Earth and Planetary Science Lettersoriginslab.uchicago.edu/sites/default/files/... · Fe isotopic composition of their mantle source and so record a fractionated, non-chondritic isotopi

$Page 1: Earth and Planetary Science Lettersoriginslab.uchicago.edu/sites/default/files/... · Fe isotopic composition of their mantle source and so record a fractionated, non-chondritic isotopi$
Earth and Planetary Science Letters 365 (2013) 63–76

Contents lists available at SciVerse ScienceDirect

Earth and Planetary Science Letters

0012-82

http://d

n Corr

St, Cam

E-m

prcradd

journal homepage: www.elsevier.com/locate/epsl

Abyssal peridotites reveal the near-chondritic Fe isotopic compositionof the Earth

Paul R. Craddock a,n, Jessica M. Warren b, Nicolas Dauphas a

a Origins Laboratory, Department of the Geophysical Sciences and Enrico Fermi Institute, The University of Chicago, 5734 South Ellis Avenue, Chicago, IL 60637, USAb Geological and Environmental Sciences, Stanford University, 450 Serra Mall, CA 94305, USA

a r t i c l e i n f o

Article history:

Received 31 July 2012

Received in revised form

1 January 2013

Accepted 11 January 2013

Editor: B. Martyin part because the Fe isotopic composition of the mantle source of terrestrial basalts is unknown. Here

Keywords:

iron

isotope

abyssal peridotite

mantle

chondritic

1X/$ - see front matter & 2013 Elsevier B.V.

x.doi.org/10.1016/j.epsl.2013.01.011

esponding author. Present address: Schlumberg

bridge, MA 02139, USA. Tel.: þ1 617 768 2042

ail addresses: [email protected],

[email protected] (P.R. Craddock).

a b s t r a c t

Terrestrial oceanic and continental basalts are enriched by approximately þ0.1% in 56Fe/54Fe ratio

relative to primitive, undifferentiated meteorites (chondrites). The d56Fe values of terrestrial basalts are

also distinct from those of basalts from Mars and asteroid Vesta, which have chondritic Fe isotopic

compositions. The processes responsible for the isotopic enrichment of terrestrial basalts are debated,

we report Fe isotopic measurements of abyssal peridotites, which are the residues of limited melting at

oceanic ridges and are thus the best proxies for the composition of the convective portion of the mantle.

Our data show that abyssal peridotites have a mean d56Fe value of þ0.01070.007% (relative to IRMM-

014), which is indistinguishable from chondrites. After correcting this data for seafloor weathering and

mantle melting, we estimate the average Fe isotopic composition of the terrestrial mantle to be

d56Fe¼þ0.02570.025%, which is also indistinguishable from chondrites, within current analytical

precision. We determine that the maximum shift in d56Fe for peridotite residues during partial mantle

melting is 0.01%. Our results argue against isotopic fractionation during core–mantle differentiation or

iron vaporization during the Moon-forming giant impact, because both processes would yield a bulk

mantle d56Fe value that is non-chondritic. In addition, our results suggest that disproportionation of

mantle Fe2þ–Fe3þ in perovskite and Fe0 metal and segregation of metal to the core could not have been

a driver for Fe isotopic fractionation in the silicate mantle. Instead, the different iron isotopic

compositions of abyssal peridotites and MORBs support mounting evidence for iron isotopic fractiona-

tion of melts but not residues during the formation of oceanic and continental crust.

& 2013 Elsevier B.V. All rights reserved.

1. Introduction

Whether the bulk iron isotopic composition of the Earth ischondritic or non-chondritic is highly contentious, yet isimportant for models of Earth formation and evolution. Terrestrialcontinental and oceanic (MORB) basalts are enriched in d56Fe{d56Fe¼[(56Fe/54Fe)sample/(

56Fe/54Fe)IRMM-014–1]�103} by approxi-mately þ0.1% relative to chondrites and to basalts from otherplanetary bodies, including Mars and Vesta (Craddock and Dauphas,2010; Poitrasson et al., 2004; Schoenberg and von Blanckenburg,2006; Teng et al., 2013; Weyer et al., 2005; Weyer and Ionov, 2007).Several models have been put forward to account for this feature,including some that assume that terrestrial basalts directly inherit theFe isotopic composition of their mantle source and so record afractionated, non-chondritic isotopic composition for the silicate Earth

All rights reserved.

er-Doll Research, 1 Hampshire

.

(d56FeBSE¼þ0.1%). One such interpretation is that kinetic isotopefractionation associated with partial vaporization of Fe during theMoon-forming giant impact enriched the silicate Earth (and Moon) inthe heavy isotopes of Fe (Poitrasson et al., 2004). Experimental studiesindicate that no detectable equilibrium iron isotopic fractionation ispresent between molten metal and silicate up to 7 GPa (Hin et al.,2012; Poitrasson et al., 2009). However, Polyakov (2009) argued thatat higher pressures relevant to core–mantle boundary conditions(4100 GPa), partitioning between ferropericlase, post-perovskite andmetallic iron imposed Fe isotopic fractionation, which enriched thebulk silicate portion of the Earth in the heavy isotopes of Fe. Mostrecently, disproportionation of ferrous Fe2þ at high pressure in thelower mantle to Fe0 metal and Fe3þ in magnesium silicate perovskitehas been put forth as another process able to generate a silicatemantle enriched in heavy Fe isotopes (Williams et al., 2012).

An alternative set of models (Dauphas et al., 2009a; Weyer andIonov, 2007) argues that the Fe isotopic composition of the bulksilicate Earth is chondritic (i.e., d56FeBSE¼0%). In these models,terrestrial basalts do not record the same Fe isotopic compositionas their mantle source, but instead are interpreted to reflect iron

www.elsevier.com/locate/epsl

www.elsevier.com/locate/epsl

http://dx.doi.org/10.1016/j.epsl.2013.01.011



mailto:[email protected]

mailto:[email protected]


P.R. Craddock et al. / Earth and Planetary Science Letters 365 (2013) 63–7664

isotopic fractionation during partial melting of the mantle (Dauphaset al., 2009a; Weyer and Ionov, 2007; Williams et al., 2004, 2005).Constraining the bulk d56Fe value of the terrestrial mantle is essentialnot only for correctly interpreting the range of Fe isotopic composi-tions observed in terrestrial igneous rocks, but also for providing abroader perspective on the origin of Fe isotope distributions in otherplanetary bodies.

Several studies of terrestrial igneous rocks have attempted toresolve the debate surrounding a chondritic versus non-chondritic Fe isotope composition for the bulk silicate Earth(BSE, defined as mantle plus crust). One focus has been Fe isotopemeasurements of orogenic peridotites and peridotite xenoliths,because these materials are the solid residues of melting and aproxy for the upper mantle (Huang et al., 2011; Schoenberg andvon Blanckenburg, 2006; Weyer et al., 2005; Weyer and Ionov,2007; Williams et al., 2005). The general conclusion from thesestudies is that the average Fe isotopic composition of peridotitesis closer to that of chondrites than to MORBs. The most extensivestudy is that of Weyer and Ionov (2007), who estimated that thefertile upper mantle has a d56Fe value of þ0.0270.03%. How-ever, bulk d56Fe among individual peridotite samples in theirstudy ranges from �0.43 to þ0.17%, which is much larger thancan be reasonably explained by simple mantle melting processesalone. The wide range of d56Fe values observed in orogenicperidotites and peridotite xenoliths has been attributed to iso-topic disturbances via late-stage metasomatism by melt or fluid(e.g., Weyer and Ionov, 2007; Zhao et al., 2010). Continentalxenoliths have long residence times in the sub-continental litho-spheric mantle and have commonly experienced metasomatism,as demonstrated by variable incompatible element enrichmentsand the presence of hydrous minerals such as phlogopite and

Fresh

5 cm

1 cm

Fig. 1. Photographs of abyssal peridotites demonstrating the range of alteration and defo

consisting of olivine, orthopyroxene and spinel. The degree of serpentinization is o1 vol% as

moderate alteration (LOI¼9%). Orthopyroxene (brown porphyroclasts), clinopyroxene (gree

unaltered. The gray material is olivine, which is partially replaced by mesh texture serpenti

from the SWIR with a high degree of alteration (LOI¼11%). Relicts of all primary minerals ar

also replaced by mesh texture serpentine and magnetite. Pyroxenes have rims of amphibo

illustrating the very fresh core and the contrasting orange weathered rinds along surfaces. W

(peridotite mylonite, EN26-26-77). (For interpretation of the references to color in this figu

amphibole (e.g., Roden and Murthy, 1985). Similarly, orogenicmassif peridotites have been modified in supra-subduction zoneenvironments and many have been metasomatized and exten-sively refertilized by reactions with melts in the lithosphere(Bodinier and Godard, 2003). Such processes obscure the identi-fication of primary mantle isotopic signatures. Thus, interpreta-tion of the Fe isotopic composition of orogenic peridotite andperidotite xenoliths remains contested (e.g., Beard and Johnson,2007; Poitrasson, 2007; Weyer et al., 2005).

Another method to constrain the Fe isotopic composition ofthe bulk silicate Earth has been the analysis of mafic andultramafic products from high degrees of mantle partial melting,such as komatiites, boninites and island-arc basalts (Dauphaset al., 2009a, 2010; Hibbert et al., 2012). If mantle melting isassociated with chemical and isotopic fractionation, high degreepartial melts are expected to have chemical and isotopic compo-sitions more representative of their mantle source than lowdegree partial melts (i.e., MORBs). Dauphas et al. (2009a) haveshown that Fe isotopic compositions of mafic magmatic rocks canbe correlated with chemical indices of the degree of mantlemelting (e.g., TiO2), consistent with positive Fe isotopic fractiona-tion during mantle melting. The Fe isotopic compositions ofproducts from high degrees of partial melting are on averagelighter than that of MORBs. For example, boninites have anaverage d56Fe of þ0.02870.008% (Dauphas et al., 2009a), whichis difficult to reconcile with a mantle source that is fractionated atþ0.1%. However, interpretation of Fe isotopic signatures inkomatiites, boninites and island-arc basalts is not straightfor-ward. Komatiites are mostly of Archean age and have beensubject to common and variable metamorphism, metasomatismand alteration after emplacement (Beswick, 1983; Lahaye and

mylonite core Weathered rinds

5 cm

5 cm

rmation fabrics among samples in this study. (A) Harzburgite from the Tonga trench

indicated by LOI¼0% (fresh peridotite, 7TOW-57-4). (B) Lherzolite from the SWIR with

n porphyroclasts) and spinel (black, not observable in hand specimen) are all relatively

ne and minor magnetite (moderately altered peridotite, VAN7-85-47). (C) Harzburgite

e present, but the degree of serpentinization is higher. Olivine has turned orange and is

le and/or serpentine (highly altered peridotite, VAN7-85-30). (D) Peridotite mylonite

eathered rinds were physically separated and analyzed for their Fe isotopic composition

re legend, the reader is referred to the web version of this article.)

P.R. Craddock et al. / Earth and Planetary Science Letters 365 (2013) 63–76 65

Arndt, 1996; Nesbitt et al., 1979). The chemical and isotopicoverprint from such processes must be well constrained in orderto identify the underlying, primary magmatic signatures. Boni-nites and island-arc basalts are the products of hydrous meltingabove subducting plates, forming from mantle sources that maythemselves have experienced previous melting and depletion(e.g., Perfit et al., 1980; Taylor et al., 1994; Woodhead et al.,1993). The restricted tectonic setting and specific conditions ofmantle melting for the genesis of these rocks has cast doubt onwhether their inferred near-chondritic mantle Fe isotopic signa-ture can be considered representative of MORB mantle sourcesand of the bulk silicate Earth (Williams et al., 2012).

