You are on page 1of 14

Earth and Planetary Science Letters 177 (2000) 287^300

www.elsevier.com/locate/epsl

Re^Os fractionation in eclogites and blueschists and the


implications for recycling of oceanic crust into the mantle
Harry Becker *
Isotope Geochemistry Laboratory, Department of Geology, University of Maryland, College Park, MD 20742, USA

Received 28 June 1999; received in revised form 31 January 2000; accepted 12 February 2000

Abstract

Metabasalts (eclogites, blueschists and mafic granulites) metamorphosed in paleosubduction zones show a range in
rhenium (Re) abundances between 3 and 1689 parts per trillion (ppt), with a median value of 331 ppt Re. The median
Re abundance of the metabasites corresponds to only V40% of the Re abundances expected for likely mafic protoliths.
Osmium (Os) abundances in the metabasites (2^42 ppt, with one sample at 909 ppt) are comparable to abundances in
mid-ocean ridge basalt (MORB). Re was lost from the protoliths either during dehydration, the most likely explanation,
or during alteration of the protoliths near the ocean ridges. Radiogenic 187 Os/188 Os and 206 Pb/204 Pb of HIMU ocean
island basalts (OIB) that are believed to contain a component of recycled altered MORB would require excessive
amounts ( s 80^90%) of V0.5^1 Gyr old recycled altered MORB in the mantle source, if constraints on Re/Os and
U/Pb from the metabasites are applied. The necessary amount of recycled component is only marginally lower (70%) if
equal amounts of altered and unaltered 2 Gyr old MORB are present in the mantle source. Considering the major
element abundances in alkaline OIB, these estimates appear unreasonably high. One possibility is that the simple bulk
mixing models commonly employed are not applicable. Rapid melting during the formation of alkaline basalts and
sluggish kinetics may prevent complete equilibration with Os-rich phases such as sulfides and alloys. Other possible
explanations include the shielding of Os-rich phases in peridotite by silicates from reaction with partial melts of eclogite
and the isolation of melt from peridotite by means of pyroxene-rich, Os-poor reaction zones along conduit walls. In the
case of disequilibrium, the actual fraction of recycled component in OIB sources could be much lower than in
equilibrium mixing models. ß 2000 Elsevier Science B.V. All rights reserved.

Keywords: rhenium; osmium; system; eclogite; blueschist; recycling; oceanic crust; ocean-island basalts

1. Introduction crust forms an important component in the source


of mantle plume-derived melts [1^8]. 187 Os/188 Os
Rhenium^osmium (Re^Os) isotopic systematics in most OIB and continental £ood basalts are
of ocean island basalts (OIB) provide key argu- higher than those of abyssal peridotites, peridotite
ments for the hypothesis that recycled oceanic massifs and xenoliths that are assumed to repre-
sent the slightly subchondritic to chondritic upper
mantle (187 Os/188 Os = 0.12^0.13, [9,10]). High val-
ues for 187 Os/188 Os in most OIB with low Os
* Tel.: +1-301-405-0084; Fax: +1-301-314-9661; abundances ( 6 30^50 parts per trillion, ppt) likely
E-mail: hbecker@geol.umd.edu result from high-level crustal contamination

0012-821X / 00 / $ ^ see front matter ß 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 0 1 2 - 8 2 1 X ( 0 0 ) 0 0 0 5 2 - 2

EPSL 5412 4-4-00


288 H. Becker / Earth and Planetary Science Letters 177 (2000) 287^300

[3,4,11]). Nonetheless, some OIB with high Os [16] may indicate that Re in subducted basalts
abundances (several hundred ppt) also have high and sediments may also be mobilized, however,
187
Os/188 Os. The 187 Os/188 Os (0.13^0.15) of these the evidence for this is insu¤cient. Substantial
lavas are di¤cult to explain by crustal contami- loss of Re from subducted oceanic crust would
nation, since contaminants such as seawater and signi¢cantly alter the e¡ect of addition of recycled
oceanic crust have very low Os abundances. For oceanic crust to plume sources and consequently
this reason, and supported by complementary Pb, increase the amounts of recycled crust or time
Sr and Nd isotopic data, Os isotopic compositions necessary to generate high 187 Os/188 Os in OIBs.
of OIBs with high Os abundances (e.g. Hawaii, In the present study, Re and Os abundances of
Austral Islands, Canary Islands) have been in- eclogites, blueschists and ma¢c high-pressure
ferred to be derived from plume sources that con- granulites from high-pressure metamorphic belts
tain up to 30^40% of recycled, altered mid-ocean (paleosubduction zones) were determined to as-
ridge basalt (MORB) or recycled MORB plus sess possible losses of these elements relative to
sediment [1^8]. Os is strongly retained in the res- likely protolith compositions. The types of rocks
idue during mantle melting while Re is moder- analyzed represent the closest proxies available to
ately enriched in melts (e.g. [12]), resulting in study element fractionation in subducted oceanic
high Re/Os in most crustal materials. Because crust, and yield an integrated record of sub-sea-
MORB and sul¢de-rich metaliferous sediments £oor alteration and metamorphic dehydration in
have relatively high Re and low Os abundances subduction zones.
[13,14], these rocks acquire very high 187 Os/188 Os
in several hundreds Myr. Lithophile radiogenic
isotope tracers in OIB, such as Sr, Nd and Pb 2. Samples and analytical techniques
isotopic compositions, are predominantly derived
from recycled components in the mantle, and Most of the eclogites and blueschists analyzed
hence, are not very sensitive indicators of the in this study represent metamorphosed basalts
mass fraction of these components in the mantle. that are remnants of subducted slabs and are
The compatible nature of Os makes 187 Os/188 Os a now exposed, along with metasediments, in
more sensitive indicator for the proportion of re- high-pressure/low-temperature metamorphic belts
cycled crust [1,2]. Because Os contents in MORB [17]. One sample (A43) probably represents a for-
are very low [12,13,15], the 187 Os/188 Os of hybrid mer gabbro, as is indicated by its relatively Ca^
mixtures of recycled crust and mantle depends Al-rich composition and the trace element compo-
essentially on the Re content of recycled crust sitions [17]. The samples (Table 1) are from the
and the age of the recycled component [1]. In Zermatt^Saas ophiolite zone in the western Alps
addition, the very low Os abundances in (LC4A, LC4B, LC5, CEV11B, TAA6A, age of
MORB, yet relatively high Os abundances of metamorphism ca. 35^40 Ma, [18]), Adula nappe
some radiogenic alkaline OIB, require speci¢c in the central Alps (AA15, age of metamorphism
proportions of recycled crust in the mantle sour- 35^40 Ma, [19]), Mu«nchberg gneiss massif, Ger-
ces if independent age constraints are available many (W1, AGR13, age of metamorphism 380^
(e.g. from Pb isotopic data, [2]). 400 Ma, [20]), Syros, Greece (BSY88, BSY87, age
It is commonly assumed that typical Re abun- of metamorphism V70 Ma, M. Broecker, person-
dances in MORB and pelagic sediments can be al communication, 1999), northern Bohemian
used to constrain the evolution of 187 Os/188 Os in massif (DIET1, DIET3, age of metamorphism
subducted oceanic crust, with age estimates for ca. 340 Ma, [21]), Corsica (SAC10, age of meta-
the latter provided by 206 Pb^207 Pb model ages morphism likely Alpine, 40^120 Ma?) and Nor-
[2]. However, evidence for suprachondritic 187 Os/ way (A43, age of metamorphism presumably
188
Os in arc peridotites indicates the mobilization 500 Ma, [22]). More details can be found in
of Os during subduction of oceanic crust [16]. [17]. The ma¢c samples show compositional and
Variable Re enrichment in some arc harzburgites textural evidence for multistage histories including

