0.07

0.08 1.46

0.04 0.05

0.04 0.21

0.05 0.07

4.99 3.91

0.14 0.25

0.10 0.10

0.04 0.04

52.61 35.83

0.47 0.41

0.47

0.17

0.04

23.01

0.21

1.75

0.14

5.11

0.51

0.22

0.05

22.38

0.61

40.92 58.46 54.53 43.78

0.60 0.33 1.13 0.54

0.01 0.00 0.07 0.30

0.06 0.01 0.07 0.07

0.09 1.43 2.97 22.09

0.04 0.84 1.21 0.61

0.04 0.16 0.47 1.82

0.06 0.17 0.10 0.13

8.65 5.53 4.10 6.51

0.30 0.80 0.29 0.26

0.07 0.06 0.10 0.18

0.04 0.02 0.04 0.03

50.13 33.45 21.90 22.49

0.48 0.37 0.90 0.35

41.16 56.86 55.38 42.85

0.33 0.44 0.35 0.35

 0.01 0.01 0.08 0.21

0.06 0.09 0.06 0.12

0.11 0.70 1.85 21.97

0.05 0.24 0.06 0.26

0.05 0.05 0.31 1.58

0.07 0.07 0.09 0.16

8.21 5.42 3.73 6.89

0.33 0.29 0.17 0.25

0.05 0.03 0.08 0.15

0.04 0.03 0.04 0.03

50.51 36.26 22.46 22.15

0.58 0.47 0.30 0.39

40.31 55.52 54.33 42.25

0.35 0.21 0.53 0.47

0.01  0.02 0.03 0.21

0.07 0.05 0.07 0.08

0.06 1.08 1.75 21.30

0.02 0.04 0.06 0.64

0.03 0.21 0.31 1.60

0.06 0.05 0.07 0.11

8.67 5.36 4.06 7.40

0.12 0.19 0.09 0.18

0.04 0.06 0.08 0.17

0.03 0.04 0.03 0.05

49.68 34.81 22.33 21.92

0.49 0.45 0.31 0.73

40.87 55.97 52.68 42.75

0.50 0.31 1.59 0.52

0.01 0.08 0.27 0.29

0.08 0.09 0.20 0.10

0.07 1.24 5.24 21.35

0.02 0.17 3.05 1.25

0.01 0.16 0.55 1.83

0.05 0.05 0.23 0.18

8.14 4.59 3.31 6.72

0.51 0.16 0.53 0.32

0.07 0.08 0.07 0.15

0.04 0.03 0.03 0.05

50.53 35.27 19.63 22.87

0.36 0.22 3.70 1.89

41.85 56.68 53.85 43.34

0.42 0.30 2.52 0.30

0.02 0.04 0.09 0.28

0.08 0.08 0.14 0.10

0.08 0.74 3.36 21.45

0.08 0.05 2.22 0.23

0.00 0.11 0.51 1.72

0.04 0.05 0.23 0.19

3.93 2.58 2.10 4.17

0.15 0.13 0.42 0.42

0.07 0.08 0.08 0.18

0.03 0.03 0.04 0.04

53.38 37.76 21.73 23.86

0.53 0.33 5.34 0.28

P. Ardia et al. / Earth and Planetary Science Letters 345-348 (2012) 104–116

M446 Ol OPx CPx Gar M435& Ol OPx CPx Gar M439& Ol OPx CPx Gar M453 Ol OPx CPx Gar M462 Ol OPx CPx Gar M442& Ol OPx CPx Gar M454 Ol OPx CPx Gar M466 Ol OPx CPx Gar M464 Ol OPx CPx Gar M456 Ol OPx CPx Gar M460 Ol OPx CPx Gar

SiO2

M473 Ol OPx CPx Gar Exp.

CaO

0.27 0.43

 0.01 0.04

0.07 0.09

0.06 1.13

0.03 0.11

0.03 0.21

0.04 0.07

7.21 4.37

0.39 0.31

0.07 0.07

0.03 0.02

51.00 35.37

0.42 1.02

0.68

0.18

0.11

21.95

0.35

1.75

0.25

5.93

0.49

0.16

0.04

22.24

0.55

Na2O

K2O

Total

#

Mg#

0.94 0.91

0.14 0.41

0.02 0.18

0.03 0.04

0.00 0.02

0.01 0.02

100.49 99.10

0.51 0.23

16 9

5.08

0.49

0.02

0.02

 0.01

0.01

101.32

0.00

8

0.11 1.35 15.42 5.26

0.04 0.18 6.91 0.09

0.02 0.09 0.55 0.02

0.02 0.03 0.23 0.01

0.02 0.03 0.03 0.03

0.01 0.03 0.02 0.02

99.57 98.83 97.86 98.95

0.64 0.67 0.89 0.38

34 13 6 10

0.93 0.94 0.93

0.11 1.38 16.69 5.34

0.07 0.07 0.28 0.55

0.06 0.17 0.81 0.50

0.03 0.04 0.06 0.17

0.03 0.02 0.03 0.12

0.02 0.01 0.01 0.08

99.75 99.04 98.89 92.87

0.61 0.29 0.48 3.23

9 10 10 3

0.93 0.94 0.93

0.14 1.46

0.03 0.46

0.03 0.17

0.03 0.03

0.01 0.00

0.02 0.01

100.18 100.46

0.66 1.42

33 6

0.93 0.93

4.61

0.39

0.01

0.02

0.00

0.02

101.88

0.58

11

0.10 1.07 15.84 2.88

0.03 0.13 4.58 0.14

0.01 0.15 1.36 0.03

0.02 0.03 0.47 0.03

0.02 0.03 0.04 0.02

0.02 0.02 0.03 0.03

99.51 98.76 98.95 99.72

0.65 0.75 2.39 0.79

13 12 16 8

0.94 0.95 0.95

0.08 1.34

0.02 0.09

0.14 0.09

0.08 0.03

0.02 0.01

0.02 0.02

99.99 99.33

0.66 0.51

20 14

0.95 0.94

5.07

0.47

0.10

0.06

0.01

0.02

101.63

0.44

8

0.13 1.72 14.11 4.64

0.02 0.13 0.99 0.22

0.02 0.30 1.20 0.04

0.02 0.11 0.12 0.04

0.02 0.00 0.01 0.01

0.02 0.02 0.02 0.02

100.08 101.10 99.45 101.87

0.81 0.31 0.72 0.48

31 2 20 7

0.91 0.92 0.90

0.09 0.82 15.00 4.40

0.02 0.42 0.31 0.19

0.02 0.14 0.99 0.05

0.03 0.06 0.06 0.02

0.02 0.02 0.03 0.02

0.02 0.02 0.02 0.01

100.21 100.31 99.93 100.29

0.77 0.56 0.64 0.63

12 14 13 12

0.92 0.92 0.91

0.08 1.79 14.83 4.44

0.01 0.48 0.21 0.27

0.01 0.17 0.91 0.05

0.03 0.04 0.04 0.04

0.02 0.04 0.03 0.02

0.02 0.02 0.02 0.01

98.91 99.02 98.67 99.35

0.47 0.34 0.63 0.59

10 5 19 14

0.91 0.92 0.91

0.09 1.41 16.51 4.61

0.02 0.10 2.59 0.25

0.02 0.30 1.30 0.15

0.02 0.03 0.32 0.09

0.02 0.03 0.03 0.02

0.02 0.02 0.03 0.03

99.84 99.14 99.60 100.74

0.59 0.32 0.95 0.51

14 8 8 11

0.92 0.93 0.91

0.07

0.03

0.06

0.05

0.03

0.02

99.50

0.68

11

0.96

0.79

0.76

0.79

0.79

0.85

0.77

0.78

0.78

0.78

0.81

gr/(prp þalm þ gr)