To better constrain the Fe isotopic composition of the bulksilicate Earth (BSE), we have measured the Fe isotopic composi-tion of a suite of well-characterized abyssal peridotites fromultra-slow oceanic spreading centers (Southwest Indian, Gakkeland Molloy Ridges). Abyssal peridotites are the solid residues oflimited asthenospheric melting at oceanic ridges and are thusthe best proxies to the composition of the globally convectingmantle (e.g., Johnson et al., 1990; Workman and Hart, 2005). Incomparison to previous studies, the peridotite samples examinedhere provide the most direct constraints on the BSE modelcomposition.

Table 1Sample locations.

Cruise name Sample Lithology/type R

Indian Ocean

Vancouver 7 Van7-85-27 Harzburgite S

Vancouver 7 Van7-85-30 Harzburgite S

Vancouver 7 Van7-85-42 Lherzolite S




Vancouver 7 Van7-96-09 Pyroxenite vein S

Vancouver 7 Van7-96-21 Lherzolite w/ pyroxenite vein S

Vancouver 7 Van7-96-21 Pyroxenite vein S





Atlantis II-107-6 AII-107-6-61-83 Peridotite mylonite; interior S

Atlantis II-107-6 AII-107-6-61-83a Peridotite mylonite; exterior S

Protea5 PROT-5-18-02 Peridotite mylonite; interior S

Protea5 PROT-5-18-02a Peridotite mylonite; exterior S

Protea5 PROT-5-18-11 Peridotite mylonite S

Protea5 PROT-5-18-40 Peridotite mylonite S

Arctic Ocean

Healy 01-02 HLY0102-04-43 Lherzolite G


Healy 01-02 HLY0102-40-24 Harzburgite G


Healy 01-02 HLY0102-41-24 – G






Polarstern ARK XVII/2 PS59-201-39 Harzburgite G

Polarstern ARK XVII/2 PS59-235-17 Lherzolite G



Polarstern ARK XVII/2 PS59-317-06 Harzburgite G

Endeavour EN26 EN26-26-71 Dunite mylonite M

Endeavour EN26 EN26-26-77 Harzburgite mylonite; interior M

Endeavour EN26 EN26-26-77a Harzburgite mylonite; exterior M

Pacific Ocean

Seven-Tow 7TOW-57-4 Harzburgite T

SWIR¼Southwest Indian Ridge; SMZ¼Sparsely Magmatic Zone; EVZ¼Eastern Volcani

2. Samples and methods

The samples chosen for this study span the range of abyssalperidotite chemical and radiogenic isotopic compositions, fromtypical depleted peridotites to melt-veined peridotites (Warrenet al., 2009). The samples have a variety of alteration histories,ranging from completely fresh, to serpentinized, to seafloorweathered (Fig. 1). Mylonites—ductily deformed rocks associatedwith faults—are included in the sample suite as the cores of thesesamples are extremely fresh, whereas the rims have undergoneseafloor oxidative weathering. In addition, a single peridotitefrom a subduction zone, collected at the Tonga trench, is includedin this study as the sample is the most chemically depleted and isalso extremely fresh.

2.1. Sample descriptions

2.1.1. Abyssal peridotites

A total of 29 abyssal peridotites from three spreading centerswere analyzed (Table 1). Thirteen samples are from the ultraslow(o20 mm/yr full rate) spreading Oblique Segment, located at91–161E on the Southwest Indian Ridge (SWIR) and dredgedduring the 2003 R/V Melville cruise Vancouver Leg 7 (Dick et al.,

idge, segment Latitude Longitude Depth FSR

(m) (mm/yr)

WIR, Oblique 52.271S 15.231E 4190 12.0

WIR, Oblique 52.271S 15.231E 4190 12.0

WIR, Oblique 52.271S 15.231E 4190 12.0

WIR, Oblique 52.271S 15.231E 4190 12.0

WIR, Oblique 52.271S 15.231E 4190 12.0

WIR, Oblique 52.131S 15.151E 3738 12.0

WIR, Oblique 53.151S 9.981E 3134 12.4

WIR, Oblique 53.151S 9.981E 3134 12.4

WIR, Oblique 53.151S 9.981E 3134 12.4

WIR, Oblique 53.151S 9.981E 3134 12.4

WIR, Oblique 53.151S 9.981E 3134 12.4

WIR, Oblique 53.151S 9.981E 3134 12.4

WIR, Oblique 53.151S 9.981E 3134 12.4

WIR, Shaka F.Z. 53.381S 9.331E 4063 14.1

WIR, Shaka F.Z. 53.381S 9.331E 4063 14.1

WIR, Prince Edward F.Z. 46.541S 33.791E 5100 14.7




akkel Ridge, SMZ 84.831N 4.661E 4294 12.9





akkel Ridge, EVZ 86.751N 64.721E 4328 11.0










olloy Ridge, Spitzbergen F.Z. 79.381N 2.651E 14.5



onga Trench, Subduction 20.251S 173.241W 9270

c Zone.

Table 2Isotopic and chemical compositions of abyssal peridotites.

Sample name d56Fe d57Fe Whole rock composition Spinel

Cr]

Cpx

Ce

(ppm)

Alteration

index

(1–5)

Weathering

index (1–3)

MgO/

SiO2n

Group

(1–3)

SiO2 Al2O3 MgO FeOT CaO Na2O K2O TiO2 MnO P2O5 LOI Total Mg]

(%) (%) (wt%)

Reference materials

BHVO-2 0.09970.027 0.12970.054 49.9 13.5 7.2 11.1 11.4 2.22 0.52 2.73 0.17 0.27 – 99.04

DTS-2b 0.02470.055 0.03570.111 39.4 0.45 49.4 6.99 0.12 0.03 – – 0.11 – – 96.49

PCC-1 0.02070.028 0.04770.049 44.41 0.72 46.2 7.91 0.55 0.03 0.01 0.01 0.13 0.00 – 100.01

UB-N 0.05770.028 0.08270.049 39.43 2.90 35.2 7.56 1.20 0.10 0.02 0.11 0.12 0.04 12.19 98.88

Abyssal peridotites

Southwest Indian Ridge

Van7-85-27 0.02170.039 0.02370.078 40.55 1.87 36.28 7.19 1.03 0.16 0.010 0.035 0.106 0.025 11.51 98.76 90.0 23.4 0.029 4 2 0.080 2

Van7-85-30 0.01770.039 0.03870.078 40.15 2.31 36.62 7.92 0.87 0.12 0.010 0.042 0.101 0.016 11.08 99.24 89.2 15.4 0.005 4 1 0.022 1

Van7-85-42 0.05070.028 0.04870.049 40.26 1.99 34.92 7.60 2.18 0.12 0.007 0.036 0.099 0.009 11.98 99.21 89.1 15.9 0.006 3 1 0.096 2

Van7-85-47 0.00070.028 �0.00770.049 41.32 2.21 36.20 8.45 1.56 0.12 0.011 0.029 0.105 0.015 9.27 99.30 88.4 17.4 0.004 3 1 0.073 2

Van7-85-49 0.00370.039 �0.01270.078 40.48 1.62 36.14 7.86 1.76 0.16 0.000 0.030 0.105 0.015 9.83 97.99 89.1 14.9 0.019 3 1 0.104 3

Van7-86-27 0.10870.039 0.16270.078 43.60 2.40 33.35 8.84 2.51 0.26 0.021 0.045 0.116 0.032 8.86 100.03 87.1 17.9 0.013 4 3 0.178 3

Van7-96-09 �0.01170.030 �0.00670.060 46.04 8.27 25.40 4.74 8.56 0.56 0.036 0.241 0.161 0.011 5.57 99.59 90.5 34.4 1.321 2 1 �0.054 1

Van7-96-21 �0.00370.030 �0.00570.049 41.70 2.15 36.59 8.42 1.30 0.17 0.017 0.085 0.115 0.023 9.12 99.68 88.6 26.5 4.196 3 1 0.078 2

Van7-96-21 �0.00270.039 0.00870.078 43.31 4.50 32.21 7.34 4.54 0.29 0.014 0.180 0.114 0.010 6.88 99.40 88.7 16.6 4.486 3 1 0.024 1

Van7-96-25 �0.05570.030 �0.09370.060 39.02 2.15 34.24 8.49 0.73 0.25 0.036 0.072 0.202 0.029 13.67 98.88 87.8 13.8 1.380 3 2 0.066 2

Van7-96-28 �0.01170.030 �0.02270.049 41.70 2.18 34.65 8.39 2.11 0.27 0.026 0.092 0.254 0.033 9.59 99.29 88.0 32.3 3.292 3 3 0.122 3

Van7-96-35 0.09670.030 0.13270.049 35.75 1.17 30.62 6.52 8.24 0.09 0.007 0.046 0.167 0.028 16.58 99.21 89.3 24.0 2.184 3 3 0.166 3

Van7-96-38 0.02170.030 0.01770.060 42.48 3.77 33.54 8.73 2.56 0.28 0.028 0.143 0.238 0.038 8.15 99.94 87.3 11.5 0.882 3 3 0.033 1

Gakkel ridge, Arctic Ocean

HLY0102-04-43 0.02170.029 0.02670.042 35.76 2.53 30.11 10.28 5.26 0.30 0.013 0.039 0.137 0.102 13.81 98.33 83.9 18.2 0.085 – 3 0.045 1

HLY0102-34-24 �0.04770.029 �0.06970.042 40.83 2.14 34.27 8.33 1.72 0.17 0.009 0.034 0.165 0.060 11.88 99.61 88.0 20.9 0.069 2 2 0.113 3

HLY0102-40-24 0.06070.039 0.08770.078 36.66 0.60 35.61 6.98 6.06 0.12 0.006 0.041 0.107 0.017 12.55 98.74 90.1 57.1 0.034 2 3 0.110 3

HLY0102-40-81 �0.04870.029 �0.05570.042 44.48 0.98 44.43 8.39 0.71 0.07 0.000 0.010 0.130 0.020 1.19 100.41 90.4 – – 1 1 0.062 2

HLY0102-41-24 0.07770.029 0.11270.042 39.89 0.47 38.19 7.50 0.28 0.16 0.009 0.037 0.110 0.033 12.72 99.41 90.1 – – – 3 0.140 3

HLY0102-70-56 0.00570.028 0.01770.049 38.88 1.55 35.82 7.35 1.82 0.17 0.004 0.018 0.099 0.015 13.08 98.82 89.7 20.0 0.855 2 – 0.076 2

HLY0102-70-75 0.00770.028 �0.00470.049 38.46 2.16 35.49 6.85 2.36 0.13 0.009 0.031 0.093 0.016 14.01 99.61 90.2 16.0 0.122 2 2 0.016 1

HLY0102-70-87 �0.01870.028 �0.02170.049 39.79 1.94 35.05 7.84 2.19 0.11 0.002 0.049 0.112 0.002 11.77 98.85 88.9 – – 2 1 0.084 2

HLY0102-70-91 0.05070.039 0.07370.078 40.66 2.62 34.58 7.02 1.86 0.10 0.004 0.045 0.111 0.013 11.93 98.94 89.8 12.0 – 3 1 0.059 2

HLY0102-92-36 0.02270.039 0.03870.078 38.23 1.66 33.77 9.33 1.96 0.29 0.013 0.035 0.102 0.073 14.02 99.48 86.6 – – 4 3 0.101 3

PS59-201-39 �0.05270.028 �0.07970.049 39.85 0.93 36.21 7.27 4.58 0.12 0.006 0.004 0.113 0.027 10.09 99.21 89.9 – 0.878 3 2 0.148 3