EPSL 5412 4-4-00


H. Becker / Earth and Planetary Science Letters 177 (2000) 287^300 289

Table 1
Re^Os data for eclogites, blueschists and ma¢c granulites
187
Sample Re Re Re blank Os Os Os blank Re/188 Osc 187
Os/188 Osc
(ppt)a (pg)b (%) (ppt)a (pg)b (%)
ORT4 Eclogite 665 1995 0.8 5.42 16 5.8 1805 þ 103 15.9 þ 0.9
W1 Eclogite 141 436 3.4 2.12 6.4 13.6 517 þ 42 4.91 þ 0.38
Dupl. 23 86 17.5 0.96 2.9 34.2 350 þ 189 15.2 þ 7.8
Dupl.d 38 107 11.2 1.32 3.3 31.3 293 þ 94 8.62 þ 2.7
AGR13 Eclogite 3.0 24 62.3 1.92 5.7 20.7 10.1 þ 13.5 2.67 þ 0.31
AA15 Eclogite 48 156 9.6 5.56 17 15.4 46.8 þ 5.1 1.16 þ 0.08
Dupl.d 49 140 8.6 3.99 10 22.4 76.3 þ 9.8 2.30 þ 0.28
CEV11B Eclogite 1108 3510 0.4 29.4 93 1.6 218 þ 2 1.66 þ 0.01
LC4A Eclogite 230 713 2.1 10.2 31 4.6 121 þ 3 0.964 þ 0.018
LC4B Eclogite 331 1068 1.4 11.4 36 4.0 154 þ 3 0.887 þ 0.014
Dupl. 282 885 1.7 11.7 36 4.0 129 þ 3 0.945 þ 0.015
Dupl.d 458 1157 1.0
LC5 Eclogite 1689 4985 0.3 28.9 85 1.7 304 þ 2 0.747 þ 0.005
A43 Eclogite 44 154 9.7 5.58 18 7.9 99 þ 11 12.3 þ 0.9
DIET1 Maf. Gran. 866 2603 0.6 4.56 14 9.9 2661 þ 282 14.8 þ 1.6
DIET3 Maf. Gran. 795 2567 0.6 5.41 17 7.9 1527 þ 95 8.99 þ 0.56
SAC10 Blueschist 757 2409 0.6 6.51 21 6.8 736 þ 24 2.53 þ 0.08
BSY88 Blueschist 331 1031 1.5 41.7 128 1.2 38.5 þ 0.5 0.1720 þ 0.001
BSY87 Blueschist 125 388 3.9 909 2717 0.06 0.661 þ 0.021 0.1279 þ 0.0005
TAA6A Blueschist 2.5 23 65.3 2.48 7.8 16.0 4.98 þ 0.32 0.277 þ 0.02
Dupl. 2.7 23 64.9 2.03 6.1 19.8 6.65 þ 0.56 0.397 þ 0.04
Dupl., duplicate; Maf., ma¢c; Fels., felsic; Gran., granulite.
a
Blank-corrected.
b
Including blank.
c 187
Errors were propagated from the uncertainties on the blank correction. In a few cases, errors on Os/188 Os are determined by
external mass spectrometric precisions.
d
Predigested in HF^HCl^ethanol.

variable ocean £oor alteration, dehydration, but nol at 100³C. After slow evaporation, the residues
very limited retrograde metamorphic overprint are transferred into Carius tubes and treated like
(e.g. formation of low-pressure amphiboles, [17]). the rest of the samples. Os was loaded on Pure-
Finely-ground powders of samples (2^3 g) were tech1 or ESPI1 Pt ¢laments and covered with a
digested and equilibrated with dilute 185 Re and mixture of Ba(OH)2 and NaOH. Re was loaded
190
Os spikes in reverse aqua regia (5 ml 14 M on fused Pt ¢laments with Ba(NO3 )2 . Os and Re
HNO3 +2.5 ml 11 M HCl) for 2 days at 240³C were run as OsO3 3
3 and ReO4 on the 12-inch, 68³
in sealed Carius tubes. Os was puri¢ed by solvent sector thermal ionization mass spectrometer at
extraction into CCl4 and back-extraction into 9 M the Isotope Geochemistry Laboratory (University
HBr, with a subsequent clean-up by micro distil- of Maryland) in pulse counting mode on the
lation. The Re fraction was separated in a two- electron multiplier. Some of the Re fractions
step column chemistry using 100^200 mesh AG1- were run using a Faraday cup detector. Multiple
X8 anion exchange resin and HCl^HNO3 . Three runs of 0.5^1 ng Os in the Johnson^Mattey
duplicates (see Table 1) were predigested together Os standard solution used at IGL yielded
187
with spike solutions in a mixture of 24 M HF, 11 Os/188 Os = 0.1140 þ 8 (2c, RSD = 0.7%, n = 19)
M HCl and ethanol in te£on bombs at 90³C for and 187 Os/188 Os = 0.1144 þ 6 (2c, RSD = 0.5%,
12 h to rule out analytical problems, such as the n = 11) for two di¡erent run periods. The external
incomplete decomposition of silicates. After evap- precision for the standard re£ects the external
oration, the procedure was repeated in HCl^etha- precision of 187 Os/188 Os for the samples only in