P (GPa)

T (1C)

H2Obulk (wt%)

5

1400

0.47

5

1400

0.54

5

1400

0.65

5

1450

0.47

5

1500

0.47

6

1400

0.66

6

1450

0.47

6

1450

0.75

6

1450

1

7

1450

0.47

8

1450

0.47

0.13

0.14

0.15

0.11

0.07

0.13

0.12

0.11

0.11

0.12 109

0.13 1.11

prp/(prp þ alm þgr)

P. Ardia et al. / Earth and Planetary Science Letters 345-348 (2012) 104–116

M446 Ol OPx CPx Gar M435& Ol OPx CPx Gar M439& Ol OPx CPx Gar M453 Ol OPx CPx Gar M462 Ol OPx CPx Gar M442& Ol OPx CPx Gar M454 Ol OPx CPx Gar M466 Ol OPx CPx Gar M464 Ol OPx CPx Gar M456 Ol OPx CPx Gar M460 Ol

41.45 56.38 n.a. 42.71

P. Ardia et al. / Earth and Planetary Science Letters 345-348 (2012) 104–116

1450

T (1C)

#

AuPd capsule not doped with Fe prior the experiment.

3.2. Phase compositions

&

13 0.53 98.94 0.02 0.03 0.07 0.35

0.03

0.02 0.19 0.02 0.27

0.08 1.40 n.a. 3.90

0.02 0.04

0.02 0.01

0.01 0.01

99.92 99.17

0.39 0.51

18 9

0.93 0.94

0.79

0.10

0.11 0.84 0.96 0.95 15 15 13 0.59 2.39 0.58 99.40 98.95 99.51 0.02 0.03 0.01 0.03 0.04 0.03 0.03 0.47 0.03 0.22 1.36 0.13 0.09 4.58 0.10 1.17 15.84 4.35

OPx CPx Gar M473 Ol OPx CPx Gar

Number of analyses considered for the average; Mg#: Mg number (MgO/40.3/(MgO/40.3045 þ FeO/71.8); prp: pyrop; alm: almandine; gr: grossular. Standard error in italic.

8

P (GPa) gr/(prp þalm þ gr) prp/(prp þ alm þgr) Mg# #

Total K2O Na2O CaO Exp.

Table 2 (continued )

olþopxþ gt7 cpx, and quenched silicate melt, which was recognized in every experiment (Fig. 1). Olivine grains in this layer are homogenously distributed and are finer-grained (  3 to 20 mm) than those in the monomineralic layer. Garnet occurs homogeneously distributed in the peridotite layer and in small grains in the olivine layer, and forms grains from o1 to  10 mm. OPx is also homogeneously distributed in the peridotite layer in all the experiments, with grains from 5 to 20 mm, and in some experiments smaller OPx grains are present in the Ol layer. CPx was identified in most experiments, concentrated at the cold end of the capsule most distal to the olivine layer, but in four experiments EDS and SEM observation did not reveal any. It may be present in regions of those charges that are not exposed. Microprobe and SEM observation confirmed the presence in all experiments of quenched melt (Fig. 1d), consisting of fine-grained aggregates of minerals and possibly amorphous phases. Textures of quenched melt vary in the experiments. Melts in experiments with comparatively low melt fractions occur interstitially and in small pools in both the peridotite and olivine layers. Experiments with higher melt fractions also contain larger melt pools on the sides of the capsule (Fig. 1c).

0.47

H2Obulk (wt%)

110

Compositions of minerals and quenched melt are given in Table 2. The Mg#s for olivine, clinopyroxene and orthopyroxene range between 91 and 96. The extreme high values are characteristic of the experimental charges that experienced higher Fe loss through annealing with the AuPd capsule during the experiment. Phase compositions do not vary spatially within the capsule, and olivine, orthopyroxene and garnet grains in the peridotite layer and in the monomineralic layer have the same composition, confirming chemical communication between the layers, that the thermal gradient across the charge was too small to influence mineral compositions, and that Fe loss is homogeneous and not concentrated along the capsule walls. Olivine compositions (Table 2) show systematic variations with total water content, and pressure. The Mg#s increase with pressure, but are also influenced by non-systematic effects associated with Fe-loss to capsule materials. Minor elements, Al2O3, Cr2O3 and CaO decrease with pressure and bulk water content. For example, at the same temperature and for equal bulk water content, Al2O3 contents diminish from 0.1170.04 to 0.0670.03 wt% from 5 to 8 GPa. Similarly, Cr2O3 and CaO concentrations decrease from 0.07 70.05 and 0.1470.03 wt% at 5 GPa down to 0.03 70.04 and 0.07 70.03 wt%, respectively. Variations in Mg#s of pyroxenes and garnets correlate to those of olivine. Alumina concentrations in orthopyroxene vary between 1.6 and 3.9 wt% and diminish with increasing pressure, consistent with the trend for pyroxenes in equilibrium with peridotite near its solidus (Hirschmann et al., 2009) (Fig. 2(a). Clinopyroxenes range from 1.9 to 5.2 wt% Al2O3 but have high uncertainties and are more variable, in part owing to the small number of analyzable grains present in the charges. At 5–6 GPa, their Al2O3 contents are consistent with the near-solidus trend; at higher pressure they may be more Al2O3-rich, but not outside of analytical uncertainty (Fig. 2a). We attempted to characterize the mean composition of the quenched melt by analyzing a large number of microprobe points and filtering them for those that may plausibly be in Fe–Mg equilibrium with olivine (see Methods section, Humayun et al., 2010; Davis et al., 2011). After careful review, we surmised that our efforts at averaging and filtering microprobe analysis of the comparatively small, and therefore severely quench modified melt pools did not successfully retrieve equilibrium melt compositions.

P. Ardia et al. / Earth and Planetary Science Letters 345-348 (2012) 104–116

3.3. SIMS

Table 3 Measured water concentration in olivine.