PS59-235-17 0.04870.030 0.09670.064 45.47 2.86 39.95 8.19 3.14 0.14 0.000 0.065 0.131 0.007 0.39 100.34 89.7 – 0.241 1 2 0.036 1

PS59-235-18 0.04970.030 0.05870.064 44.46 2.65 41.29 8.47 2.29 0.10 0.000 0.056 0.137 0.010 0.53 99.99 89.7 – – 2 1 –0.002 1

PS59-257-15 �0.01270.030 0.00770.064 40.22 1.52 36.02 7.68 1.03 0.14 0.006 0.035 0.100 0.028 12.85 99.63 89.3 25.6 – 3 1 0.110 3

PS59-317-06 0.02470.030 0.03870.064 38.81 1.60 37.17 7.33 0.26 0.13 0.007 0.021 0.106 0.023 14.37 99.82 90.0 20.7 0.004 4 2 0.035 1

Tonga trench

7TOW-57-4 0.03970.030 0.04870.064 44.18 0.51 45.57 8.18 0.50 0.02 0.002 0.000 0.127 0.000 �0.04 99.05 90.9 – 0.017 1 1 0.067 2

Mylonite peridotites

Southwest Indian Ridge

AII-107-6-61-83 0.02070.030 0.02370.064 43.35 2.07 41.27 8.22 1.90 0.10 0.009 0.028 0.131 0.005 1.96 99.05 89.9 16.6 0.024 1 1 0.017 1

AII-107-6-61-

83a

0.08570.039 0.13270.078 45.01 2.91 36.05 8.13 2.84 0.25 0.087 0.039 0.106 0.000 3.62 99.05 88.8 – – – 3 0.108 3

P.R

.C

rad

do

cket

al.

/E

arth

an

dP

lan

etary

Science

Letters3

65

(20

13

)6

3–

76

66

PR

OT

-5-1

8-0

2�

0.0

277

0.0

30�

0.0

437

0.0

64

45

.41

1.8

84

3.0

68

.25

1.2

20

.07

0.0

00

0.0

45

0.1

38

0.0

07�

0.2

19

9.8

79

0.3

29

.5–

11

0.0

43

1

PR

OT

-5-1

8-0

2a

�0

.02

17

0.0

30�

0.0

347

0.0

64

––

––

––

––

––

––

––

––

3–

–

PR

OT

-5-1

8-1

10

.02

47

0.0

28

0.0

427

0.0

74

44

.94

1.5

24

3.5

28

.63

0.5

10

.04

0.0

00

0.0

20

0.1

51

0.0

07

0.4

69

9.7

89

0.0

––

11

0.0

51

2

PR

OT

-5-1

8-4

0�

0.0

947

0.0

28�

0.1

537

0.0

49

44

.75

2.9

44

0.5

38

.53

2.1

00

.15

0.0

02

0.1

05

0.1

44

0.0

07�

0.1

19

9.1

58

9.4

17

.6–

11

�0

.00

11

Mo

llo

yR

idg

e,A

rcti

cO

cea

n

EN

26

-26

-71

0.0

127

0.0

41

0.0

307

0.0

86

44

.33

1.0

04

0.7

48

.65

1.4

00

.10

0.0

18

0.0

08

0.1

22

0.0

05

1.7

89

8.1

48

9.4

30

.2–

11

0.1

40

3

EN

26

-26

-77

0.0

197

0.0

41

0.0

367

0.0

86

43

.05

1.7

74

2.0

38

.19

1.4

10

.28

0.0

71

0.0

31

0.1

21

0.0

09

1.7

89

8.7

49

0.1

––

11

0.0

17

1

EN

26

-26

-77

a0

.05

27

0.0

30

0.1

237

0.0

64

41

.73

2.6

23

4.6

98

.97

4.0

30

.47

0.0

85

0.0

77

0.1

32

0.0

47

6.5

09

9.3

58

7.3

––

–3

0.0

84

2

Wh

ole

rock

ma

jor

ele

me

nts

of

pe

rid

oti

tes

are

fro

mth

isst

ud

y.N

eg

ati

ve

va

lue

sfo

rLO

Ia

rew

ith

inth

ee

stim

ate

dre

pro

du

cib

ilit

yo

f7

0.5

%.T

he

ind

ice

so

fse

rpe

nti

niz

ati

on

(alt

era

tio

n)

an

dw

ea

the

rin

ga

refr

om

vis

ible

insp

ect

ion

of

sam

ple

sin

ha

nd

spe

cim

en

an

d/o

rp

oli

she

dse

ctio

n.

We

ath

eri

ng

ind

ex

fro

m1¼

fre

shto

3¼

ex

ten

siv

ely

we

ath

ere

d.

Alt

era

tio

nin

de

xfr

om

1¼

fre

shto

5¼

com

ple

tely

alt

ere

d.

Mg

O/S

iO2n

isth

ed

iffe

ren

ceb

etw

ee

nth

em

ea

sure

da

nd

pre

dic

ted

Mg

O/S

iO2

rati

os

ba

sed

on

the

terr

est

ria

lg

eo

che

mic

al

fra

ctio

na

tio

ntr

en

da

sa

nin

dic

ato

ro

fth

ed

eg

ree

of

we

ath

eri

ng

(Gro

up

s1

–3

).S

pin

el

Cr]

an

dcl

ino

py

rox

en

eC

ea

refr

om

pu

bli

she

dst

ud

ies

of

the

sam

esa

mp

les

(re

fere

nce

sli

ste

din

the

tex

t).


2003; Standish et al., 2008). Seafloor peridotite exposures arerelatively common along this segment due to normal faulting thataccommodates a significant proportion of the spreading (Dicket al., 2003). The peridotites used in this study have beencharacterized extensively for their mineral, trace element andradiogenic isotope compositions (Warren et al., 2009; Warren andShirey, 2012). Samples from dredge Van7-85 are typical depletedharzburgites and lherzolites, while those from dredge Van7-96include pyroxenite-veined peridotites. Van7-96 dredge samplesspan the radiogenic isotope field of MORBs and abyssal perido-tites, from depleted to enriched. The pyroxenites have the samerange of trace element and radiogenic isotopic compositions astheir peridotite hosts and are thought to have originated duringthe passage of the nearby Bouvet hotspot (Warren et al., 2009).

Fifteen samples are from the Gakkel ridge in the high ArcticOcean, collected during a coordinated two ice-breaker cruise(USCGC Healy and PFS Polarstern) in Summer 2001 (Michaelet al., 2003; Thiede and Shipboard Scientific Party, 2002). TheGakkel ridge extends east for �1800 km from the Lena trough,north of Greenland, toward the continental margin of the LaptevSea. Full spreading rate decreases west to east from 14 to 7 mm/yr,making the entire ridge ultra-slow spreading. Sampling focused onthe western half of the Gakkel ridge from 51W to 851E, withperidotites recovered primarily between 31 and 291E in the SparselyMagmatic Zone. Seafloor spreading in this section is associated withlittle or no volcanism (Michael et al., 2003). Both harzburgite andlherzolite occur in the western part of the Sparsely Magmatic Zone,while lherzolite is dominant to the east (Michael et al., 2003).

A single peridotite from the Tonga trench at 211S is alsoincluded in this study. This sample (7TOW-57-4) was recoveredat a depth between 9270 and 9130 m by dredge during the 1970cruise Seven-Tow aboard R/V Thomas Washington (Bloomer andFisher, 1987). The sample is associated with the edge of the TongaPlate, which is over-riding the Pacific Plate. Peridotites from thisdredge are predominantly harzburgite with almost no alteration(o1 vol% serpentine). Sample 7TOW-57-4 from the Tonga ridge isthe freshest of the abyssal peridotites in this study (Fig. 1), andone of few abyssal peridotites containing 0–1 vol% serpentine.

2.1.2. Peridotite mylonites

Six abyssal peridotite mylonites were also studied for theiriron isotopic compositions (Table 1). Mylonite is a generic termthat refers to rocks that have undergone intense ductile deforma-tion, resulting in grain size reduction, often by orders of magni-tude (White et al., 1980). In peridotite mylonites, olivine hastypically recrystallized to a grain size of 1–100 mm, compared toan original grain size of 1–10 mm (Jaroslow et al., 1996). Abyssalperidotite mylonites are often very fresh, with serpentinizationlimited to cross-cutting veins. Seafloor oxidative weathering isreadily observed on the mylonite samples as thin layers (o5 mm)stained with Fe-oxyhydroxide (Fig. 1). For this study, weatheredrinds were isolated from the fresh interiors and were analyzedseparately.

Four of the peridotite mylonites in this study are from theSWIR, collected during cruises AII-107-6 (R/V Atlantis II, 1980) andProtea5 (R/V Melville, 1984) from the 91E Shaka and 331–351EPrince Edward fracture zones, respectively. Detailed petrologicdescriptions of these samples are provided by Farmer and Dick(1981), Fisher et al. (1986) and Jaroslow et al. (1996). Mylonitesfrom the Shaka F.Z. are predominantly harzburgite. The fewmylonites described lithologically from the Prince Edward F.Z.are lherzolite. The two other peridotite mylonites are from theMolloy ridge in the Arctic Ocean at 791N, 31E (depth 2000 m)between the Knipovich Ridge and the Lena Trough. These samples

20

40

60

0 2 40

0.2

0.4

0.6

spin

el C

r#

whole rock Al2O3 (wt.%)

who

le ro

ck N

a 2O

(wt.%

)

peridotites

mylonites

Gakkel

SWIRVan7-85SWIRVan7-96

Tonga

SWIRGakkel

5%10%

15%

Fig. 2. Indices of geochemical fractionation: Spinel Cr][¼molar Cr/(CrþAl)] and

whole-rock Na2O versus whole rock Al2O3 on an anhydrous basis. The peridotites

studied here span the range of whole rock compositions observed globally (crosses

in bottom panel; Niu, 2004), which are interpreted as reflecting a range of degrees

of partial melting. The orange trend (top panel) is geochemical fractionation in the

residue modeled for 0–15% partial melting of average depleted mantle (Workman

and Hart, 2005). The modeled compositions overlap those measured in our

peridotite suite. The calculation was carried out using MELTS thermodynamic

software to track major element distributions among melt and mineral phases

(Asimow and Ghiorso, 1998; Ghiorso and Sack, 1995). Melting was fractional and

induced by decompression from 2.5 to 0.2 GPa along an adiabat, starting at a

temperature of 1350 1C (which is above the solidus at 2.5 GPa) and ending at

1257 1C. The oxygen fugacity, fO2, was fixed at QFM to calculate the ferric and

ferrous Fe contents of the solid at starting temperature and pressure.

(For interpretation of the references to color in this figure legend, the reader is

referred to the web version of this article.)


are from dredge EN26-26, recovered during the 1978 cruise EN26aboard R/V Endeavour (Neumann and Schilling, 1984).