EPSL 5412 4-4-00


290 H. Becker / Earth and Planetary Science Letters 177 (2000) 287^300

the few cases where the blank contribution on


187
Os/188 Os is insigni¢cant (Table 1). Total analyt-
ical blanks were 15 þ 12 pg Re (12 þ 5 pg for sam-
ples with HF predigestion, processed in fall 1998),
and 1.5 þ 0.5 pg for Os (1.0 þ 0.3 pg for ORT4
and one of the W1 samples processed in fall
1997, and 3 þ 1 pg for samples with HF prediges-
tion processed in fall 1998). 187 Os/188 Os of the
blank was 0.175 þ 0.005. Abundances and isotopic
ratios in Table 1 were corrected for blank contri-
bution.
Fig. 2. Y^Re variation diagram. Symbols as in Fig. 1. Note
the di¡erent scales for Y and Re axes. Both elements are
3. Results mildly incompatible and correlate well in MORB (e.g.
[12,23]). Most of the Re data in this diagram are from [13].
Values for Y were calculated from the Tb data in [13], using
Os behaves as a compatible element during Tb/Y = 0.025 for average N-MORB [53]. Yttrium data for
melting and fractionation in the mantle (e.g. metabasites from [17]. Many high-P rocks have Y abundan-
[12]) and consequently, it is advantageous to com- ces comparable to MORB, but signi¢cantly lower Re, indi-
pare it with other compatible elements, such as cating substantial loss of the latter.
Ni, that show predominantly lithophile character-
istics in the crust and mantle (Fig. 1). Os abun- the data lie within the MORB ¢eld in the Ni^Os
dances in the metabasites range between 1 and diagram (Fig. 1). Re is mildly incompatible during
42 ppt (except for sample BSY87, see below). mantle melting [12,23] and shows good correla-
The median Os abundance in the metabasites is tions with other mildly incompatible elements,
6 ppt (n = 15) or 4.9 ppt for `common' Os (the such as Y (Fig. 2). Most eclogites and blueschists
non-radiogenic Os component), little di¡erent (3^1689 ppt) have much lower Re abundances
from the median of 4.2 ppt reported for modern than MORBs, but comparable Y abundances (Y
MORB (n = 40, [12,13,15]). Consequently, most of data from [17]). The metabasalts yield a median
Re of 331 ppt, 62% lower than the median of
MORB (860 ppt, n = 37, [12,13]). Only two ma¢c
granulites and two eclogites overlap with the
MORB ¢eld (Fig. 1b). Two very Na-rich and
Ca-depleted blueschists (TAA6A and BSY87)
that presumably underwent intense high-temper-
ature hydrothermal sub-sea£oor alteration [17]
show variable Re depletion. In sample TAA6A,
Re and Y may be depleted as a result of this
process. Part of the Ca (presumably also Y) lost
from this sample was deposited nearby, and is
preserved as epidote^clinozoisite-rich pockets.
Fig. 1. Ni^Os variation diagram. Shown are abundances in
MORB (+, [12,13]) and OIB (X, [2,23,11,34]). To facilitate
These pockets also contain sul¢des and it is pos-
comparisons, the Os abundance data for metabasites were re- sible that the sul¢des contain some of the missing
calculated using a chondritic 187 Os/188 Os, because most of Re. The other blueschist sample (BSY87) has high
the samples contain a large fraction of radiogenic Os. Nickel Os, Cr and Ni abundances, and a mantle-like
data for metabasites from [17]. Symbols: F, metabasaltic 187
Os/188 Os, suggesting that the low Re abun-
eclogites and blueschists; E, metagabbroic eclogites; O, low-
temperature blueschist (not Ca-depleted); R, Na-rich, Ca-de-
dance in this sample re£ects the depleted nature
pleted blueschists; 8, ma¢c high-pressure granulites. For de- of the protolith (either a cumulate or a MgO-rich
tails on samples, see text and [17]. melt).

EPSL 5412 4-4-00


H. Becker / Earth and Planetary Science Letters 177 (2000) 287^300 291

The results for Re and Os abundances on du-


plicates processed including predigestion in HF^
HCl^ethanol are comparable to results obtained
without predigestion, indicating either that Re in
the silicates is released during both digestion pro-
cedures, or that the silicates contain minor
amounts of Re only. The agreement between du-
plicates of the same sample varies between poor
(samples W1, LC4B) and good (AA15, TAA6A),
most likely re£ecting the heterogeneous distribu-
tion of sul¢des in these medium-grained rocks.
Rutile, a common trace mineral in eclogites, is
not or only partially dissolved by these digestion
procedures. Rutile may accommodate some Re in
its structure if Re has the proper valence (Re4‡ ).
However, it is unlikely that Re in rutile can bal-
ance the low Re abundances in samples AGR13,
AA15 or TAA6A (Table 1), because this would
require 3^6 orders of magnitude higher Re con-
tents in rutile, compared with the amounts of Re
analyzed.
Because of the high Re/Os of most of the meta-
basites relative to Re/Os of the mantle, the meta-
basites have developed high to very high 187 Os/
188
Os (up to 16, Table 1, Fig. 3). Only one sample
has high Os abundances (909 ppt, blueschist
BSY87) and an `upper mantle'-like present-day Fig. 3. (a) Re^Os isochron diagram. Symbols as in Fig. 1.
187 Error bars are shown or are smaller than symbol size. Errors
Os/188 Os (0.1279). Some metabasites, for which
include propagated uncertainties from blank corrections. Iso-
the blank correction on Os and Re was very large, chrons depict the evolution of 40, 200, 500 and 1000 Myr
have 187 Re/188 Os that are too low for their 187 Os/ old rocks evolving from upper mantle-like 187 Os/188 Os (0.13).
188
Os compared to other metabasalts (Fig. 3). (b) Expanded isochron diagram showing samples at low
187
This is particularly the case for two duplicates Re/188 Os.
of sample W1 that contain very little Os and re-
quired a large blank correction (Table 1). The roughly consistent with the ages of metamorphism
aliquot of this sample that yielded the highest and the assumption that igneous formation did
Os content lies close to the 500 Myr isochron, not precede metamorphism by more than 50^
suggesting that the two duplicates that plot to 100 Myr (Fig. 3). In Fig. 4, a model for the Os
the left of the isochron were overcorrected for isotopic evolution of a metabasite is shown with a
their Os or Re blank. Two other samples (AA15 protolith age of 480 Myr and metamorphism and
and A43) show a similar behavior, but possibly concurrent decrease in Re/Os at 400 Myr. Several
for di¡erent reasons as will be discussed further aspects of Re^Os behavior in altered basalts
below. should be noted. The decrease in Re/Os will re-
tard the increase of 187 Os/188 Os compared to the
unmetamorphosed protolith. Thus, if the proto-
4. Impact of alteration on Re^Os model ages and lith age is known, calculated initial 187 Os/188 Os
calculated initial 187 Os/188 Os will be higher (point B in Fig. 4) than the actual
initial 187 Os/188 Os (point A in Fig. 4). Conversely,
187
Most blank-corrected Os/188 Os values are if the protolith age is unknown, a Re^Os model

EPSL 5412 4-4-00


292 H. Becker / Earth and Planetary Science Letters 177 (2000) 287^300

grew before the metamorphism. The evolution


of 187 Os/188 Os in this sample can be modeled by
rapid increase in a 1200 Ma old mantle-derived
ma¢c protolith with 187 Re/188 Os of 1000 and
metamorphism, and Re loss at 500 Ma, with sub-
sequently retarded increase in 187 Os/188 Os. This
interpretation is consistent with the Svecokareli-
an^Grenville age of other eclogite protoliths in
Scandinavia [24] and suggests that the protolith
of A43 may have been a gabbro that intruded
into continental crust or accreted Proterozoic oce-
anic crust.