Concentrations of H in peridotite-saturated olivine coexisting with hydrous melts, reported as ppm H2O by weight, C ol H2 o , (Table 3), increase with pressure at 1450 1C, rising from 57726 ppm water in olivine at 5 GPa to 254760 ppm at 8 GPa (Fig. 3a). Within analytical error, experiments conducted with ol 0.47 wt% H2O do not demonstrate a significant change in C ol H2 o C H 2 o over the relatively narrow temperature interval investigated (Fig. 4a), though a modest reduction is evident in similar experiments at 12 GPa (Tenner et al., 2011). Variations in bulk H2O content from 0.47 to 1.0 wt% have no discernable influence on observed H2O storage capacities within analytical uncertainty (Fig. 5).

OPx

CPx

this study:

OPx

Exp.

Mg#

H2OOl (ppm)

St. dev.

M446 M435& M439& M453 M462 M442& M454 M466 M464 M456 M460 M473

0.94 0.93 0.93 0.93 0.94 0.95 0.91 0.92 0.91 0.92 0.96 0.93

129.7 103.3 145.2 57.0 105.8 n.a. 207.4 262.0 288.7 110.8 236.8 272.2

33.7 18.0 22.1 26.1 19.0 77.6 34.1 100.7 38.1 38.1 81.8

P (GPa)

T (1C)

H2Obulk (wt%)

5 5 5 5 5 6 6 6 6 7 8 8

1400 1400 1400 1450 1500 1400 1450 1450 1450 1450 1450 1450

0.47 0.54 0.65 0.47 0.47 0.66 0.47 0.75 1.00 0.47 0.47 0.47

st. dev.: standard deviation of analyses of multiple crystals from the same experiment; n.a.: not available.

CPx

&

12

AuPd capsule not doped with Fe prior the experiment.

4. Discussion

10 Al2O3 (wt%)

111

Al2O3 in Px = f [P]

8 6 4 2

The effect of pressure on the storage capacity of water in olivine in equilibrium with peridotite and melt can be evaluated at 1450 1C from combined consideration of the experiments in this study at 5–8 GPa and those obtained from similar experiments by Tenner et al. (2011) from 10 to 13.4 GPa (Fig. 3a). The data, taken together, define a linear trend such that C olivine H2 o ðppmÞ ¼ 61:9ð 7 5:6ÞP-225ð 744Þ

ð1Þ

0 0

2

4

6 8 pressure (GPa)

10

12

20

15

with an r2 of 0.78. With increasing temperature, the water storage capacity of olivine in equilibrium with peridotite diminishes (Fig. 4b). Increases in temperature stabilize hydrous partial melts of peridotite with greater proportions of silicate and correspondingly smaller dissolved H2O concentrations. These in turn reduce the activity of H2O in the melt (amelt H2 O ) as well as the H2O fugacity, 0

0

f H2 o ,as f H2 o ¼ amelt H2 o f H2 o ; where f H2 O is the fugacity of pure H2O at

DH O 2

Px/Ol

0

10 DPx/Ol H2O = f [Al2O3]

5

the T and P of interest. Additionally, f H2 o diminishes because f H2 O decreases with temperature (Pitzer and Sterner, 1994). These effects are partly offset by the increase in solubility of H2O in olivine with increasing temperature, which is evident chiefly at low pressures and temperatures, where the composition of fluid coexisting with olivine is nearly pure H2O (Zhao et al., 2004).

0 0

2

4

6 Al2O3 (wt%)

8

10

20 Px/Ol DH 2O = f [P]

2

D Px/Ol H O

15

10 +/- from panel (a) 5

0 0

2

4

6 8 pressure (GPa)

10

12

px=ol

Fig. 2. Relationships between pyroxene/olivine partitioning of H2O, DH2 O , and Al2O3 contents of pyroxenes in peridotite at temperatures similar to the mantle geotherm as a function of pressure. In (a) Al2O3 concentrations of opx (circle) and cpx (triangle) from high pressure experiments are shown as a function of their pressure, black symbols represent data from Al2O3 free systems. Larger symbols indicate our analyses (Table 2). Data are from this study (larger symbols) as well as from the compilation of Hirschmann et al. (2009) and from Tenner et al. (2011), and Withers et al. (2011). The curve is the proposed relationship between the Al2O3 content of peridotitic pyroxene and pressure along the mantle geotherm from Hirschmann et al. (2009) shown with the parametrization uncertainly (grey px=ol line). In (b) experimentally-determined values of DH2 O from Aubaud et al. (2004, 2007), Hauri et al. (2006), O’Leary et al. (2010), Tenner et al. (2011), and Withers et al. (2011) are plotted vs. the Al2O3 content of the pyroxenes. The linear px=ol relationship ignores data with high values of DH2 O from experiments with low Al2O3 in pyroxene from O’Leary et al. (2010), as olivine in these experiments had H2O concentrations near the limits of detection (20–23 ppm). Combining the relationship for pyroxene Al2O3 content as a function of pressure in (a) with that px=ol px=ol between pyroxene Al2O3 and DH2 O from (b) produces a predicted trend for DH2 O as a function of pressure near the mantle geotherm and this is shown in (c) together with available experimental determinations. Note that low values of px=ol DH2 O at 3 and 6 GPa are for Al2O3-free pyroxene (Withers et al., 2011) and are therefore expected to be below the predicted trend for peridotite.

112

P. Ardia et al. / Earth and Planetary Science Letters 345-348 (2012) 104–116

200

600

water in olivine (ppm)

water in Ol (ppm)

800

61.9(+/-5.6) x P - 225 (+/-44)

400

200

data at 1450 °C this study Tenner et al., 2011 linear fit

0

5 GPa, 0.47wt% bulk water

150 100

50

0 1400

600

1500

2000

500

8 GPa

400

39.1

300

xP

- 66 1500°C

1450°C

200 1400°C

100

data from this study recalculated; Tenner et al., 2011 linear fit

0 5

6

7

8 9 10 Pressure (GPa)

11

12

13

Fig. 3. (a) Experimental determinations of H2O storage capacity at 1450 1C in peridotite-saturated olivine as a function of pressure from this study and from Tenner et al. (2011)10–13 GPa. (b) Peridotite-saturated olivine storage capacity of water along the mantle geotherm (Stixrude and Lithgow-Bertelloni, 2007). Data from this study at 5 GPa, 1400 1C and 8 GPa, 1450 1C were considered for the calculation together with the calculated values at 13.4 GPa, 1500 1C from Tenner et al. (2011). Error bars are calculated similar to the previous one.