Major and trace element compositions of the peridotite sam-ples studied here (Table 2) span the range observed globally, fromrelatively fertile to chemically depleted (Fig. 2). For example,whole rock Al2O3 contents of the peridotites, excluding thepyroxenite veins, overlap the range reported by previous studies(Coogan et al., 2004; Godard et al., 2008; Niu, 2004; Paulick et al.,2006). Spinel Cr]{¼molar Cr/(CrþAl)�100} in our samplesranges from 12 to 57, compared to a global variation from 10 to59 for abyssal peridotites (Dick and Bullen, 1984). The reciprocalcorrelation between spinel Cr] and whole rock Al2O3 reflectsgeochemical fractionation during partial mantle melting, owing tothe more incompatible behavior of Al relative to Cr (Dick andBullen, 1984). Most of the samples have pyroxene trace elementconcentrations typical of abyssal peridotites (Johnson et al.,1990). The exceptions are the abyssal peridotites and pyroxenitesfrom dredge Van7-96 from the Southwest Indian Ridge, which aretrace element enriched (Warren et al., 2009). These samples areinterpreted as reflecting a contribution from recycled componentsin the Bouvet plume, which was co-located under the westernend of the Oblique Segment spreading axis at �15 Ma and iscurrently associated with Bouvet Island, South Atlantic (Hartnady

and Le Roex, 1985). The range of chemical compositions withinwhich our samples fall (e.g., whole rock Al2O3, spinel Cr]) isconsistent with the abyssal peridotites representing the residuesof o15–20% mantle partial melting (Dick and Bullen, 1984).Modeling with MELTS can reproduce the major element varia-bility among our samples by 4–16% melting (Fig. 2a). This is inagreement with recent estimates of mantle melting at mid-oceanridges of between 6% and 10% globally (Salters and Stracke, 2004;Workman and Hart, 2005).

2.2. Analytical methods

From hand specimen, approximately 5 g of interior materialwas obtained by cutting with a wafer blade. Where possible,weathered rims were removed from cut samples by gentlygrinding exterior surfaces, but for some abyssal peridotites inwhich weathering appeared pervasive this removal was incom-plete. For the mylonites, oxidative rinds were easily isolated fromthe fresh interiors retained for isotopic analysis. Subsequentsample preparation and Fe isotope measurements followed theanalytical procedures developed in our laboratory and describedin detail previously (Dauphas et al., 2004, 2009b). Samples werecrushed to powder using hand pestle and mortar. Between 5 and15 mg of powdered sample was digested by sequential acidtreatment and then purified by ion exchange chromatography(Bio-Rad AG1-X8, chloride form, 1 mL volume) in HCl medium.The purified sample solutions were analyzed for their Fe isotopiccomposition using a Thermo Scientific Neptune MC-ICPMS at theUniversity of Chicago. Instrumental mass fractionation was cor-rected using standard–sample bracketing. Iron isotopic composi-tions are reported in permil deviation relative to the referencematerial IRMM-014, or IRMM-524a (metal from which IRMM-014is prepared), which have an identical iron isotopic composition(Craddock and Dauphas, 2010; Taylor et al., 1992). All Fe isotopicvalues are reported at the 95% confidence level. Dauphas et al.(2009b) have shown that iron isotopic measurements are accu-rate at the 70.025% level.

To demonstrate the accuracy and long-term reproducibility ofisotope ratio measurements in our laboratory, we report here theFe isotopic compositions of well-characterized mafic and ultra-mafic reference materials (basalts BHVO-1/BHVO-2, peridotitePCC-1, dunite DTS-2b and serpentine UB-N). In all cases, ourmeasured Fe isotope ratios are identical within analytical uncer-tainties to other laboratory values (Poitrasson et al., 2004;Schoenberg and von Blanckenburg, 2005; Schuessler et al.,2009; Weyer et al., 2005), including those reported recently byMillet et al. (2012) using a Fe double spike.

Major element analyses of whole rock samples (Table 2) werecarried out by the GeoAnalytical Lab at Washington StateUniversity, following procedures detailed by Johnson et al.(1999). Spinel Cr] and clinopyroxene trace element concentra-tions are from previous studies (Jaroslow et al., 1996; Strackeet al., 2011; Warren and Hirth, 2006; Warren et al., 2009).

The alteration present among samples in this study is typicalof abyssal peridotites. In hand specimen, abyssal peridotitesdisplay a range of degrees of alteration (Fig. 1). While abyssalperidotites can appear highly altered, thin section analysis oftenreveals that alteration is not pervasive, with all primary mineralphases still present. Commonly, peridotites have experiencedhydrothermal alteration, which results in partial replacement ofolivine by a mesh network of serpentine (e.g., Bach et al., 2006).This partial serpentinization is often responsible for changes incolor and texture of the sample (compare Fig. 1b and c to a). Wedeveloped an alteration index (Table 2) for abyssal peridotites ona scale of 1 (fresh) to 5 (completely altered). Peridotites have analteration index of 2 (relatively fresh) if they preserve primary

Chondritesn=42

Peridotitesn=37

15

20

-0.2 -0.1 0 0.1 0.2

10

15

20

5


pyroxenes with minimal or no alteration to serpentine. In con-trast, a value of 3 indicates peridotites with pyroxenes beingpartially replaced and rimmed by amphibole or serpentine, andolivine being more pervasively replaced. Samples have an index of4 if one or more mineral phase is completely altered and an indexof 5 if they are completely altered.

Marine weathering of seafloor samples also occurs at very lowtemperature when peridotites are exposed on the seafloor forextended periods of time (e.g., Snow and Dick, 1995). Oxidativemarine weathering is visible as discoloration of the peridotitesfrom the original greenish hue to tan or orange, in particularalong exposed surfaces and cracks (Fig. 1d). In some cases,replacement of olivine by clay minerals occurs. We developed aweathering index on a scale of 1–3 to categorize the degree ofweathering in our samples (Table 2). Peridotites with a weath-ering index of 1 are visibly unweathered with no orange colora-tion; an index of 2 represents samples with minor amounts oforange coloring and clay mineral formation; an index of 3 repre-sents highly weathered samples that are orange in color and haveextensive formation of clay minerals.

5

10

15

20MORBs

n=52

10

-0.2 -0.1 0 0.1 0.2

-0.2 -0.1 0 0.1 0.2

5

δ56Fe (‰)

Fig. 3. Iron isotope frequency distributions of chondrites (Craddock and Dauphas,

2010; Dauphas et al., 2009a), abyssal peridotites and peridotite mylonites (this

study), and mid-ocean ridge basalts (Beard et al., 2003; Schoenberg and von

Blanckenburg, 2006; Teng et al., 2013; Weyer and Ionov, 2007). The Fe isotopic

composition (d56Fe) of the upper mantle inferred from our abyssal peridotite data

is indistinguishable from that of chondrites.

3. Results

The d56Fe values of all whole rock abyssal peridotites andperidotite mylonites fall within a limited range from �0.094 toþ0.108% (Fig. 3; Table 2). This range of d56Fe values measured inabyssal peridotites and peridotite mylonites is significantly smallerthan among orogenic peridotites. Abyssal peridotites with thehighest degree of alteration and weathering tend to have theheaviest Fe isotopic compositions (Table 2), whereas the Fe isotopiccompositions of the other whole rock samples cluster about 0%. Themean of all abyssal peridotite samples (d56Fe¼þ0.01070.007%,n¼37) is indistinguishable from that of carbonaceous, ordinary andenstatite chondrites (d56Fe¼þ0.00570.008%, n¼42), but is sig-nificantly lighter than that of their MORB complements(d56Fe¼þ0.11070.003%, n¼52; Figs. 3 and 4; e.g., Beard et al.,2003; Schoenberg and von Blanckenburg, 2006; Teng et al., 2013;Weyer and Ionov, 2007). This is demonstrated by two-tailed Studentt-tests. There is no statistically significant difference between themean d56Fe compositions of peridotites and chondrites [t(77)¼0.75,critical t (po0.025)¼2.29]. However, the sample populations ofperidotites and MORBs have statistically different mean d56Fecompositions [t(85)¼11.43, critical t (po0.025)¼2.28]. There isno indication of systematic inter-ridge differences in d56Fe valuesamong either peridotite or basalt populations (Fig. 4). The averaged56Fe of peridotite samples from SWIR (þ0.01670.012%, n¼13)and Gakkel (þ0.00970.011%, n¼15) are the same at the currentlevel of analytical precision.

4. Discussion

This study of abyssal peridotites and peridotite mylonitesdemonstrates that oceanic mantle rocks have a narrow distribu-tion of Fe isotopic values clustered about 0%, with a mean d56Fecomposition that is identical to that of chondrites, within analy-tical uncertainty (Fig. 3). Abyssal peridotites are the best samplesto directly investigate elemental and isotopic composition of themantle, as they are the residues of asthenospheric melting withlimited melt removed. Below, we present arguments that demon-strate that the Fe isotopic compositions of abyssal peridotitesduring melting preserve that of their asthenospheric source andare thus representative of the convective portion of the bulksilicate Earth. The abyssal peridotite data provide the mostconvincing evidence to date that the terrestrial mantle has a bulk

Fe isotopic composition nearly identical to that of chondrites.Further, the data demonstrate that MORBs have an Fe isotopiccomposition fractionated by þ0.1% in d56Fe relative to theirmantle source, supporting evidence for Fe isotope fractionationduring terrestrial mantle partial melting.

4.1. Iron isotope redistribution in peridotites owing to alteration

and weathering?

We consider first whether hydrothermal alteration or weath-ering have systematically affected the primary Fe isotopic com-position of these mantle rocks before discussing the broadimplications from our Fe isotopic data. It is commonly observedthat abyssal peridotites are variably affected by post-magmatichydrothermal alteration and marine weathering (Niu, 2004; Snowand Dick, 1995), and the majority of peridotites in this study have

-0.20

-0.10

0.00

0.10

0.20

δ56 F

e (v

s. IR

MM

-014

)

Peridotites

MORBs

Southwest Indian Ridge Gakkel Mid-AtlanticRidge

Peridotites

MORBs

MORBs MORBs±0.025 ‰

Chondrites

Abyssal peridotitesPeridotite mylonitesMid-ocean ridge basalts

Fig. 4. Summary of the stable Fe isotope compositions (d56Fe vs. IRMM-014) of abyssal peridotites and peridotite mylonites (this study) and of mid-ocean ridge

basalts (MORBs) from oceanic spreading centers (Weyer and Ionov, 2007; Teng et al., 2013). The weighted mean d56Fe value and associated 95% confidence envelope are

shown for peridotites and basalts according to ridge location. Our most conservative estimate of analytical uncertainty is 70.025% (shown by the

vertical error bar) is based on the accuracy of Fe isotope measurements determined by Dauphas et al., (2009b). Peridotite and MORB populations have statistically

different d56Fe compositions, with MORBs heavier by þ0.1%. There is no statistical difference in the Fe isotopic composition within peridotite and MORB populations

sampled from different spreading centers. At current limits of analytical precision, the isotopic fractionation between peridotite (mantle residue) and basalts (melt) is the

same along all ridge axes.

weatheringterrestrialarray

20

30

40

50

0.60.81.01.20

0.05

0.10

Al 2

O3/

SiO

2

MgO/SiO2

MgO

(w

t.%)

serpentinization

0 5 10 15

-0.1

0

0.1

0.2

0 0.1 0.2

MgO/SiO2*LOI (wt.%)

δ56 F

e (‰

)

δ56 F

e (‰

)

weathering

-0.1

0

0.1

0.2

0 5 10 15

LOI (wt.%)

peridotitesmylonites

WI =3WI =2WI =1

WI =3WI =1

Fig. 5. Chemical and isotopic indicators of peridotite alteration (serpentinization)

and seafloor weathering. (a) Hydrothermal alteration is driven primarily by

serpentinization at elevated temperatures (4150–400 1C) associated with hydra-

tion of primary olivine and pyroxenes to secondary minerals including serpentine

and brucite (e.g., Bach et al., 2006). Loss on ignition (LOI, wt%) is a proxy for the

degree of hydration (serpentinization). (b) Low temperature oxidative weathering

at the seafloor is characterized by preferential loss of MgO (Snow and Dick, 1995),

as shown by departure from the terrestrial magmatic fractionation array that

defines primary MgO/SiO2 ratios (Jagoutz et al., 1979; Hart and Zindler, 1986).