Fig. 4. Os isotope evolution diagram showing the evolution 5. Re and Os loss in high-pressure metabasites
of 187 Os/188 Os (bold lines) in a hypothetical example of a
Phanerozoic basalt crystallized at 480 Ma and metamor- It is di¤cult to evaluate whether Os was lost or
phosed at 400 Ma with concurrent decrease of Re/Os. The gained in the metabasites because there is only a
Re^Os model age of this sample would be too old. If the
age of the protolith is known from independent constraints,
weak correlation between the compatible elements
the calculated initial 187 Os/188 Os at the time of protolith for- Os and Ni in MORB. In contrast, the low Re
mation would be too high. With increasing time, intervals abundances in most of the eclogites and blue-
between protolith formation and metamorphism di¡erences schists relative to MORB suggest that substantial
between real and calculated protolith ages and initial ratios amounts of Re were lost either during ocean £oor
become much more pronounced, particularly if metamor-
phism occurred recently.
alteration or as a result of dehydration. The e¡ect
of low- and high-temperature alteration at ocean
ridges on Re and Os abundances in MORB is
age at the intersection with the evolution of aver- poorly constrained. Data on pyrite-rich hydro-
age chondrites will yield an age that is older than thermal deposits from the TAG hydrothermal
the true age of the protolith (Fig. 4). These e¡ects ¢eld suggest that near the surface, Re may be
are increasingly ampli¢ed as the metamorphism mobile under oxidizing conditions, but generally
approaches the present. Large time di¡erences be- immobile under reducing conditions deeper in the
tween protolith formation and metamorphism are mineralized zones [14]. It is also unknown
the likely explanation for some samples showing whether low-temperature weathering of MORB
high 187 Os/188 Os and low Re/Os (Fig. 3). For ex- results in the mobilization of Re. It is conceivable
ample, two separate analyses of eclogite AA15 that Re loss occurs during dehydration of altered
show reproducibly low Re/Os that correlate with metabasalts in subduction zones because low-
relatively high 187 Os/188 Os (Table 1). Given the pressure experiments at 500³C show a strong in-
young age of metamorphism for this sample crease in the solubility of ReO2 in hydrothermal
(35^40 Myr), the high 187 Os/188 Os was likely ac- solutions with increasing chloride concentration
quired before metamorphism and loss of Re. The of the £uid [25]. Speciation and solubilities of
relatively high 187 Os/188 Os would require a longer the PGE and Re in complex supercritical £uids
time span between formation of the igneous pro- at high pressures ( s 1^2 GPa) are virtually un-
tolith (Variscan?) and metamorphism. The eclo- known. However, some observations provide cir-
gite sample from Norway (A43) underwent eclo- cumstantial evidence that at least some removal
gite facies metamorphism about 400^500 Myr ago of Re and Os from altered oceanic metabasalt
([22] and references therein). This sample also has occurs during dehydration in subduction zones.
a very high 187 Os/188 Os (12.3, Table 1) for its Re/ First, variable Re enrichment in some depleted
Os, indicating that most of the radiogenic Os arc harzburgites that also show suprachondritic

EPSL 5412 4-4-00


H. Becker / Earth and Planetary Science Letters 177 (2000) 287^300 293

187
Os/188 Os [16] suggests that Re may be trans-
ported from the subducted slab into the subarc
mantle. Second, blueschists and eclogites often
contain sul¢des as a trace phase. These sul¢des
are likely inherited from the igneous protolith or
from previous alteration at low temperatures. The
high sul¢de^silicate partition coe¤cient for Re
(values between 43 and 1200 have been reported
[26,27]) indicates that as long as sul¢des are stable
in eclogites and blueschists, the Re budget of the
bulk rocks will be dominated by sul¢des. Sul¢de-
bearing silicate veins that formed during dehydra- Fig. 5. 187 Os/188 Os^206 Pb/204 Pb diagram showing OIB data
tion at high pressures were reported from eclo- ¢ltered for Os abundances v50 ppt. Ma, Mangaia; Tu, Tu-
gites [28] and are evidence that sul¢des can be buai; Ru, Rurutu; Ta, Tahaii; Sav, Savaii; Ra, Raratonga
mobilized in hydrous subduction zone £uids. [2]; Can, Canary islands [4,8]; Az, Azores [11]; Ha, various
Hawaiian volcanoes [6,7,33,34]. OIB endmember composi-
The heterogeneous distribution of Re and Os in tions from [2,6,54]. UM, upper mantle (e.g. [9,10]); EPM, en-
subducted altered metabasalts (Table 1) makes it riched plume mantle [11,31]. OIBs with high 187 Os/188 Os and
di¤cult to constrain the depletion of these ele- 206
Pb/204 Pb are believed to be derived from a source contain-
ments precisely for the purpose of a general mod- ing recycled MORB (e.g. [2]). The low 206 Pb/204 Pb in some
el. Some eclogites and the blueschists still contain Hawaiian lavas may re£ect old pelagic sediment in a recycled
component [6,7].
hydrous minerals, hence, the possibility exists that
more Re and Os could be lost upon further dehy-
dration. Because of the large variability of the OIB sources, previous isotopic evolution models
abundance data, the median abundances of the used age constraints derived from Pb isotopic re-
metabasites (see Section 3) will be used to charac- lations and assumptions on the likely Re and Os
terize the subducted altered MORB reservoir. The contents of recycled altered MORB [1^8]. From
median values for Re (331 ppt) and Os (6 ppt) these models, it was concluded that the mantle
yield a 187 Re/188 Os of 325 for subducted altered sources of OIB that show the most extreme
metabasalt, signi¢cantly lower than the value for HIMU signature (e.g. Mangaia, Tubuai, St. Hel-
MORB (V1000). ena) may contain between 20 and 30% of recycled
MORB, with a model-dependent age of ca. 2 Gyr
[2]. The 2 Gyr age constraint is based on the as-
6. Os^Pb isotopic constraints on recycled altered sumption that the 206 Pb^207 Pb isotopic correla-
MORB in HIMU-type OIBs tions de¢ned by OIB have age signi¢cance (e.g.
[29]). Some OIBs that show less radiogenic Pb
The new data presented here warrant a re-eval- isotopic compositions (e.g. Canary islands), but
187
uation of Os isotopic constraints on the presence Os/188 Os that are as high as those of extreme
of recycled MORB in OIBs. On the basis of trace HIMU lavas, have been interpreted to contain
element and Pb isotopic systematics, HIMU-type either old recycled MORB with lower 238 U/204 Pb
OIBs (e.g. Tubuai, Mangaia, St. Helena, Canary (W) or younger recycled MORB in their sources
islands, Cape Verde islands) have been interpreted [4,8]. Distributions of Os and Pb isotopic data for
to contain a distinctive signal of recycled altered ocean islands that show a dominant HIMU sig-
MORB (e.g. [29,30]). The discussion here will fo- nature require at least one or two additional
cus on HIMU OIB, because other OIBs contain components to explain their compositions
additional recycled sediment components [6,7] (Fig. 5). Besides the HIMU component, a compo-
that may complicate a quantitative assessment nent that may be either upper mantle or `enriched
of recycling models. In order to obtain estimates plume mantle' (EPM, putative lower mantle with
for the proportion of recycled altered MORB in somewhat higher 206 Pb/204 Pb and 187 Os/188 Os than