Because peridotite is a complex chemical system consisting of numerous chemical components, its thermodynamic variance is comparatively high. Consequently, the H2O storage capacity of peridotite-saturated olivine at a given temperature and pressure is not strictly fixed. Variations in melt and mineral compositions owing to the effects of bulk composition or to changes in melt fraction associated with variable amounts of H2O or other fluxing agents can influence the activity of H2O and hence the concentration of H2O in olivine. However, our experiments spanning a range of H2O bulk concentrations at fixed temperature and pressure show no discernable trend in H2O concentration in olivine (Fig. 5), and this suggests that at a given temperature and pressure, the peridotite-saturated H2O storage capacity of olivine is not a strong function of bulk composition or the total H2O available. As noted in the introduction increases in total H2O content beyond the storage capacity produce enhanced melt fractions at approximately the same H2O activity, and the H2O concentration in the olivine coexisting with this melt varies little. Based on this reasoning, the H2O storage capacities measured for bulk H2O concentrations between 0.5 and 1 wt% bulk H2O may be quite similar to those at bulk H2O concentrations approaching that of the storage capacity itself. Consequently, we assume that the H2O storage capacities that we measure for peridotite with 0.5–1 wt% H2O are also applicable at bulk H2O concentrations near the storage capacity itself (i.e., at low total H2O

water in olivine (ppm)

water in Ol (ppm)

1450

1500 5.65 GPa 1000

6 GPa 500 8 GPa 6 GPa 0

1000

1200 1400 Temperature (°C)

1600

Fig. 4. Peridotite saturated olivine water storage capacity, C ol H2 O , as a function of temperature for (a) this study, with equal bulk water content of 0.47 wt%, at 5 GPa for a narrow temperature range of investigation. (b) determinations of peridotitesaturated storage capacities of olivine (this study, solid circles) at 5, 6, and 8 GPa compared to storage capacity measurements at the same pressure from experiments in which olivine coexisted only with orthoyroxene and fluid or only fluid. Storage capacities diminish with increasing pressure owing to reduced activity of H2O in fluids that dissolve greater proportions of silicate, but are significantly lower for olivine coexisting with the full lherzolite assemblage owing to greater dissolved silicate component and commensurately lower activity of H2O. Hexagons are from Withers and Hirschmann (2008) (excluding experiments buffered at low SiO2 activity), diamonds from Withers and Hirschmann (2007), triangles from Mosenfelder et al. (2006) and squares from Kohlstedt et al. (1996), empty symbols are for 8 GPa, in grey for 6 GPa and black for 5.65 GPa. All previously published data are adjusted to conform with the infrared calibration of Withers et al. (submitted for publication) for OH stretching bands in olivine synthesized at high pressure. Adjustment of the Kohlstedt et al. (1996) data assumed sample thickness of 100 mm and absorption indicatrix axial polarized relative band intensities that are typical of high pressure olivines (Withers et al., submitted for publication). An additional correction has been applied to the SIMS data of Withers and Hirschmann (2007) and Withers and Hirschmann (2008) to compensate for underestimation of C H2 O in the standards used in those studies (Withers et al., 2011).

concentrations, where melt fractions are too small to confidently document saturation experimentally). This assumption does not account for changes in melt composition at small melt fraction, such as enrichment in alkalis, which should stabilize melt at lower H2O activity and therefore support a somewhat lower H2O storage capacity. We note, however, that near solidus alkali enrichments for garnet peridotite are small. For example, an incipient partial melt of fertile lherzolite has just 2.5 wt% Na2O

P. Ardia et al. / Earth and Planetary Science Letters 345-348 (2012) 104–116

400 6 GPa, 1450°C

300

200 5 GPa, 1400°C 100

Measured water in olivine (ppm)

400

water in Ol (ppm)

113

300

200

100

0

data at 1400°C data at 1450°C data at 1500°C

1:1 0.4

0.6 0.8 bulk water content (wt%)

1.0

Fig. 5. Olivine H2O storage capacity at 1400 and 5 GPa and 1450 and 6 GPa for three different starting water concentrations. The data suggest that C ol H2 O of peridotite-saturated olivine in the presence of hydrous melt is not influenced by the bulk water content. We infer that at a given temperature and pressure increases in bulk H2O content generate greater proportions of peridotite partial melt with similar H2O concentrations, leading to similar water activity in both the melt and coexisting olivine.

at 3 GPa (Davis et al., 2011), much less than the 48 wt% expected at 1 GPa (Baker et al., 1995; Hirschmann et al., 1998). Nearsolidus K2O concentrations are yet lower for depleted K2O-poor mantle; for example an incipient (0%) partial melt of depleted mantle peridotite with 60 ppm K2O (Workman and Hart, 2005) should have 2 wt% K2O (Davis et al., 2011). Therefore we argue that any such diminishment of storage capacity in the mantle at depths 4100 km are not significant, though further experiments with alkali-rich bulk compositions are desirable. The measured peridotite-saturated H2O storage capacities of olivine can be compared to those predicted by the model of Hirschmann et al. (2009). This model predicts the concentration of H2O dissolved in silicate melt required to stabilize melt at a given temperature and pressure below the dry peridotite solidus and specifies an olivine/melt partition coefficient, and so predicts the concentration of H2O in olivine under those conditions. Thus, the experimental storage capacity determinations provide an independent check on the accuracy of the model. The H2O concentrations calculated in this way match the experiments with an average disagreement of 34% (Fig. 6). A preponderance of the experimental measurements have H2O concentrations greater than those predicted, which suggests that olivine storage capacities are greater than those predicted by the model. The disagreement may arise largely owing to the relatively small olivine/melt partition coefficient (0.0017) applied in the Hirschmann et al. (2009) model, which derives chiefly from experimental measurements at 0.5–2 GPa. ol=liq Hirschmann et al. (2009) argued that larger values of DH2 O likely apply at high pressure, and the larger concentrations in olivine measured in the present experiments support that contention.

4.1. Olivine storage capacity along the mantle geotherm The change in olivine storage capacity with depth in the mantle convolves the combined effects of changing pressure and temperature along the geotherm. The increase in storage capacity with pressure (Fig. 3a) may be partly offset by the influence of temperature (Fig. 4). Following the sub-ridge geotherm of Stixrude and Lithgow-Bertelloni (2005, 2007), temperature increases from 1410 to 1445 1C over the pressure interval from 5 to 8 GPa, so within experimental and analytical error, determinations at 5–8 GPa and

0 0

100 200 300 Predicted water in olivine (ppm) (Hirschmann et al., 2009)

400

Fig. 6. Observed olivine storage capacities compared to those calculated from the model of Hirschmann et al. (2009). The latter are calculated from parameterization of experiments that constrain the concentration of H2O required to stabilize hydrous partial melt of peridotite at the temperature and pressure of interest (4.5–13 wt%, with the maximum concentrations corresponding to 8 GPa and 1450 1C, the minimum to 5 GPa and 1500 1C) and from the olivine/melt partition px=ol coefficient of H2O, DH2 O (0.0017), derived from low pressure experimental determinations. The good correlation between experimental data and calculated values corroborates the validity of the model. On average, direct experiments suggest larger olivine storage capacities than those from the predictive model, px=ol perhaps because the applicable value of DH2 O is larger than 0.0017 at high pressure.