Our geochemical estimate of weathering, MgO/SiO2n. Dashed lines correspond to

MgO/SiO2n values of 0.1 and 0.2. (c) d56Fe versus LOI. (d) d56Fe versus MgO/SiO2

n

Blue bands in (c) and (d) define the best fit and 95% confidence interval of the

regression through each dataset. Samples are plotted according to their weath-

ering index (W.I.), which is based on visual examination of hand specimens. See

Supplementary material for the same data plotted according to their geographic

location, which shows that no systematic variation occurs as a function of

location. (For interpretation of the references to color in this figure legend, the

reader is referred to the web version of this article.)


undergone some degree of hydrothermal alteration and/or oxida-tive marine weathering. Hydrothermal alteration of peridotite isdriven primarily by serpentinization at elevated temperatures(�150–400 1C) associated with hydration of primary olivine andpyroxenes to secondary minerals including serpentine and brucite(e.g., Bach et al., 2006), whereas marine weathering is distin-guished as a very low temperature, oxidative process operating atthe seafloor (Snow and Dick, 1995).

The extent of serpentinization in the abyssal peridotite samplesranges widely, from minimal (o1 vol% serpentine) to high(420 vol% serpentine; Fig. 1). The degree to which abyssal perido-tites are serpentinized can be assessed from whole rock watercontents, represented by loss on ignition (LOI, wt%). LOI values forsamples in this study range from 0% to 17% (Fig. 5), indicatingvariable degrees of serpentinization among abyssal peridotites, inagreement with petrographic observations. Samples with LOI o2%,and thus minimal serpentinization, include the Tonga sample, threeGakkel samples and the cores of the peridotite mylonites (Table 2).

Serpentinization does not appear to have disturbed—at leastnot in a systematic manner—the primary Fe isotope compositionof the peridotites. The d56Fe values of heavily serpentinizedperidotites are not obviously different from those of un-serpentinized or weakly serpentinized samples (d56Fe vs. LOI,Fig. 5). The best fit through the data indicates that the meand56Fe value of peridotites is invariant with increasing degrees ofserpentinization. The whole rock Fe contents of the abyssalperidotites and peridotite mylonites, excluding the pyroxenite-veined samples, range between 8 and 12 wt% FeOT on an anhy-drous basis. Thus, they do not exhibit the extensive Fe depletionthat would be required to fractionate the isotopes of Fe in thealtered residue. These observations can be explained by the factthat serpentinization is near isochemical except for the addition ofwater (i.e., major element ratios are unchanged), without mobili-zation and loss of Fe at the whole-rock scale (Bach et al., 2006; Niu,2004). The lack of large-scale mobilization of Fe during serpenti-nization (i.e., acting as a closed system process) is expected not toproduce Fe isotopic fractionation. Our interpretation is similar tothat of Rouxel et al. (2003) who found that the calculated averageFe isotopic composition of basalt alteration products was found tobe very similar to that of bulk basaltic crust, despite a wide range

-0.10

-0.10

-0.10 -0.05 0.00 0.05 0.10 0.15

-0.05 0.00 0.05 0.10 0.15

-0.05 0.00 0.05 0.10 0.15

δ56Fe (‰)

2

4

2

4

2

4

Group 1 (n =13)MgO/SiO2

Least weathered

Group 2 (n =12)Moderately weathered

Group 3 (n =12)MgO/SiO2

Most weathered

Fig. 6. Iron isotope frequency distributions in abyssal peridotites and peridotite

mylonites (this study), grouped according to their geochemical index of weath-

ering MgO/SiO2* . Group 1 is the least weathered (MgO/SiO2

nr0.045); group 2 is

moderately weathered; and group 3 is the most geochemically weathered (MgO/

SiO2*Z0.101). Peridotites from Groups 1 and 2 have similar distributions of Fe

isotope values, with an average d56Fe composition of 0% within uncertainty. The

dispersion of these data can largely be explained by analytical uncertainty. Group

3 peridotites have a significantly larger dispersion of d56Fe values, likely reflecting

isotopic fractionation during weathering.


of Fe isotopic composition (42% in d56Fe) among individualaltered samples (Rouxel et al., 2003). The net flux of Fe duringhigh-temperature alteration of basalt or peridotite appears to besmall at the macro (410 cm) scale, with no net change in whole-rock primary d56Fe compositions.

Marine weathering of abyssal peridotites at the seafloor is knownto affect primary chemical compositions (Snow and Dick, 1995), andmay have disturbed Fe isotopic compositions. One proxy for theextent of weathering is MgO/SiO2, because Mg is lost preferentiallyfrom the rock during this process (Snow and Dick, 1995). Seafloorweathering can be identified by a plot of MgO/SiO2 versus Al2O3/SiO2

(Fig. 5) in which weathering is characterized by departure from theterrestrial geochemical fractionation array (Hart and Zindler, 1986;Jagoutz et al., 1979). Our proxy for weathering, MgO/SiO2

n, is aquantitative measure of the departure of the measured peridotiteMgO/SiO2 ratio from the geochemical fractionation trend. MgO/SiO2

n iscalculated as the difference between measured MgO/SiO2 and pri-mary MgO/SiO2 ratios, with the primary ratio of each sampleestimated from the MgO/SiO2 versus Al2O3/SiO2 magmatic fractiona-tion trend, assuming that Al and Si are immobile during marineweathering (Snow and Dick, 1995). As MgO/SiO2

n ratios and theinferred degree of weathering increase, so does the dispersion ofthe Fe isotopic compositions (Fig. 5). This suggests that weatheringhas affected the Fe isotopic compositions of the peridotites. Toevaluate this issue more carefully, we have divided the data intothree geochemical groups of approximately equal size, showingdifferent extents of weathering. Group 1 contains the 13 leastweathered samples (out of 37 samples total) that have the lowestMgO/SiO2

n ratios (o0.045). Group 3 contains the 12 most weatheredsamples with the highest MgO/SiO2

n ratios (40.101). Group 2 containsthe remaining 12 samples that fall between groups 1 and 3. Thehistograms in Fig. 6 show that groups 1 and 2 have similardistributions (MSWD�6), whereas group 3 shows significantly moredispersion (MSWD �13) and extends to more positive d56Fe values.The high MSWD value of group 1 is entirely explained by mylonitePROT-5-18-40, which has a d56Fe value of �0.09470.028%. Exclud-ing this sample, the MSWD of the remaining 12 least weatheredsamples is 2.4.

We use the population of least weathered samples, excludingmylonite PROT5-18-40, in estimating the Fe isotopic compositionof the mantle. The average d56Fe value of this subset isþ0.01570.013%. The accuracy of our measurements has onlybeen tested down to a level of 70.025% (Dauphas et al., 2009b).To account for the possibility of systematic error, we increase theerror bars to that value. The best estimate d56Fe value of theabyssal peridotites prior to weathering is therefore þ0.01570.025%. This is a robust estimate that does not depend heavily onthe cut-off adopted for weathering. For instance, taking the 25least weathered samples (�2/3 of the whole data set) yields anaverage of þ0.00870.015% (70.025% including possible sys-tematic errors).

4.2. Can partial melting at oceanic ridges fractionate Fe

isotopes in peridotites?

Several studies have previously examined whether mantle melt-ing at oceanic ridges can fractionate Fe isotope compositions of thesolid residue relative to the mantle source (e.g., Weyer and Ionov,2007; Williams et al., 2005, 2009). The results of these calculationsdiffer substantially, with some models suggesting that large Feisotopic fractionation can be expressed in the residue. Modelingquantitatively the Fe isotopic evolution of melts and residues duringmelting is not without uncertainty, in particular because equilibriumFe isotope fractionation factors between mantle minerals and melts attemperatures and pressures relevant to oceanic ridge melting are notyet known. These uncertainties can be avoided by focusing on isotope

mass balance, which is not dependent upon the type of meltingmodel used (e.g., batch versus fractional) or the mineral-melt isotopefractionation factors chosen. Isotope mass balance for mantle meltingis governed by the relationship,

d56Femantle source ¼ d56FemeltUf Femeltþd56FeresidueUf Feresidue ð1Þ

where f is the mass fraction of Fe in the melt (basalt) or solid residue(peridotite). This equation can be recast in terms of the mantleresidue:

d56Femantle source ¼ ðD56Femelt�solidþd

56FeresidueÞUð12f FeresidueÞþd56FeresidueUf Feresidue

ð2Þ

0.00 0.05 0.10 0.15Degree of Melting, F

Fe2+res

Feliq

Fe3+

Feliq2+

Feliq3+

0.0

0.2

0.4

0.6

0.8

1.0

f

Fractional Melting

0.00

0.05

0.10

0.15

1:1

fFe l

iq

res

Fig. 7. Geochemical evolution of melt and residue during fractional melting of

average upper oceanic mantle. The parameter f represents the mass fraction of

ferrous and ferric Fe in the melt/liquid (basalt) or solid residue (peridotite).

Geochemical fractionation was modeled using MELTS (Ghiorso and Sack, 1995).

The starting composition is the model composition for depleted MORB mantle

from Workman and Hart (2005). The initial ferric/ferrous ratio of the bulk system

was constrained by the buffer fO2¼QFM at starting temperature (1350 1C) and

pressure (2.5 GPa). Oxygen fugacity was subsequently buffered during melting.

Melting occurred by decompression in 0.1 GPa increments along the adiabat,

resulting in�15% melting at a final pressure of 0.2 GPa and a temperature of

1257 1C. The lower panel shows that total Fe partitions into the liquid as a

function of the degree of melting such that fFeliq�F, with the 1:1 correlation

shown. The geochemical distribution of Fe for batch melting of MORB mantle (not

shown) is identical to fractional melting.

0 2 4

-0.2

-0.1

0

0.1

0.2

0 20 40 60whole rock Al2O3 (wt.%)

δ56 F

e (‰

)

peridotites

mylonitesGakkel

SWIR Van7-85SWIR Van7-96

TongaSWIRGakkel

spinel Cr#

Fig. 8. d56Fe versus indices of ridge mantle melting: whole rock Al2O3 (left) and

Spinel Cr] (right). Samples are plotted according to their geographic location.

Abyssal peridotites and peridotite mylonites show no correlation between d56Fe

and indices of melting, as supported by results of partial mantle melting models.

There is no identifiable difference between the average d56Fe values of peridotites

from different oceanic spreading centers, at current limits of analytical precision.


where D56Femelt-solid is the difference between d56Febasalt andd56Feperidotite (¼þ0.1%). Given that the bulk partition coefficient,Kd, of Fe between mineral and melt is close to one at conditionsrelevant to mantle melting (e.g., Pearce and Parkinson, 1993), themass distribution of Fe (fFemelt) can be approximated by the degree ofmelting (F) and it follows:

d56Feresidue2d56Femantle source �2FUD56Femelt�solid

d56Femelt2d56Femantle source � ð12FÞUD56Femelt�solid ð3Þ

To strengthen this point, we modeled quantitatively the distribu-tion of FeO and Fe2O3 between minerals and melt using MELTSsoftware for thermodynamic modeling of phase equilibria (Asimowand Ghiorso, 1998; Ghiorso and Sack, 1995). Melting was modeledas fractional and induced by decompression from 2.5 to 0.2 GPaalong an adiabat. The starting temperature was 1350 1C at 2.5 GPa(which corresponds to being above the solidus) and the finaltemperature was 1257 1C. The oxygen fugacity was initially fixedat QFM (quartz–fayalite–magnetite), from which the bulk FeO andFe2O3 contents were calculated at the starting temperature andpressure. Oxygen fugacity was subsequently buffered during melt-ing. At each increment of pressure and temperature, MELTS wasused to calculate the degree of melting and the distribution of oxidecomponents between minerals and melt. In our calculations, up to�15% melting was produced. Extraction of Fe as both Fe3þ andFe2þ into the melt scales almost unitarily as a function of melting(i.e., fFeliq�F) such that, at F�0.15, the total fraction of Fe in themelt is �0.12 (Fig. 7).