EPSL 5412 4-4-00


294 H. Becker / Earth and Planetary Science Letters 177 (2000) 287^300

the upper mantle, e.g. [11,31,32]) is present in the


HIMU OIB source (Fig. 5).
It has been noted previously (e.g. [6,33]) that
certain OIB suites tend to de¢ne straight correla-
tions between 187 Os/188 Os and other radiogenic
tracers (Pb, He, Sr, Nd). Because Os behaves as
a compatible element in the mantle, the linear
correlations with incompatible isotopic tracers im-
ply that equilibrium interactions between Os-de-
pleted partial melts and high-Os mantle peridotite
must be severely restricted, otherwise a decou-
pling of these tracers should be observed
[2,6,34]. Consequently, it has been proposed that
correlations of 187 Os/188 Os with other radiogenic
tracers predominantly re£ect the mixing of mag-
mas derived from isotopically distinct mantle
[2,6,34]. Most data for uncontaminated HIMU
OIB from individual islands or groups of islands Fig. 6. Evolution of 300, 700, 1200 and 2000 Myr old sub-
ducted MORB in a 187 Os/188 Os^206 Pb/204 Pb diagram. Note
seem to form clusters in this diagram (Fig. 5).
logarithmic 187 Os/188 Os axis. (a) Evolution model for altered
MORB (indicated by long arrow) calculated assuming chon-
6.1. Standard models with new data dritic 187 Os/188 Osi , radiogenic growth for 50 Ma and 187 Re/
188
Os = 1000, and retarded growth and 187 Re/188 Os = 325 for
The 187 Os/188 Os^206 Pb/204 Pb diagram in Fig. 6 the rest of the time (300, 650, 1150 and 1950 Myr). 206 Pb/
204
Pbi calculated assuming a single-stage evolution model
shows the evolution of recycled altered MORB
and 238 U/204 Pb = 11. Pb isotopic evolution after metamor-
and unaltered MORB for variable ages. Present- phism calculated using 238 U/204 Pb = 37, derived from eclogites
day compositions for 350, 700, 1200 and 2000 [17]. Strongly curved thin lines are mixing lines between
Myr old recycled altered MORB were obtained modern upper mantle peridotite (3300 ppt Os, 187 Os/
188
from evolution models using the median Re and Os = 0.130, 20 ppb Pb, 206 Pb/204 Pb = 18.3) and subducted
altered MORB (6 ppt Os, 196 ppt Pb, [17]). The ¢eld for
Os abundances from the present study and a
HIMU OIB is shown as a bold bar. (b) Model for 350^2000
model W value of 37, derived from the high-pres- Myr old unaltered MORB (187 Re/188 Os = 1000, 238 U/204 Pb
sure rocks ([17], for details see Fig. 6). Initial iso- = 11 [35,53], 187 Os/188 Osi and 206 Pb/204 Pbi ) calculated the
topic compositions were based on simple evolu- same way as in (a). Also shown is a mixing line between
tion models for the upper mantle (see caption of 2000 Myr old unaltered and altered MORB (from (a)). Thick
marks represent 20% increments. A mixture of 60% altered
Fig. 6) and the evolution for the ¢rst 50 Myr was
and 40% unaltered 2 Gyr old MORB would develop 206 Pb/
calculated using 187 Re/188 Os (1000) assumed to be 204
Pb comparable to the most extreme HIMU OIB (Fig. 5).
typical for altered MORB (the e¡ect of a some-
what higher W is negligible). Evolution during the
remaining time was calculated using 187 Re/ because of the very di¡erent Os/Pb in these end-
188
Os = 325 and W = 37 for subducted altered members (Figs. 6, 7). This e¡ect is largely caused
MORB. For unaltered MORB, it was assumed by the very di¡erent Os concentrations.
that no dehydration occurs during subduction, For a W value of 37, the 206 Pb/204 Pb of HIMU
since the low water contents of unaltered OIB sources suggests ages between 0.5 and V1
MORB may be accommodated in nominally an- Gyr, if the recycled component in these lavas is
hydrous phases (see [17]). Re, Os, U and Pb abun- indeed subducted altered MORB (Figs. 6a, 7a).
dances of average fresh MORB were used for un- The extreme HIMU samples from Mangaia
altered MORB (see caption of Fig. 6). Mixing (Fig. 5) require melting of a mixture of ca. 80%
lines between depleted upper mantle or EPM V1 Gyr old recycled altered MORB and 20%
and recycled altered MORB are strongly curved peridotite, if complete equilibration between par-