1400–1450 1C can be assumed to represent storage capacities along the geotherm. The geotherm temperature rises to  1500 1C at the base of the upper mantle. Using olivine storage capacity estimates at 1500 1C from Tenner et al. (2011), the peridotite-saturated olivine storage capacity along the mantle geotherm from 150 to 400 km is approximately linear C olivine H2 O ðppmÞ ¼ 39:1  PðGPaÞ66:0

ð2Þ

The modest H2O storage capacity of peridotite-saturated olivine at 5–8 GPa contrasts with large concentrations of H2O found in olivine in simpler systems at similar pressures but lower temperatures. For example, at 1000–1200 1C, Kohlstedt et al. (1996) found 1000 ppm H2O in olivine coexisting with orthopyroxene at 5 GPa and Kohlstedt et al. (1996), Mosenfelder et al. (2006) and Withers and Hirschmann (2008) found 1000– 2000 ppm at 8 GPa (Fig. 4b) (these concentrations are adjusted to be consistent with the new calibration of Withers et al., submitted for publication; see figure caption), which are much greater than the 100–300 ppm found over this pressure interval in this study. This large difference is attributable partly to the effects of temperature and partly to the influence of additional components in peridotite. Withers and Hirschmann (2008) showed that H2O concentrations in olivine coexisting with pyroxene diminish substantially with increased temperature at 8 GPa (Fig. 4b). This is attributable to reduced H2O activity in high temperature fluids coexisting with olþ opx, which have greater dissolved silicate component than lower temperature fluids. At 1450 1C, the trend documented by Withers and Hirschmann (2008) suggests 800 ppm H2O in olivine (Fig. 4b). Quantitative comparison between earlier results and those presented here is complicated by differences between different FTIR and SIMS calibrations, but the smaller storage capacity documented in this study (100–300 ppm) compared to that suggested by the trend of Withers and Hirschmann (2008) at 1450 1C and 8 GPa is owing to

114

P. Ardia et al. / Earth and Planetary Science Letters 345-348 (2012) 104–116

reduced activity of H2O in hydrous partial melts of peridotite as compared to those coexisting with more simple olivine þ pyroxene assemblages. 4.2. Peridotite storage capacity Our experiments constrain directly the H2O storage capacity of peridotite saturated olivine, but the H2O storage capacity requires also constraints on the H2O contained within coexisting pyroxenes and garnet. If such experimental determinations for each phase, i, C iH2 O were available, then the H2O storage capacity of peridotite, (C per H2 O ), could be calculated given the modal proportions of each phase at (Xi) at a particular P and T (C iH2 O ½P,T ): X i C H2 O X i ð3Þ C per H2 O ¼ P,T

Because peridotite-saturated storage capacities for pyroxene and olivine are not available for conditions along the mantle geotherm, the peridotite storage capacity can be calculated by an alternative method that employs olivine storage capacities and partition coefficients between the other minerals and olivine i=ol (DH2 O ), together with the modes of the phases (Xi), ! X i=ol ol C per ¼ C x þ D X ð4Þ i ol H2 O H2 O H2 O i

This approach has the advantage that mineral/mineral partition coefficients are less subject to strong pressure and temperature variations than measurements of peridotite-saturated storage capacity (Hirschmann et al., 2009; Tenner et al., 2011). px=ol A key influence on values DH2 O needed for Eq. (4) is the Al2O3 content of the pyroxenes present in peridotite along the mantle geotherm, as the stability of OH in pyroxene is strongly related to coupled substitutions with Al3 þ (Aubaud et al., 2004; Grant et al., 2006; Mierdel et al., 2007; O’Leary et al., 2010). The Al2O3 content in pyroxene, of course, varies with depth in the mantle and in particular, diminishes with increasing depth after the onset of garnet stability. In Fig. 2a C px from the cpx and opx in the Al2 O3 present experiments are plotted with those from previously published peridotite-saturated experiments and compared to the empirical relation proposed by Hirschmann et al. (2009). The new experimental pyroxene compositions follow the trend proposed by Hirschmann et al. (2009) within the confidence limits that they outlined, except for two pyroxenes at 6 and 8 GPa, for which small crystal sizes inhibited precise analyses. Experimentally-determined values of pyroxene/olivine partitioning of water increase with Al2O3, although there is considerable scatter deriving chiefly from experiments where olivine determinations are near detection limits, and are consistent with the linear relationship (Fig. 2b): px=ol

DH2 O ¼ 2:63UC px þ 0:78 Al2 O3

ð5Þ

along the mantle Combining this with the dependence of C px Al2 O3 geotherm as a function of pressure (Hirschmann et al., 2009) px=ol yields a predictive relationship for DH2 O as a function of pressure (Fig. 2c). gar=ol The value of DH2 O is not constrained from peridotite-saturated experiments at the pressures and temperatures of interest. The closest constraints come from the experiments of Mookherjee and Karato (2010), in which garnet and olivine coexist with a gar=ol hydrous melt and yield DH2 O of 4.2, 9.0 and 0.9, respectively at 5, 7 and 9 GPa and 1100 1C and 0.5 at 9 GPa and 1200 1C. These values account for a correction to olivine concentrations from those reported by Mookherjee and Karato (2010), who applied the method of Paterson (1982), to concentrations consistent with the Withers et al. (submitted for publication) calibration applied