Our peridotite sample compositions can be reproduced by 4–16%fractional melting (Fig. 2), based on melting models that start withan average mantle composition from Workman and Hart (2005).This melting range agrees with previous estimates for the degree ofpartial melting involved in MORB generation (Johnson et al., 1990;Klein and Langmuir, 1987; Montesi et al., 2011; Salters and Stracke,2004; Workman and Hart, 2005). Assuming F¼1075% andD56Femelt-residue¼þ0.1%, mass balance (Eq. (3)) indicates that theshift in the Fe isotopic composition of the residue should be�0.01070.005%. This is not identifiable at current limits ofanalytical precision. Indeed, abyssal peridotites show no correlationbetween d56Fe and indices of melting (Fig. 8), even in the fewperidotite samples that have the most depleted chemical composi-tions. We have corrected the average value of the least weatheredmantle peridotites (group 1; d56Fe¼þ0.01570.025%) by this shiftto yield the present best estimate of the d56Fe composition of theconvective upper mantle of þ0.02570.025%.

Williams et al. (2005) presented calculations suggesting thatpartial mantle melting could impart large isotopic shifts in the Feisotopic composition of the residues of melting. In Fig. 3 of theirpaper, Williams et al. (2005) report a fractionation in d56Fe of theresidue relative to the starting source of �0.13% at F¼0.02,which is impossible from a mass balance perspective. Fractiona-tion between the residue and source of �0.13% at only 2% ofmelting requires that the complementary melt have an Fe isotopiccomposition fractionated by D56/64Fe¼þ6.5% (Eq. 3). This valueis inconsistent with the D56/64Femelt-source values shown by theirmodel calculations and is clearly inconsistent with the smallrange of Fe isotope compositions measured in MORBs (Figs. 3 and 4).We conclude that the graphical results for the partial meltingcalculations presented by Williams et al. (2005) must be incorrectand that there is no measurable Fe isotopic fractionation in theresidue during mantle melting at oceanic ridges.

4.3. A chondritic Fe isotopic composition for the Earth

Our arguments above support the view that mantle melting atoceanic ridges cannot significantly fractionate the Fe isotopic

composition of the residue of melting. Therefore, abyssal peridotitesmust preserve the original Fe isotopic composition of the mantlesource. The spread in the Fe isotopic values of abyssal peridotitesthat are geochemically the least weathered (MgO/SiO2

no0.045, 35%of samples) can be explained almost entirely by analytical uncer-tainty, with the exception of a single mylonite with very negatived56Fe. Any remaining variation among samples may reflect subtledifferences in the Fe isotopic composition among mantle domainssampled along ridge segments. Overall, our results indicate that theconvective upper mantle has a d56Fe value of þ0.02570.025%.


Thus, our best estimate for the Fe isotopic composition of theglobally convecting portion of the silicate Earth is indistinguishableat current levels of analytical precision from that of chondrites, butdifferent from that of MORBs (Craddock and Dauphas, 2010;Dauphas et al., 2009a; Schoenberg and von Blanckenburg, 2006;Teng et al., 2013; Zhu et al., 2001).

This result has several key implications. First, the Moon-forming impact could not have given rise to extensive vaporiza-tion and loss to space of isotopically light Fe at high temperature,because this process would have produced a mantle compositionwith a heavy, fractionated Fe isotopic composition different fromthat of chondrites (Poitrasson et al., 2004).

Secondly, our results indicate that metal-silicate differentia-tion (core formation) at high pressures on Earth also did notfractionate the isotopes of Fe between the bulk mantle and core,as suggested previously (Polyakov, 2009). The idea of Polyakov(2009) was based on the apparent difference in the reducedpartition function ratios, or b-factors, for (Fe,Mg)SiO3 post-perovskite (pv) and metallic Fe (m) determined from nuclearresonant inelastic X-ray scattering (NRIXS), where equilibriumisotopic fractionation between these phases is given by dpv–dmE1000� (ln bpv� ln bm). The b-factors in that study werederived from a series expansion in moments of the phonondensity of state (PDOS) extracted from NRIXS spectra at highpressures. In a recent study, Dauphas et al. (2012) presented analternative method to determine b-factors directly from momentsof the NRIXS energy spectrum. These authors showed that Feisotopic fractionation at high temperatures relevant to mantleminerals is given by the third moment of the NRIXS, whichcorresponds to the mean force constant of Fe bonds. Dauphaset al. (2012) further showed that estimates of the mean forceconstant are highly dependent upon the high energy tails of theNRIXS spectrum or the PDOS. For high pressure minerals (e.g.,ferropericlase, post-perovskite), the derived constants of Polyakov(2009) are highly uncertain and probably inaccurate owing tohigh noise and truncation of the extended high energy tail duringdata acquisition. The absence of measurable Fe isotopic fractiona-tion during core formation as evidenced by the indistinguishableFe isotope composition of undifferentiated chondrites andthe terrestrial mantle supports the need for a reevaluation ofb-factors in high-pressure minerals.

In addition, our findings indicate that there is no Fe isotopicfractionation expressed in the globally convecting mantle as aresult of FeO disproportionation in the lower mantle to Fe metaland ferric Fe in perovskite. This conclusion contrasts with recentresults from experimental studies of Fe disproportionation byWilliams et al. (2012), who reported a large equilibrium isotopefractionation between coexisting perovskite and Fe metal(�0.45%/amu). Williams et al. (2012) suggested that this fractio-nation would be expressed as a positive bulk Fe isotopic shift forthe silicate mantle of þ0.1% in d56Fe, which is the same asterrestrial basalts. However, the isotope balance of Williams et al.(2012) erroneously assumes that all of the Fe isotopic fractiona-tion between perovskite and metal is expressed only in perovs-kite. In Appendix A, we derive the correct chemical and isotopicmass balance for disproportionation. We demonstrate that 6/7 ofthe Fe isotopic fractionation during disproportionation would beexpressed in the metal phase, whereas only 1/7 would beexpressed between perovskite and starting FeO. The calculationsof Williams et al. (2012) over-predict, by a factor of 7, theexpected Fe isotopic fractionation between perovskite and initialmantle FeO.

Our calculations show that no net Fe isotopic fractionation inthe bulk mantle would be detectable by the disproportionationprocess. The difficulty of producing any Fe isotopic shift in thesilicate Earth by disproportionation stems from the low Fe3þ/SFe

ratio of the fertile upper mantle �0.036 (Canil et al., 1994). Basedon the stoichiometry of FeO disproportionation (3Fe2þ-

2Fe2þþFe0), only 1.5% of total Fe as metal (i.e., equivalent to 3%

of total Fe as Fe3þ) could have been segregated from the mantleto the core, leaving 98.5% of the Fe budget in the bulk mantle. Thefraction of Fe removed is insufficient to leverage an isotopicfractionation on the mantle. Furthermore, our isotopic balancealso demonstrates that continuous FeO disproportionation andincremental Fe metal removal to the core (i.e., Rayleigh distilla-tion) yields the same, limited Fe isotopic fractionation in per-ovskite as would single-stage disproportionation (Appendix A).Hence, iron isotopes cannot be used to discriminate the role of Fedisproportionation in the terrestrial mantle.

Our peridotite Fe isotopic data demonstrate that oceanicbasalts are enriched in heavy Fe isotopes relative to that of theirmantle source. This supports previous ideas for Fe isotopicfractionation between melt and solid during terrestrial mantlemelting (e.g., Dauphas et al., 2009a; Weyer et al., 2005; Williamset al., 2004; 2005). Our results also suggest that future work oniron isotopic fractionation during mantle partial melting shouldfocus on constraining the equilibrium iron isotopic fractionationfactors between melts and minerals as a function of pressure.

5. Conclusions

Abyssal peridotites are the residues of limited mantle meltingbeneath oceanic ridges and are thus the most representative samplesof the convective portion of the mantle. The average d56Fe value ofabyssal peridotites from this study is indistinguishable from that ofchondrites. The Fe isotopic composition of peridotites is not measur-ably fractionated from their mantle source during melting (within70.01%) and so peridotites preserve the original d56Fe compositionof their source. Accounting for possible Fe isotopic fractionationduring marine weathering of peridotites and for limited fractionalmelting, we estimate a d56Fe composition of the convective mantlefrom the least weathered peridotites to be d56Fe¼þ0.02570.025%.Abyssal peridotites provide compelling evidence that the d56Fe valueof the mantle is chondritic, whereas MORBs are fractionated atþ0.1% during partial melting. The different average d56Fe values ofperidotites and basalts support the idea that there is Fe isotopefractionation during partial mantle melting of the terrestrial mantle.

Our data preclude fractionation of Fe isotopes at high pres-sures during terrestrial core formation (metal-silicate differentia-tion) and also rule out partial iron vaporization and loss to spaceof isotopically light Fe during the Moon-forming giant impact.Both of these processes would have yielded a mantle with afractionated, non-chondritic d56Fe composition, which is incon-sistent with abyssal peridotites recording d56Fe values for theconvective mantle identical to that of chondrites, within analy-tical uncertainty.

Iron isotopes have been proposed as a proxy to identifydisproportionation of FeO in mantle to Fe0 in metal and Fe3þ inperovskite, because experimentally determined equilibrium Feisotope fractionation between perovskite and metal is large(Williams et al., 2012). However, our study demonstrates thatthis process could not be a driver for Fe isotopic fractionation inthe mantle and that Fe isotopes cannot be used to discriminateFeO disproportionation as a means to internally oxidize theEarth’s mantle.

Acknowledgments

The authors thank Henry Dick at Woods Hole OceanographicInstitution (WHOI) for access to the peridotite sample archive.


Mark Ghiorso is thanked for helpful discussions regarding geo-chemical modeling of mantle melting using MELTS/pMELTS.Reviews by Frank Poitrasson and an anonymous reviewer onour original submission are greatly appreciated. Editorial assis-tance by Bernard Marty is also appreciated. This study was fundedby the NASA grant NNX09AG59G, the NSF grant EAR-0820807 anda Packard Fellowship to N.D.

Appendix A. Derivation of Fe isotope balance during FeOdisproportionation

The disproportionation of Fe2þ can be represented in simpli-fied form by the reaction,

Fe2þ-2

3Fe3þ

þ1

3Fe0,

For n starting atoms of Fe2þ , of which a fraction x dispropor-tionates, the fractions of Fe2þ , Fe3þ , and Fe0 after disproportiona-tion are,

n Fe2þ� �

¼ 1�xð Þn,

n Fe3þ� �

¼2

3xn,

n Fe0� �

¼1

3xn,

The fraction of Fe3þ in perovskite after disproportionation is�1/3 (Williams et al., 2012), so it follows,

nðFe3þÞ

nðFe2þÞþnðFe3þ

Þ¼

ð2=3Þxn

ð1�xÞnþð2=3Þxn¼

1

3:

From this, we can estimate,

x¼3

7:

This establishes the disproportionation reaction for FeO insilicate mantle as,

7Fe2þOðsilicateÞ

-2½ðFe2þOÞ2ðFe3þO1:5Þ�ðperovskiteÞ

þ Fe0

ðmetalÞ:

Iron isotope balance for single stage (bulk) or incremental(Rayleigh) disproportionation follow accordingly.