EPSL 5412 4-4-00


H. Becker / Earth and Planetary Science Letters 177 (2000) 287^300 295

and high estimates of average Re abundances in


recycled altered MORB. The new metabasite data
(Table 1) and the Re data on MORB now avail-
able [12,13,15] show that many earlier estimates
of Re abundances in recycled MORB were too
high. The present results are most similar to the
770 Ma, 500 ppt Re endmember model for HIMU
basalts discussed by Hauri and Hart [2]. Even if
the average age of recycled altered MORB in OIB
sources would be 2000 Myr, this still requires 60^
70% of recycled component.
A possible way to relax the age constraints im-
posed by the high W and 206 Pb/204 Pb of recycled
altered MORB could be that HIMU OIB sources
contain mixtures of recycled altered and unaltered
MORB. It is conceivable that unaltered MORB
does not lose any water (and hence Pb and Re) in
Fig. 7. (a) Expanded view of Fig. 6 at low 187 Os/188 Os. subduction zones [17]. Consequently, subducted
EPM, UM and OIB data ¢elds from Fig. 5. Shown are mix- unaltered MORB would develop relatively unra-
ing lines between 350, 700, 1200 and 2000 Myr old altered
MORB and peridotite (see Fig. 6a), and mixing relationships
diogenic Pb isotopic compositions, but highly ra-
between hybrid 2000 Myr old MORB (60% altered, 40% un- diogenic 187 Os/188 Os (Fig. 6b). If the sources of
altered) and peridotite (from Fig. 6b). Thick marks indicate HIMU OIB contain both altered and unaltered
20% intervals of the fraction of recycled altered MORB in a recycled MORB, the age of the recycled material
mixture of eclogite and peridotite. This model assumes that is more di¤cult to evaluate, because such a hybrid
partial melting results in complete equilibration of melt with
both eclogite and peridotite. If the assumptions of the model
likely has a lower W than recycled altered MORB
are correct, very high percentages (80^90%) of relatively (most commonly, values for W in fresh MORB
young ( 6 1000 Myr) recycled altered MORB are required to range between 8 and 13, [35]). Figs. 6b and 7a
explain HIMU OIB compositions (see text). If HIMU OIB show an example that assumes that the source
sources contain mixtures of recycled altered and unaltered of HIMU OIB with the most radiogenic 206 Pb/
MORB, ages of recycled components may be signi¢cantly 204
older, but large amounts of recycled MORB are still indi-
Pb contains altered and unaltered 2 Gyr old
cated. (b) Disequilibrium between partial melt of recycled al- recycled MORB. For a 206 Pb/204 Pb of 21.7, a val-
tered MORB eclogite and peridotite. Endmember composi- ue found in extreme HIMU lavas from Mangaia,
tions are the same as in (a) except that the melt reacts only and the parameters in the caption of Figs. 6b and
with a fraction (300 ppt) of the Os in peridotite. Disequilib- 7a, the proportion of altered to unaltered MORB
rium may result from e¡ective shielding of Os-bearing phases
(sul¢des, alloys) by silicate or oxide host phases that do not
is 60:40% if the crust is 2 Gyr old. The Re con-
react with melt because of compositional and/or kinetic ef- centration of such a mixture is 543 ppt, and the
fects (see text). present-day 187 Os/188 Os is 16.9. In this model, the
proportion of recycled crust in the source of
HIMU OIB has to be at least 70% to explain
tial melt of eclogite and peridotite occurred (Fig. the radiogenic 187 Os/188 Os of basalts from Man-
7a). Some basalts from the Canary islands would gaia and Tubuai (Fig. 7a). This example illus-
require more than 90% of V500 Ma old altered trates that even the presence of recycled unaltered
MORB in their source to explain their high 187 Os/ MORB does not su¤ciently ease the requirement
188
Os in these simple mixing models (Figs. 5, 6a, of large amounts of these materials in HIMU OIB
7a). Signi¢cantly lower estimates for the amount sources.
of recycled crust in HIMU OIB sources were ob- The high proportion of recycled altered MORB
tained in previous studies, resulting from the as- required in the bulk mixing models in Figs. 6 and
sumption of older ages of the recycled component 7a poses a problem that is not easily resolved.

EPSL 5412 4-4-00


296 H. Becker / Earth and Planetary Science Letters 177 (2000) 287^300

Major element compositions of HIMU basalts causing bulk mixing scenarios to be inappropri-
suggest that such high proportions of recycled ate. These alternatives will be discussed in the
MORB in HIMU sources are unlikely. For in- next two sections.
stance, MgO contents of 15^20 wt% and low
SiO2 (43^45 wt%) in some Tubuai and Mangaia 6.2. Radiogenic 187 Os/188 Os of HIMU OIB may
samples [23,30] are inconsistent with partial melt- not re£ect a contribution from recycled altered
ing of a mixture of 80% MORB eclogite with 20% MORB
peridotite. Melting modes of such a composition
will contain high proportions of clinopyroxene Previous studies of lithophile trace element (e.g.
and garnet (e.g. [36], and references therein) and Ce/Pb and Nb/U) and isotopic compositions of
one might expect silica-rich rather than silica-poor HIMU lavas provide strong arguments for the
melt compositions (trondhjemitic^tonalitic presence of recycled altered MORB in the source
liquids, e.g. [37]). Abundances of highly incom- of these lavas (e.g. [30]). One possibility is that the
patible elements such as Nb and Th are enriched budget of lithophile tracers in HIMU OIB is do-
by a factor of 10^40 in HIMU OIB relative to minated by the recycled component while the ra-
MORB. These enrichments have been attributed diogenic Os isotopic composition re£ects the pres-
to low degrees of partial melting of a peridotitic ence of a component with su¤ciently high Os
mantle source [38]. If the mantle source of HIMU abundances that is not recycled MORB. Signi¢-
OIB is dominated by eclogite, then for 60 parts cant amounts of recycled pelagic sediment in
per million (ppm) Nb in alkaline OIB, and 3 ppm HIMU OIB are commonly ruled out as sources
Nb in a MORB eclogite source, 5% melting of of the radiogenic Os because it would dominate
eclogite would be required to explain the trace the Pb budget and impose a low W, high Th/U
element enrichment in the lavas. Such low degrees signal on HIMU OIB [41]. However, as for al-
of partial melting of a clinopyroxene^garnet-rich tered MORB, it is likely that dehydration in sub-
source should yield silica-rich melts [37] that are duction zones will dramatically change the par-
not observed. On the contrary, HIMU-type OIB ent^daughter ratios of lithophile isotopic
are silica-undersaturated. In the past, this has systems. The extent of this modi¢cation is un-
been attributed to the e¡ects of the decrease of known. Even if the lithophile isotopic systematics
the stability ¢eld of olivine during melting of peri- of HIMU OIB can be accommodated with old
dotite at high pressures and in the presence of recycled sediments, it is uncertain whether the
CO2 (e.g. [39]). The behavior of hybrid clinopyr- moderate Os abundances and relatively low Re/
oxene- and garnet-rich mantle during melting in Os in pelagic sediments [42] are su¤cient to ac-
the presence of C^H^O species is not well known. count for the radiogenic Os and mixing relations
Experiments on ma¢c compositions suggest [40] in HIMU OIB.
that melting of eclogite-enriched hybrid mantle Walker et al. [43,44] proposed that radiogenic
187
that contains minor amounts of carbonate should Os in some plume-derived basalts could re£ect
produce melts enriched in SiO2 . contributions from either the outer core or from
These considerations suggest that the Re^Os the lower mantle. Elevated 186 Os/188 Os (the rele-
bulk mixing models of recycled oceanic crust vant isotopic ratio for the 190 Pt^186 Os decay sys-
with peridotite yield results that are di¤cult to tem) in some OIB from Hawaii may indicate a
reconcile with other evidence. Possible explana- contribution from the outer core [32]. However,
tions for this are : (1) the endmember in HIMU the models for a contribution from the outer core
OIB that contains radiogenic 187 Os/188 Os and yield only slightly suprachondritic 187 Os/188 Os
206
Pb/204 Pb may not be recycled MORB, or (2) (V0.132^0.137). These values are di¤cult to rec-
partial melting of mixed lithologies of eclogite (re- oncile with radiogenic 187 Os/188 Os of HIMU OIB
cycled crust) and peridotite may either be a dis- of up to 0.15. In principle, a strati¢ed mantle with
equilibrium process or may result in unexpected high Re/Os in the lower mantle as proposed by
partitioning e¡ects during melting and reaction Walker et al. [44] could also explain the radiogen-