elsewhere in this work. We evaluate peridotite storage capacity gar=ol applying values of DH2 O of 0.9, 4.2, and 9 and we infer that the lower values likely apply at higher pressure, but clearly the available constraints have considerable uncertainty. Combining Eqs. (2), (4) and (5) allows calculation of peridotite water storage capacities along a convecting mantle geotherm (Fig. 7). gar=ol Three curves derive from the distinct values of DH2 O employed and each curve has an envelope that represents additional possible variation that arises from uncertainties in the Al2O3 content of pyroxenes (Fig. 2). The predicted bulk water storage capacities in gar=ol peridotite at 4 GPa ( 130 km depth) are 339761 ppm (DH2 O ¼0.9), gar=ol gar=ol 381761 ppm (DH2 O ¼4.2), and 443762 (DH2 O ¼9) and at 8 GPa gar=ol gar=ol they rise to 3477151 ppm (DH2 O ¼0.9), 4847151 ppm (DH2 O ¼ gar=ol 4.2), and 6807152 (DH2 O ¼9). The calculated peridotite-saturated storage capacities are consistent with those predicted by the model of Hirschmann et al. (2009) (thick black dashed line in Fig. 7). Although the two approaches share assumptions related to the relative partitioning of H2O between olivine and other mantle minerals, we emphasize that they derive ultimately from entirely different experimental constraints. The model of Hirschmann et al. (2009) is based principally on experimental mineral-melt partition coefficients of H2O (e.g., Aubaud et al., 2004, 2008; Hauri et al., 2006; Tenner et al., 2009), which constrain the concentration of H2O in nearsolidus partial melts, and experimental determinations of the influence of that H2O on the melting temperature of peridotite, which determine the temperature at which such near-solidus partial melts become stable as a function of depth. Consequently, the H2O storage capacity determinations presented here represent independent verification of the accuracy of the Hirschmann et al. (2009) model. 4.3. Application to hydrous melting in the mantle It is well-established that small amounts of H2O in nominally anhydrous peridotite reduce the melting temperature and increase the depth at which melting commences in upwelling mantle (Hirth and Kohlstedt, 1996; Asimow and Langmuir, 2003). However, significant debate remains regarding the magnitude of this effect. Recent contributions argue that normal concentrations of H2O ( 100–200 ppm) are sufficient to induce partial melting in the low velocity zone (or asthenosphere) beneath mature oceanic lithosphere at depths of 70–220 km (Mierdel et al., 2007; Green et al., 2010; Ni et al., 2011), whereas others have considered that dehydration partial melting of nominally anhydrous mantle is limited to regions of upwelling beneath ridges and oceanic islands (Hirth and Kohlstedt, 1996; Asimow and Langmuir, 2003; Aubaud et al., 2004; O’Leary et al., 2010). Our experimental results at 5–8 GPa apply most directly to the deeper portions of the low velocity zone at 150–250 km depth. They demonstrate that the H2O concentration required to induce partial melting in peridotite at this depth interval is between 270 and 855 ppm (Fig. 7). This is significantly greater than the 50–200 ppm H2O present in mantle sampled by MORB (Michael, 1995; Saal et al., 2002; Simons et al., 2002; Workman and Hart, 2005), suggesting to us that pervasive hydrous partial melting does not occur at these depths. We note, however, that 350–900 ppm H2O may be present in mantle beneath oceanic islands (Sobolev and Chaussidon, 1996; Dixon et al., 2002; Simons et al., 2002; Aubaud et al., 2005) or in enriched mantle domains beneath ridges (Dixon et al., 2002), and that local partial melting may occur owing to H2O. Owing to a similarity in experimental approach, our conclusions about the H2O required to induce hydrous partial melting in the asthenosphere invites comparison with the contrasting interpretation of Green et al. (2010). Green et al.’s (2010) conclusion that hydrous

Pressure (GPa)

5

6

115

Mantle water content (average) Hirschmann et al., 2009 Green et al., 2010 Tenner et al., 2011

LOW VELOCITY ZONE

4

009 H2

G 2010

P. Ardia et al. / Earth and Planetary Science Letters 345-348 (2012) 104–116

This study: calc. peridotite water storage capacity with D Gar/Ol peridotite saturated: H2O 9.0 4.2 0.9 calc. water concentration in olivine

7

8

Mantle

9 0

100

Tenner et al., 2011

200 300 400 500 600 700 800 bulk water content in peridotite, 0% incipient melting (ppm)

900

Fig. 7. Calculated peridotite H2O storage capacity as a function of depth in the upper mantle (Eq. (2), adiabat given by Stixrude and Lithgow-Bertelloni (2007)), are plotted px=ol with their uncertainties (see Fig. 3) indicated by the horizontal lines which is defined by variations in alumina content (Eq. (5)) and DH2 O . The grey line was calculated gar=ol assuming DH2 O of 4.2, thin black lines for values of 9.0, and thick black lines for values of 0.9 (Mookherjee and Karato, 2010). The thick dotted line is C ol H2 O calculated for the ridge adiabat (Eq. (2)). The vertical line refers to the estimated C ol H2 O for Green et al. (2010). The thick dashed black line is the estimated H2O concentration in peridotite from the study of Hirschmann et al. (2009) based on the freezing point depression model, where the H2O content for the peridotitic assemblage is calculated for 0% melting. The graded grey zone between 50 and 200 ppm H2O represents the estimated H2O concentration in the upper mantle. At the bottom in black, we show the range of H2O dissolved in peridotite calculated by Tenner et al. (2011) for higher-pressure (410 GPa).

partial melting would occur in an asthenosphere containing 180 ppm H2O is based chiefly on experimental observation of H2O-poor minerals coexisting with material that they interpret as having quenched from a fluid-saturated hydrous partial melt containing  30 wt% H2O. For example, at 2.5 GPa and 1050 1C, Green et al. (2010) infer vapor-saturated partial melt to coexist with olivine with 40 ppm H2O. If their observations are robust, then relatively small amounts of H2O in olivine (and by extension, peridotite) are needed to stabilize a highly H2O-enriched melt in the asthenosphere. Superficially, a storage capacity of 40 ppm H2O at 2.5 GPa seems consistent with the trend of olivines from our higher pressure experiments (Fig. 3b), but the latter were annealed at high temperature (1400– 1500 1C) and therefore coexisted with comparatively low H2O melts (4–13 wt%, see figure caption of Fig. 6) far from vapor saturation. In other words, the vapor-saturated olivines at 2.5 GPa and 1050 1C should have substantially more H2O than the vapor-undersaturated trend in Fig. 3b. Indeed, 40 ppm H2O is a factor of 3–10 times less than that found for vapor-saturated olivine at equivalent conditions by Kohlstedt et al. (1996) and Mosenfelder et al. (2006). Either the quenched material in the experiments of Green et al. (2010) were not actually vapor-saturated melts with  30 wt% H2O or there is a discrepancy between their methods for quantifying H2O in olivine with those applied in our work, Kohlstedt et al. (1996), and Mosenfelder et al. (2006). Green et al. (2010) argued that the small concentration of H2O (50–200 ppm) generally inferred in the source regions of MORB are lower than the concentrations in the convecting mantle below, reflecting instead the H2O remaining in mantle residues following removal of small-degree partial melts at depths beneath those of principal sub-ridge basalt generation. They suggest that the H2O removed at greater depth erupts with offaxis magmas or is vented by ‘‘mantle degassing’’. It is not clear why deep melts formed during upwelling might get diverted away from ridge or how mantle could degas from depths greater than 70 km, given the very high solubility of H2O in basalt. Plank and Langmuir (1992), considered the possibility that partial melt

from mantle upwelling far from ridges, in the so-called ‘‘wings’’ of the partial-melting region, may not be sampled at ridges, but the loss of small-degree melts from the sub-ridge source just prior to generation of MORB should also remove other incompatible elements and fractionate highly incompatible elements from each other and from modestly incompatible elements, thereby leaving an imprint on long-term (Rb–Sr, U–Pb) and short-term (U–Th) chronometers. In fact, comparison of Rb/Sr and U/Pb ratios to 87 Sr/86Sr and 206Pb/204Pb isotopes of MORB suggests that excess of small amounts of deep low degree partial melt are as common as depletions (Galer and O’Nions, 1986; White, 1993). Thus, there is little support for the hypothesis that partial melting in the asthenosphere occurs under conditions more hydrous than that generally supposed for the sources of MORB.