1. Bulk equilibration

Bulk equilibration follows the single stage equilibrationbetween the lower and upper mantle modeled by Williamset al. (2012). Using the numerical constraints above, we writethe isotopic mass balance between perovskite and metal as,

1�xð Þnþ2

3xn

� �dpvþ

1

3xndm ¼ nd2þ

0 ,

where subscripts pv and m refer to perovskite and metal,respectively. Taking the starting isotopic composition of FeO inthe mantle to be d2þ

0 ¼ 0 and introducing Dpvm ¼ dpv�dm as the

difference in the isotopic composition of Fe between perovskiteand metal (Williams et al., 2012), we obtain,

1-xð Þþ2

3x

� �dpvþ

1

3x dpv�Dpv

m

� �¼ 0:

We therefore have,

dpv ¼x

3Dpv

m

Introducing the value of x derived above,

dpv ¼1

7Dpv

m � 0:143Dpvm

dm ¼�6

7Dpv

m :

2. Rayleigh distillation

In a Rayleigh model, Fe disproportionation and Fe metalremoval occurs incrementally. Let us write npv and Rpv, thenumber of atoms and Fe isotopic ratio of perovskite, respectively.We also note Rpv-m, the Fe isotopic ratio of the flux of iron inperovskite that is transformed in metal and is removed into thecore. The isotopic mass-balance at each increment of metalremoval is,

dðnpvRpvÞ ¼ Rpv-mdðnpvÞ

Introducing ampv ¼ Rm=Rpv, the equilibrium isotopic fractiona-

tion between metal and perovskite, we have,

npvdRpvþRpvdnpv ¼ ampvRpvdðnpvÞ:

Division by npvRpvand rearrangement yields,

1

RpvdRpv ¼ am

pv�1� � 1

npvdnpv

from which we obtain the Rayleigh distillation equation,

dlnRpv ¼ ðampv�1Þdlnnpv:

Integration of the Rayleigh equation gives,

lnðRpv=Rpv,0Þ ¼ ðampv�1Þlnðnpv=npv,0Þ:

We can introduce the d notation by multiplying by 1000 andnoting that DE1000 � (a�1);

dpv ¼�Dpvm

npv

npv,0

� :

For perovskite with 1/3 Fe3þ , we need to disproportionate 3/7of the initial Fe2þ . For each Fe2þ atom that disproportionates, 1/3Fe0 atom is formed. Therefore, 1/7 of the initial Fe2þ is convertedFe metal and 6/7 is retained in perovskite. We have,

dpv ¼�Dpvm ln

6

7

� � 0:15Dpv

m :

Batch equilibration and Rayleigh distillation produce the sameFe isotopic fractionation in perovskite, i.e., � 1=7Dpv

m .

Appendix A. Supporting information

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.epsl.2013.01.011.

References

Asimow, P., Ghiorso, M.S., 1998. Algorithmic modifications extending MELTS tocalculate subsolidus phase relations. Am. Mineral. 83, 1127–1131.

Bach, W., Paulick, H., Garrido, C.J., Ildefonse, B., Meurer, W.P., Humphris, S.E., 2006.Unraveling the sequence of serpentinization reactions: petrography, mineralchemistry, and petrophysics of serpentinites from MAR 151N (ODP Leg 209,Site 1274). Geophys. Res. Lett. 33, L13306, doi:13310.11029/12006GL025681.

Beard, B.L., Johnson, C.M., 2007. Comment on ‘‘Iron isotope fractionation duringplanetary differentiation’’ by S. Weyer et al., Earth Planet. Sci. Lett. V240, pages251–264. Earth Planet. Sci. Lett., 256(3–4), 633–637.

Beard, B.L., Johnson, C.M., Skulan, J.L., Nealson, K.H., Cox, L., Sun, H., 2003.Application of Fe isotopes to tracing the geochemical and biological cyclingof Fe. Chem. Geol. 195, 87–117.

Beswick, A.E., 1983. Primary fractionation and secondary alteration within anArchean ultramafic lava flow. Contrib. Mineral.Petrol. 82, 221–231.

Bloomer, S.H., Fisher, R.L., 1987. Petrology and geochemistry of igneous rocks fromthe Tonga trench—a non-accreting plate boundary. J. Geol. 95, 469–495.

Bodinier, J.-L., Godard, M., 2003. Orogenic, ophiolitic and abyssal peridotites.Treatise Geochem. 2, 103–170.



Canil, D., O’Neill, H.S.C., Pearson, D.G., Rudnick, R.L., McDonough, W.F., Carswell,D.A., 1994. Ferric iron in peridotites and mantle oxidation states. Earth Planet.Sci. Lett. 123, 205–230.

Coogan, L.A., Thompson, G.M., MacLeod, C.J., Dick, H.J.B., Edwards, S.J., HosfordScheirer, A., Barry, T.L., 2004. A combined basalt and peridotite perspective on14 million years of melt generation at the Atlantis Bank segment of theSouthwest Indian Ridge: evidence for temporal changes in mantle dynamics?Chem. Geol. 207, 13–30.

Craddock, P.R., Dauphas, N., 2010. Iron isotopic compositions of geologicalreference materials and chondrites. Geostand. Geoanal. Res. 35, 101–123.

Dauphas, N., Craddock, P.R., Asimow, P.D., Bennett, V.C., Nutman, A.P., Ohnen-stetter, D., 2009a. Iron isotopes may reveal the redox conditions of mantlemelting from Archean to present. Earth Planet. Sci. Lett. 288, 255–267.

Dauphas, N., Janney, P.E., Mendybaev, R.A., Wadhwa, M., Richter, F.M., Davis, A.M.,van Zuilen, M., Hines, R., Foley, C.N., 2004. Chromatographic separation andmulticollection-ICPMS analysis of iron. Investigating mass-dependent andindependent isotope effects. Anal. Chem. 76, 5855–5863.

Dauphas, N., Pourmand, A., Teng, F.-Z., 2009b. Routine isotopic analysis of iron byHR-MC-ICPMS: how precise and how accurate? Chem. Geol. 267, 175–184.

Dauphas, N., Roskosz, M., Alp, E.E., Golden, D.C., Sio, C.K., Tissot, F.L.H., Hu, M.,Zhao, J., Gao, L., Morris, R.V., 2012. A general moment NRIXS approach to thedetermination of equilibrium Fe isotopic fractionation factors: application togoethite and jarosite. Geochim. Cosmochim.Acta 94, 254–275.

Dauphas, N., Teng, F.-Z., Arndt, N.T., 2010. Magnesium and iron isotopes in 2.7 Ga Alexokomatiites: mantle signatures, no evidence for Soret diffusion, and identification ofdiffusive transport in zoned olivine. Geochim. Cosmochim. Acta 74, 3274–3291.

Dick, H.J.B., Bullen, T., 1984. Chromian spinel as a petrogenetic indicator in abyssaland alpine-type peridotites and spatially associated lavas. Contrib. Mineral.Petrol. 86, 54–76.

Dick, H.J.B., Lin, J., Schouten, H., 2003. An ultrslow-spreading class of ocean ridge.Nature 426, 405–412.

Farmer, H.G., Dick, H.J.B., 1981. Description of WHOI Rock Dredge Samples. WHOITechnical Report 81–48. Woods Hole Oceanographic Institution, Woods Hole,MA, p. 325.

Fisher, R.L., Dick, H.J.B., Natland, J.H., Meyer, P.S., 1986. Mafic/ultramfic suites ofthe slowly spreading southwest Indian ridge: protea exploration of theAntarctic Plate boundary, 241E–471E, 1984. Ofioliti 11, 147–178.

Ghiorso, M.S., Sack, R.O., 1995. Chemical mass transfer in magmatic processes. IV.A revised and internally consistent thermodynamic model for the interpola-tion and extrapolation of liquid–solid equilibria in magmatic systems atelevated temperatures and pressures. Contrib. Mineral. Petrol. 119, 197–212.

Godard, M., Lagabrielle, Y., Alard, O., Harvey, J., 2008. Geochemistry of the highlydepleted peridotites drilled at ODP Sites 1272 and 1274 (fifteen–twentyfracture zone, Mid-Atlantic Ridge): Implications for mantle dynamics beneatha slow spreading ridge. Earth Planet. Sci. Lett. 267, 410–425.

Hart, S.R., Zindler, A., 1986. In search of a bulk-Earth composition. Chem. Geol. 57,247–267.

Hartnady, C.J.H., Le Roex, A.P., 1985. Southern Ocean hotspot tracks and theCenozoic absolute motion of the African, Antarctic, and South American plates.Earth Planet. Sci. Lett. 75, 245–257.

Hibbert, K.E.J., Williams, H.M., Kerr, A.C., Puchtel, I.S., 2012. Iron isotopes in ancientand modern komatiites: evidence in support of an oxidized mantle fromArchean to present. Earth Planet. Sci. Lett. 321–322, 198–207.

Hin, R.C., Schmidt, M.X., Bourdon, B., 2012. Experimental evidence for the absenceof iron isotope fractionation between metal and silicate liquids at 1 GPa and1250–1300 1C and its cosmochemical consequences. Geochim. Cosmochim.Acta 93, 164–181.

Huang, F., Zhang, Z., Lundstrom, C.C., Zhi, X., 2011. Iron and magnesium isotopiccompositions of peridotite xenoliths from Eastern China. Geochim. Cosmo-chim. Acta 75, 3318–3334.

Jagoutz, E., Palme, H., Baddenhausen, H., Blum, K., Cendales, M., Dreibus, G.,Spettel, B., Lorenz, V., Wanke, H., 1979. The abundances of major, minor andtrace elements in the earth’s mantle as derived from primitive ultramaficnodules. In: Proceedings of the 10th Lunar and Planetary Science Conference,pp. 2031–2050.

Jaroslow, G.E., Hirth, G., Dick, H.J.B., 1996. Abyssal peridotite mylonites: implica-tions for grain-size sensitive flow and strain localization in the oceaniclithosphere. Tectonophysics 256, 17–37.

Johnson, D.M., Hooper, P.R., Conrey, R.M., 1999. XRF analysis of rocks and mineralsfor major and trace elements on a single low dilution Li-tetraborate fusedbead. Adv. X-ray Anal. 41, 843–867.

Johnson, K.T.M., Dick, H.J.B., Shimizu, N., 1990. Melting in the oceanic uppermantle: An ion microprobe study of diopsides in abyssal peridotites. J.Geophys. Res. 95, 2661–2678.

Klein, E.M., Langmuir, C.H., 1987. Global correlations of ocean ridge basalt chemistrywith axial depth and crustal thickness. J. Geophys. Res. 92, 8089–8115.

Lahaye, Y., Arndt, N.T., 1996. Alteration of a komatiite flow from Alexo, Ontario,Canada. J. Petrol. 37, 1261–1284.

Michael, P.J., Langmuir, C.H., Dick, H.J.B., Snow, J.E., Goldstein, S.L., Graham, D.W.,Lehnert, K., Kurras, G., Jokat, W., Muhe, R., Edmonds, H.N., 2003. Magmatic andamagmatic seafloor generation at the ultraslow-spreading Gakkel ridge, ArcticOcean. Nature 423, 956–961.

Millet, M.-A., Baker, J.A., Payne, C.E., 2012. Ultra-precise stable Fe isotopemeasurements by high resolution multiple-collector inductively coupledplasma mass spectrometry with a 57Fe–58Fe double spike. Chem. Geol. 304–305, 18–25.