EPSL 5412 4-4-00


H. Becker / Earth and Planetary Science Letters 177 (2000) 287^300 297

ic 187 Os/188 Os of HIMU OIB. However, at lavas suggests that melt extraction occurs more
present, no direct evidence is available that the rapidly than under dry conditions. These circum-
lower mantle has a higher Re/Os than the upper stances render it likely that disequilibrium proc-
mantle. esses may be the norm rather than the exception
Metasomatism of lithospheric mantle by small- in the source of HIMU OIB. The high MgO, Cr,
degree (e.g. carbonatitic) melts was proposed as Ni and Os contents of many HIMU OIB indicate
an alternative explanation for the high 206 Pb/204 Pb that peridotitic mantle must have contributed sig-
of some HIMU basalts. Metasomatism could im- ni¢cantly to the formation of these magmas. A
part a high-U/Pb signal on lithospheric mantle scenario that could potentially explain some of
that would result in radiogenic 206 Pb/204 Pb in the inconsistencies encountered here involves
less than 100 Ma [45]. However, studies of litho- melting of carbon-bearing eclogite and limited
spheric peridotites metasomatized by small-degree or no reaction of the melt with peridotite during
melts [46] do not indicate that this process adds separation from the eclogitic residue. Since a sili-
or removes signi¢cant amounts of Re and Os to ceous partial melt of eclogite is not in equilibrium
produce radiogenic 187 Os/188 Os in less than with lherzolite, once the melt leaves the eclogite
100 Ma. In conclusion, it is di¤cult to identify domain, reaction of melt with olivine and forma-
alternative materials or processes that could be tion of pyroxene-bearing assemblages may be the
suitable to explain both the high 187 Os/188 Os and consequence. The reaction of siliceous melt with
206
Pb/204 Pb in HIMU OIB. olivine in peridotite likely results in a gain of Mg,
a loss of Si from the melt [37] and consequently a
6.3. Are bulk mixing models appropriate? decrease of viscosity and increase in silica under-
saturation of the melt. When conduits develop in
If no source other than recycled altered crust the host peridotite, it can be expected that melt^
can produce high 187 Os/188 Os and 206 Pb/204 Pb, peridotite reaction produces layers of pyroxene on
then the assumptions made for the bulk mixing the walls of the conduit, because the alkaline ba-
models must be re-evaluated. It is commonly as- salts are Ca-rich and may be saturated with py-
sumed that under the conditions of partial melting roxene at an early stage. The e¡ect of these pyrox-
in the asthenospheric mantle, local equilibration enite layers on the next melt batches that use the
of melts and mantle occurs in less than 105 yr [47]. conduits would be to limit interaction of the mag-
This is likely true for asthenosphere-derived large- ma with peridotite.
volume magmas such as MORB, or £ood basalts, For the Os mixing relationships, it is important
tholeiites, picrites and komatiites that may re£ect that the melt equilibrates or interacts with only a
higher than average mantle temperatures (Hawaii, small fraction of the peridotitic Os. Another way
Iceland, Gorgona Island). It is less clear if the this could be achieved is if most Os is contained in
equilibrium assumption is met by the smaller-vol- accessory sul¢des or alloys [50] that are included
ume alkali basalts, basanites and nephelinites that in a peridotite mineral that does not participate in
show the HIMU signature. For instance, experi- the reaction with siliceous melt. If alloys are sta-
mentally determined mineral dissolution rates [48] ble and the dominant host for Os in peridotite in
and assessments of xenolith settling rates [49] sug- the presence of silicate melt, very low solubilities
gest that mantle xenoliths preserved in some of of Os in silicate melts [51] could translate into
the alkali basalts require rapid ascent velocities very high peridotite^melt partition coe¤cients
of these magmas from their source regions (in for Os. Several problems exist with this model.
the order of days or weeks). These fast ascent First, it is not known how homogenously the
velocities can be largely attributed to the low den- bulk of the Os is distributed in peridotites, and
sity of volatile-rich magmas and the increased vol- if peridotite minerals may be excluded from dis-
ume and magma pressure relative to dry magma solution^precipitation reactions for given P^T
during partial melting [49]. The volatile-rich man- conditions and melt compositions. Second is the
tle source required by the composition of these unknown extent of kinetic control on these reac-