Acknowledgements We appreciate the comments of an anonymous referee and the editor, Tim Elliott, and gratefully acknowledge the support of the National Science Foundation through Grants OCE0623550, EAR0757903, and EAR1161023. References Asimow, P., Langmuir, C., 2003. The importance of water to oceanic mantle melting regimes. Nature 421, 815–820. Aubaud, C., Hauri, E., Hirschmann, M., 2004. Hydrogen partition coefficients between nominally anhydrous minerals and basaltic melts. Geophys. Res. Lett. 31, L26011. Aubaud, C., Pineau, F., He´kinian, R., Javoy, M., 2005. Degassing of CO2 and H2O in submarine lavas from the society hotspot. Earth Planet. Sci. Lett. 235, 511–527. Aubaud, C., Withers, A.C., Hirschmann, M.M., Guan, Y., Leshin, L.A., Mackwell, S.J., Bell, D.R., 2007. Intercalibration of FTIR and SIMS for hydrogen measurements in glasses and nominally anhydrous minerals. Am. Mineral. 92, 811–828. Aubaud, C., Hirschmann, M.M., Withers, A.C., Hervig, R.L., 2008. Hydrogen partitioning between melt, clinopyroxene, and garnet at 3 GPa in a hydrous MORB with 6 wt% H2O. Contrib. Mineral. Petrol. 156, 607–625.

116

P. Ardia et al. / Earth and Planetary Science Letters 345-348 (2012) 104–116

Bai, Q., Kohlstedt, D.L., 1992. Substantial hydrogen solubility in olivine and implications for water storage in the mantle. Nature 357, 672–674. Baker, M., Hirschmann, M., Ghiorso, M., Stolper, E., 1995. Compositions of nearsolidus peridotite melts from experiments and thermodynamic calculations. Nature 375, 308–311. Barr, J.A., Grove, T.L., 2010. AuPdFe ternary solution model and applications to understanding the fO(2) of hydrous, high-pressure experiments. Contrib. Mineral. Petrol. 160, 631–643. Behn, M.D., Hirth, G., Elsenbeck II, J.R., 2009. Implications of grain size evolution on the seismic structure of the oceanic upper mantle. Earth Planet. Sci. Lett. 282, 178–189. Dasgupta, R., Hirschmann, M.M., Withers, A.C., 2004. Deep global cycling of carbon constrained by the solidus of anhydrous, carbonated eclogite under upper mantle conditions. Earth Planet. Sci. Lett. 227, 73–85. Davis, F.A., Tangeman, J.A., Tenner, T.J., Hirschmann, M.M., 2009. The composition of KLB-1 peridotite. Am. Mineral. 94, 176–180. Davis, F.A., Hirschmann, M.M., Humayun, M., 2011. The composition of the incipient partial melt of garnet peridotite at 3 GPa and the origin of OIB. Earth Planet. Sci. Lett. 308, 380–390. Dixon, J., Leist, L., Langmuir, C., Schilling, J., 2002. Recycled dehydrated lithosphere observed in plume-influenced mid-ocean-ridge basalt. Nature 420, 385–389. Faul, U., Jackson, I., 2005. The seismological signature of temperature and grain size variations in the upper mantle. Earth Planet. Sci. Lett. 234, 119–134. Galer, S.J.G., O’Nions, R.K., 1986. Magmagenesis and the mapping of chemical and isotopic variations in the mantle. Chem. Geol. 56, 45–61. Grant, K.J., Kohn, S.C., Brooker, R.A., 2006. Solubility and partitioning of water in synthetic forsterite and enstatite in the system MgO  SiO2  H2O7 Al2O3. Contrib. Mineral. Petrol. 151, 651–664. Green, D.H., Hibberson, W.O., Kova´cs, I., Rosenthal, A., 2010. Water and its influence on the lithosphere-asthenosphere boundary. Nature 467, 448 (U97). Hauri, E., Gaetani, G., Green, T., 2006. Partitioning of water during melting of the Earth’s upper mantle at H2O-undersaturated conditions. Earth Planet. Sci. Lett. 248, 715–734. Hirschmann, M.M., 2010. Partial melt in the oceanic low velocity zone. Phys. Earth Planet. Int. 179, 60–71. Hirschmann, M., Baker, M., Stolper, E., 1998. The effect of alkalis on the silica content of mantle-derived melts. Geochim. Cosmochim. Acta 62, 883–902. Hirschmann, M., Aubaud, C., Withers, A., 2005. Storage capacity of H2O in nominally anhydrous minerals in the upper mantle. Earth Planet. Sci. Lett. 236 (1–2), 167–181. Hirschmann, M.M., Tenner, T., Aubaud, C., Withers, A.C., 2009. Dehydration melting of nominally anhydrous mantle: the primacy of partitioning. Phys. Earth Planet. Int. 176, 54–68. Hirth, G., Kohlstedt, D., 1996. Water in the oceanic upper mantle: implications for rheology, melt extraction and the evolution of the lithosphere. Earth Planet. Sci. Lett. 144, 93–108. Hirth, G., Kohlstedt, D.L., 2003. Rheology of the upper mantle and the mantle wedge: a view from the experimentalists. AGU Memoir 138, 83–105. Humayun, M., Davis, F.A., Hirschmann, M.M., 2010. Major element analysis of natural silicates by laser ablation ICP-MS. J. Anal. At. Spectrom. 25, 998. Jarosewich, E., Nelen, J.A., Norberg, J.A., 1980. Reference samples for electron microprobe analysis. Geostand. Newslett. 4, 43–47. Karato, S., Jung, H., 1998. Water, partial melting and the origin of the seismic low velocity and high attenuation zone in the upper mantle. Earth Planet. Sci. Lett. 157, 193–207. Koga, K., Hauri, E., Hirschmann, M., Bell, D., 2003. Hydrogen concentration analyses using SIMS and FTIR: comparison and calibration for nominally anhydrous minerals. Geochem. Geophys. Geosyst. 4, 1–20. (Paper no.1019). Kohlstedt, D., Keppler, H., Rubie, D., 1996. Solubility of water in the alpha, beta and gamma phases of (Mg,Fe)2SiO4. Contrib. Mineral. Petrol. 123, 345–357. Lemaire, C., Kohn, R., Brooker, R.A., 2004. The effect of silica activity on the incorporation mechanisms of water in synthetic forsterite: a polarised infrared spectroscopic study. Contrib. Mineral. Petrol. 147, 48–57. Litasov, K.D., Shatskiy, A.F., Katsura, T., Ohtani, E., 2009. Water solubility in forsterite at 8–14 GPa. Dokl. Earth Sci. 425, 432–435. Lu, R., Keppler, H., 1997. Water solubility in pyrope to 100 kbar. Contrib. Mineral. Petrol. 129, 35–42. Mei, S., Kohlstedt, D., 2000. Influence of water on plastic deformation of olivine aggregates 1. Diffusion creep regime. J. Geophys. Res.-Sol. Ea. 105, 21457–21469. Michael, P., 1995. Regionally distinctive sources of depleted MORB—evidence from trace-elements and H2O. Earth Planet. Sci. Lett. 131, 301–320. Mierdel, K., Keppler, H., Smyth, J.R., Langenhorst, F., 2007. Water solubility in aluminous orthopyroxene and the origin of Earth’s asthenosphere. Science 315, 364–368.