Montesi, L.G.J., Behn, M.D., Hebert, L.B., Lin, J., Barry, J.L., 2011. Controls on meltmigration and extraction at the ultraslow Southwest Indian Ridge 101–161E. J.Geophys. Res. 116, B10102, doi:10110.11029/12011JB008259.

Nesbitt, R.W., Sun, S.-S., Purvis, A.C., 1979. Komatiites: geochemistry and genesis.Can. Mineral. 17, 165–186.

Neumann, E.-R., Schilling, J.-G., 1984. Petrology of basalts from the Mohns-Knipovich Ridge; the Norwegian-Greenland Sea. Contrib. Mineral. Petrol. 85,209–223.

Niu, Y., 2004. Bulk-rock major and trace element compositions of abyssalperidotites: Implications for mantle melting, melt extraction and post-melting processes beneath mid-ocean ridges. J. Petrol. 45, 2423–2458.

Paulick, H., Bach, W., Godard, M., de Hoog, J.C.M., Suhr, G., Harvey, J., 2006.Geochemistry of abyssal peridotites (Mid-Atlantic Ridge, 151200N, ODP Leg209): implications for fluid/rock interaction in slow spreading environments.Chem. Geol. 234, 179–210.

Pearce, J.A., Parkinson, I.J., 1993. Trace element models for mantle melting:application to volcanc arc petrogenesis. In: Prichard, H.M., Alabaster, T., Harris,N.B.W., Neary, C.R. (Eds.), Magmatic Process and Plate Tectonics. GeologicalSociety, London, pp. 373–403, Special Publications.

Perfit, M.R., Gust, D.A., Bence, A.E., Arculus, R.J., Taylor, S.R., 1980. Chemicalcharacteristics of island-arc basalts: implications for mantle sources. Chem.Geol. 30, 227–256.

Poitrasson, F., 2007. Does planetary differentiation really fractionate iron isotopes?Earth Planet. Sci. Lett. 256, 484–492.

Poitrasson, F., Halliday, A.N., Lee, D.-C., Levasseur, S., Teutsch, N., 2004. Iron isotopedifferences between Earth, Moon, Mars and Vesta as possible records ofcontrasted accretion mechanisms. Earth Planet. Sci. Lett. 223, 253–266.

Poitrasson, F., Roskosz, M., Corgne, A., 2009. No iron isotope fractionation betweenmolten alloys and silicate melt to 2000 1C and 7.7 GPa: experimental evidenceand implications for planetary differentiation and accretion. Earth Planet. Sci.Lett. 278, 376–385.

Polyakov, V.B., 2009. Equilibrium iron isotope fractionation at core–mantleboundary conditions. Science 323, 912–914.

Roden, M.F., Murthy, V.R., 1985. Mantle metasomatism. Annu. Rev. Earth Planet.Sci. 13, 269–296.

Rouxel, O.J., Dobbek, N., Ludden, J., Fouquet, Y., 2003. Iron isotope fractionationduring oceanic crust alteration. Chem. Geol. 202, 155–182.

Salters, V.J.M., Stracke, A., 2004. Composition of the depleted mantle. Geochem.Geophys. Geosyst. 5, Q05B07, http://dx.doi.org/10.1029/2003GC000597.

Schoenberg, R., von Blanckenburg, F., 2005. An assessment of the accuracy of stable Feisotope ratio measurements on samples with organic and inorganic matrices byhigh-resolution multicollector ICP-MS. Int. J. Mass Spectrom. 242, 257–272.

Schoenberg, R., von Blanckenburg, F., 2006. Modes of planetary-scale Fe isotopefractionation. Earth Planet. Sci. Lett. 252, 342–359.

Schuessler, J.A., Schoenberg, R., Sigmarsson, O., 2009. Iron and lithium isotopesystematics of the Hekla volcano, Iceland—evidence for Fe isotope fractiona-tion during magma differentiation. Chem. Geol. 258, 78–91.

Snow, J.E., Dick, H.J.B., 1995. Pervasive magnesium loss by marine weathering ofperidotite. Geochim. Cosmochim.Acta 59, 4219–4235.

Standish, J.J., Dick, H.J.B., Michael, P.J., Melson, W.G., O’Hearn, T., 2008. MORBgeneration beneath the ultraslow spreading Southwest Indian Ridge (9–251E):major element chemistry and the importance of process versus source.Geochem. Geophys. Geosyst. 9, Q05004, doi:05010.01029/02008GC001959.

Stracke, A., Snow, J.E., Hellebrand, E., von der Handt, A., Bourdon, B., Birbaum, K.,Gunther, D., 2011. Abyssal peridotite Hf isotopes identify extreme mantledepletion. Earth Planet. Sci. Lett. 308, 359–368.

Taylor, P.D.P., Maeck, R., De Bievre, P., 1992. Determination of the absolute isotopiccomposition and atomic weight of a reference sample of natural iron. Int. J.Mass Spectrom. Ion Processes 121, 111–125.

Taylor, R.N., Nesbitt, R.W., Vidal, P., Harmon, R.S., Auvray, B., Croudace, I.W., 1994.Mineralogy, chemistry and genesis of boninite series volcanics, Chichijima,Bonin Islands, Japan. J. Petrol. 35, 577–617.

Teng, F-Z., Dauphas, N., Huang, S., Marty, B., 2013, Iron isotopic systematics ofoceanic basalts, Geochim. Cosmochim. Acta, http://dx.doi.org/10.1016/j.gca.2012.12.027.

Thiede, J., Shipboard Scientific Party, 2002. POLARSTERN ARKTIS XVII/2 CruiseReport: AMORE 2001 (Arctic Mid-Ocean Ridge Expedition). Reports on Polarand Marine Research 421, 390 pp.

Warren, J.M., Hirth, G., 2006. Grain size sensitive deformation mechanisms innaturally deformed peridotites. Earth Planet. Sci. Lett. 248, 438–450.

Warren, J.M., Shimizu, N., Sakaguchi, C., Dick, H.J.B., Nakamura, E., 2009. Anassessment of upper mantle heterogeneity based on abyssal peridotite isotopiccompositions. J. Geophys. Res. 114, B12203, doi:12210.11029/12008JB006186.

Warren, J.M., Shirey, S.B., 2012. Lead and osmium isotopic constraints on theoceanic mantle from single abyssal peridotite sulfides. Earth Planet. Sci. Lett.359–360, 279–293.

Weyer, S., Anbar, A.D., Brey, G.P., Munker, C., Mezger, K., Woodland, A.B., 2005. Ironisotope fractionation during planetary differentiation. Earth Planet. Sci. Lett. 240,251–264.

Weyer, S., Ionov, D.A., 2007. Partial melting and melt percolation in the mantle:the message from Fe isotopes. Earth Planet. Sci. Lett. 259, 119–133.

White, S.H., Burrows, S.E., Carreras, J., Shaw, N.D., Humphreys, F.J., 1980. Onmylonites in ductile shear zones. J. Struct. Geol. 2, 175–187.

Williams, H.M., McCammon, C.A., Peslier, A.H., Halliday, A.N., Teutsch, N., Levas-seur, S., Burg, J.P., 2004. Iron isotope fractionation and the oxygen fugacity ofthe mantle. Science. 304, 1656–1659.

http://dx.doi.org/10.1029/2003GC000597

http://dx.doi.org/10.1029/2003GC000597

http://dx.doi.org/10.1029/2003GC000597

dx.doi.org/10.1016/j.gca.2012.12.027

dx.doi.org/10.1016/j.gca.2012.12.027


Williams, H.M., Nielsen, S.G., Renac, C., Griffin, W.L., O’Reilly, S.Y., McCammon, C.A.,Pearson, N., Viljoen, F., Alt, J.C., Halliday, A.N., 2009. Fractionation of oxygen andiron isotopes by partial melting processes: implications for the interpretation ofstable isotope signatures in mafic rocks. Earth Planet. Sci. Lett. 283, 156–166.

Williams, H.M., Peslier, A.H., McCammon, C., Halliday, A.N., Levasseur, S., Teutsch, N.,Burg, J.P., 2005. Systematic iron isotope variations in mantle rocks and minerals:the effects of partial melting and oxygen fugacity. Earth Planet. Sci. Lett. 235,435–452.

Williams, H.M., Wood, B.J., Wade, J., Frost, D.J., Tuff, J., 2012. Isotopic evidence forinternal oxidation of the Earth’s mantle during accretion. Earth Planet. Sci.Lett. 321–322, 54–63.

Woodhead, J.D., Eggins, S.M., Gamble, J., 1993. High field strength and transitionelement systematics in island arc and back-arc basin basalts: evidence formulti-phase melt extraction and a depleted mantle wedge. Earth Planet. Sci.Lett. 114, 491–504.

Workman, R.K., Hart, S.R., 2005. Major and trace element composition of thedepleted MORB mantle (DMM). Earth Planet. Sci. Lett. 231, 53–72.

Zhao, X., Zhang, H., Zhu, X., Tang, S., Tang, Y., 2010. Iron isotope variations in spinelperidotite xenoliths from North China Craton: implications for mantle meta-somatism. Contrib.Mineral. Petrol. 160, 1–14.

Zhu, X.K., Guo, Y., O’Nions, R.K., Young, E.D., Ash, R.D., 2001. Isotopic homogeneityof iron in the early solar nebula. Nature 412, 311–313.

Supplementary Online Material

Abyssal Peridotites reveal the Near‐Chondritic Fe Isotopic Composition of the Earth

Paul R. Craddock, Jessica M. Warren and Nicolas Dauphas

SOM Fig. 1

SOM Fig. 2

SOM Fig. 1. Chemical indicators of peridotite alteration and weathering (see also Fig. 5). Hydro-thermal alteration is driven by serpentinization and the addition of water, for which loss on ignition (LOI, wt. %) is a proxy. Low temperature oxidative weathering at the sea�oor is characterized by preferential loss of Mg (Snow and Dick, 1995) as shown by departure from the terrestrial magmatic fractionation array that de�nes primary MgO/SiO2 ratios (Jagoutz et al., 1979; Hart and Zindler, 1986). Samples are distinguished according to their geographic location.

weathering

terrestrial array

Al2O3/SiO2

MgO

/SiO

2

20 30 40

0

5

10

15

0 0.05 0.1

0.6

0.8

1

1.2

MgO (wt.%)

LOI (

wt.%

)

peridotites

mylonites

Gakkel

SWIR Van7-85SWIR Van7-96

Tonga

SWIRGakkel

SOM Fig. 2. Impact on δ56Fe values of abyssal peridotites and peridotite mylonites from hydro-thermal alteration (serpentinization) and sea�oor weathering (see also Fig. 6). (a) δ56Fe versus LOI, (b) δ56Fe versus MgO/SiO2*. Hydrothermal alteration is driven primarily by serpentinization at elevated temperatures (> 150 – 400 °C) associated with hydration of primary olivine and pyrox-enes to secondary minerals including serpentine and brucite (Bach et al., 2006), whereas marine weathering is distinguished as a low temperature, oxidative process operating at the sea�oor (Snow and Dick, 1995). Samples are plotted according to their geographic location, as in SOM Fig. 1. Blue bands in each plot de�ne the best-�t and 95 % con�dence interval of the regression through each dataset.

weatheringserpentinization

0 5 10 15-0.1

0

0.1

0.2

0 0.1 0.2

MgO/SiO2 *LOI (wt.%)

δ56 Fe

(‰)

a b

Documents

Earth and Planetary Science Lettersoriginslab.uchicago.edu/sites/default/files/... · Fe isotopic composition of their mantle source and so record a fractionated, non-chondritic isotopi