Mookherjee, M., Karato, S.-I., 2010. Solubility of water in pyrope-rich garnet at high pressures and temperature. Geophys. Res. Lett., 37. Mosenfelder, J.L., Deligne, N.I., Asimow, P.D., Rossman, G., 2006. Hydrogen incorporation in olivine from 2–12 GPa. Am. Mineral. 91, 285–294. Ni, H., Keppler, H., Behrens, H., 2011. Electrical conductivity of hydrous basaltic melts: implications for partial melting in the upper mantle. Contrib. Mineral. Petrol. O’Leary, J.A., Gaetani, G.A., Hauri, E.H., 2010. The effect of tetrahedral Al3 þ on the partitioning of water between clinopyroxene and silicate melt. Earth Planet. Sci. Lett. 297, 111–120. Paterson, M., 1982. The determination of hydroxyl by infrared-absorption in quartz, silicate-glasses and similar materials. Bull. Mineral. 105, 20–29. Pitzer, K., Sterner, S., 1994. Equations of state valid continuously from zero to extreme pressures for H2O and CO2. J. Chem. Phys. 101, 3111–3116. Plank, T., Langmuir, C., 1992. Effects of the melting regime on the composition of the oceanic-crust. J. Geophys. Res., 19749–19770. Poe, B.T., Romano, C., Nestola, F., Smyth, J.R., 2010. Electrical conductivity anisotropy of dry and hydrous olivine at 8 GPa. Phys. Earth Planet. 181, 103–111. Presnall, D., Gudfinnsson, G., Walter, M., 2002. Generation of mid-ocean ridge basalts at pressures from 1 to 7 GPa. Geochim. Cosmochim. Acta 66, 2073–2090. Rauch, M., Keppler, H., 2002. Water solubility in orthopyroxene. Contrib. Mineral. Petrol. 143, 525–536. Saal, A., Hauri, E., Langmuir, C., Perfit, M., 2002. Vapour undersaturation in primitive mid-ocean-ridge basalt and the volatile content of Earth’s upper mantle. Nature 419, 451–455. Smyth, J.R., Frost, D.J., Nestola, F., Holl, C.M., Bromiley, G., 2006. Olivine hydration in the deep upper mantle: Effects of temperature and silica activity. Geophys. Res. Lett. 33, L15301. Simons, K., Dixon, J., Schilling, J.G., Kingsley, R., Poreda, R., 2002. Volatiles in basaltic glasses from the Easter-Salas y Gomez Seamount Chain and Easter Microplate: implications for geochemical cycling of volatile elements. Geochem. Geophys. Geosyst., 3. Sobolev, A., Chaussidon, M., 1996. H2O concentrations in primary melts from supra-subduction zones and mid-ocean ridges: implications for H2O storage and recycling in the mantle. Earth Planet. Sci. Lett. 137, 45–55. Stanley, B., Hirschmann, M.M., Withers, A.C., 2011. CO2 solubility in Martian basalts and Martian atmospheric evolution. Geochim. Coscmochim. Acta 75 (20), 5987–6003. Stixrude, L., Lithgow-Bertelloni, C., 2005. Thermodynamics of mantle minerals—I. Phys. Earth Planet. Int. 162, 610–632. Stixrude, L., Lithgow-Bertelloni, C., 2007. Influence of phase transformations on lateral heterogeneity and dynamics in Earth’s mantle. Earth Planet. Sci. Lett. 263, 45–55. Tenner, T.J., Hirschmann, M.M., Withers, A.C., Hervig, R.L., 2009. Hydrogen partitioning between nominally anhydrous upper mantle minerals and melt between 3 and 5 GPa and applications to hydrous peridotite partial melting. Chem. Geol., 1–15. Tenner, T., Hirschmann, M., Withers, A., Ardia, P., 2011. H2O storage capacity of olivine and low-Ca pyroxene from 10 to 13 GPa: consequences for dehydration melting above the transition zone. Contrib. Mineral. Petrol., 1–24. Till, C.B., Grove, T.L., Withers, A.C., 2012. The beginnings of hydrous mantle wedge melting. Contrib. Mineral. Petrol. 163, 669–688. White, W.M., 1993. 238U/204Pb in MORB and open system evolution of the depleted mantle. Earth Planet. Sci. Lett. 115, 211–226. Withers, A.C., Wood, B.J., Carroll, M.R., 1998. The OH content of pyrope at high pressure. Chem. Geol. 149, 161–171. Withers, A.C., Hirschmann, M.M., 2007. H2O storage capacity of MgSiO3 clinoenstatite at 8–13 GPa, 1100–1400 1C. Contrib. Mineral. Petrol. 154, 663–674. Withers, A.C., Hirschmann, M.M., 2008. Influence of temperature, composition, silica activity and oxygen fugacity on the H2O storage capacity of olivine at 8 GPa. Contrib. Mineral. Petrol. 156, 595–605. Withers, A., Hirschmann, M., Tenner, T., 2011. The effect of Fe on olivine H2O storage capacity: consequences for H2O in the Martian mantle. Am. Mineral. 96, 1039–1053. Withers, A.C., Bureau, H., Raepsaet, C., Hirschmann, M.M.. Calibration of infrared spectroscopy by elastic recoil detection analysis of H in synthetic olivine. Chem. Geol. (submitted for publication). Workman, R.K., Hart, S.R., 2005. Major and trace element composition of the depleted MORB mantle (DMM). Earth Planet. Sci. Lett. 231, 53–72. Zhao, Y.H., Ginsberg, S.B., Kohlstedt, D.L., 2004. Solubility of hydrogen in olivine: dependence on temperature and iron content. Contrib. Mineral. Petrol. 147, 155–161.