EPSL 5412 4-4-00


298 H. Becker / Earth and Planetary Science Letters 177 (2000) 287^300

tions and on the dissolution of Os-bearing phases Os abundances. The implication is that Re was
in melt. Di¡usion coe¤cients of lithophile major lost either during dehydration, the most likely ex-
and trace elements in silicate melts (10310 ^10312 planation, or during alteration of the protoliths
cm2 s31 at 1200^1300³C, e.g. [47]) suggest that for near the ocean ridges. Models for the origin of
some of these elements, cm-scale equilibration HIMU OIB that are believed to contain a com-
may occur in approximately 300^30000 yr. These ponent of recycled altered MORB would require
time scales may be too long to reach equilibrium excessive amounts ( s 80^90%) of the latter to ex-
if volatile-charged magmas are involved. Note, plain their radiogenic 187 Os/188 Os and 206 Pb/204 Pb
however, that theoretical work on the physics of if constraints on Re/Os and U/Pb from the meta-
melt segregation in the mantle suggests longer basites are used. In the light of other evidence,
time scales for melt^residue separation [52]. these estimates appear unreasonably high. It is
Fig. 7b shows an example in which a `reactive' suggested that the bulk mixing models commonly
Os content of mantle of 300 ppt is assumed. The employed may not be applicable and that the ac-
mixing relationships in Fig. 7b require signi¢- tual fraction of recycled component in these OIB
cantly lower proportions of eclogite to peridotite sources is much lower than implied by these mod-
in the source of HIMU OIBs and result in curves els. Possible explanations for the `misbehavior' of
that appear to be more consistent with data ¢elds Os may include kinetic reasons and the shielding
for HIMU OIBs. Disequilibrium models would of Os-rich phases by silicates from reaction with
lessen the problem of explaining relatively radio- melt.
genic 187 Os/188 Os and relatively high Os abundan-
ces in HIMU OIB by recycled MORB with very
low Os abundances. In the case of disequilibrium Acknowledgements
melting and reaction, time scales as well as mag-
ma temperatures and composition likely deter- Al Brandon, Sonia Esperanc°a, John Morgan
mine how much Os from the wall rock will be and Rich Walker provided invaluable intellectual
transferred into the melt. It is interesting to note input through various conversations in the lab
that abundances of lithophile incompatible ele- and during lunch at China Cafe and other places.
ment abundances in HIMU OIBs are generally Thanks go also to other Re^Os folks (the DTM
assumed to re£ect low degrees of partial melting crew, Gerhard Bru«gmann, John Chesley, Robbie
under the assumption that the melts equilibrated Frey, Munir Humayun, Jon Snow, Liz Widom)
with a peridotitic residue (e.g. [38]). The main for sharing techniques and insight. A thorough
argument supporting these models is the strong review of a ¢rst draft by Rich Walker is very
enrichment of LREE and highly incompatible el- much appreciated. O¤cial journal reviews by
ements in many HIMU basalts [30,38]. These Chris Hawkesworth and Jan Kramers helped to
models are di¤cult to reconcile with a disequi- improve the manuscript signi¢cantly. I am grate-
librium scenario unless only Os does not equili- ful to Sonia Esperanc°a and Tim Mock for their
brate with melt but the lithophile elements do, or support in the lab. Work on this project was
the enrichment of incompatible elements in made possible through various NSF research
HIMU basalts re£ects the pre-enrichment of the grants to Rich Walker.[AH]
mantle source by metasomatic melts (e.g. [45])
and disequilibrium during melting.
References

7. Conclusions [1] R.J. Walker, L.M. Echeverria, S.B. Shirey, M.F. Horan,
Re^Os isotopic constraints on the origin of volcanic
rocks, Gorgona Island, Columbia Os isotopic evidence
Eclogites, blueschists and ma¢c granulites tend for ancient heterogeneities in the mantle, Contrib. Miner-
to contain only V40% of the Re abundances ex- al. Petrol. 107 (1991) 150^162.
pected for likely ma¢c protoliths, but comparable [2] E.H. Hauri, S.R. Hart, Re^Os isotope systematics of

EPSL 5412 4-4-00


H. Becker / Earth and Planetary Science Letters 177 (2000) 287^300 299

EMII and HIMU oceanic island basalts from the south and Eocene subduction of the Zermatt-Saas-Fee ophio-
Paci¢c Ocean, Earth Planet. Sci. Lett. 114 (1993) 353^371. lites implications for the geodynamic evolution of the
[3] L. Reisberg, A. Zindler, F. Marcantonio, W. White, D. Central and Western Alps, Contrib. Mineral. Petrol. 132
Wyman, B. Weaver, Os isotope systematics in ocean is- (1998) 269^287.
land basalts, Earth Planet. Sci. Lett. 120 (1993) 149^167. [19] H. Becker, Garnet peridotite and eclogite Sm^Nd mineral
[4] F. Marcantonio, A. Zindler, T. Elliott, H. Staudigel, Os ages from the Lepontine dome (Swiss Alps) new evidence
isotope systematics of La Palma, Canary Islands evidence for Eocene high-pressure metamorphism in the central
for recycled crust in the mantle source of HIMU ocean Alps, Geology 21 (1993) 599^602.
islands, Earth Planet. Sci. Lett. 133 (1995) 397^410. [20] H.-G. Stosch, G.W. Lugmair, Geochemistry and evolu-
[5] M. Roy-Barman, C.J. Alle©gre, 187 Os/186 Os in oceanic is- tion of MORB-type eclogites from the Mu«nchberg Mas-
land basalts tracing crust recycling in the mantle, Earth sif, southern Germany, Earth Planet. Sci. Lett. 99 (1990)
Planet. Sci. Lett. 129 (1995) 145^161. 230^249.
[6] E.H. Hauri, J. Lassiter, D.J. DePaolo, Osmium isotope [21] A.v. Quadt, The Saxonian granulite massif new aspects
systematics of drilled lavas from Mauna Loa, Hawaii, from geochronological studies, Geol. Rundsch. 82 (1993)
J. Geophys. Res. 101 (1996) 11793^11806. 516^530.
[7] J. Lassiter, E.H. Hauri, Os-isotope variations in Hawaiian [22] B. Jamtveit, D.A. Carswell, E.W. Mearns, Chronology of
lavas; evidence for recycling of oceanic lithosphere in the the high-pressure metamorphism of Norwegian garnet
Hawaiian plume, Earth Planet. Sci. Lett. 164 (1998) 483^ peridotites/pyroxenites, J. Metamorph. Geol. 9 (1991)
496. 125^139.
[8] E. Widom, K.A. Hoernle, S.B. Shirey, H.-U. Schmincke, [23] E.H. Hauri, S.R. Hart, Rhenium abundances and system-
Os Isotope Systematics in the Canary Islands and Ma- atics in oceanic basalts, Chem. Geol. 139 (1997) 185^205.
deira Lithospheric Contamination and Mantle Plume Sig- [24] M.B.E. Mo«rk, E.W. Mearns, Sm^Nd isotopic systematics
natures, J. Petrol. 40 (1999) 1^40. of a gabbro-eclogite transition, Lithos 19 (1986) 255^267.
[9] J.E. Snow, L. Reisberg, Erratum of `Os isotopic system- [25] Y. Xiong, S.A. Wood, Experimental determination of the
atics of the MORB mantle results from altered abyssal solubility of ReO2 and the dominant oxidation state of
peridotites', Earth Planet. Sci. Lett. 136 (1995) 723^733. rhenium in hydrothermal solutions, Chem. Geol. 158
[10] T. Meisel, R.J. Walker, J.W. Morgan, The osmium iso- (1999) 245^256.
topic composition of the Earth's primitive upper mantle, [26] J.H. Jones, M.J. Drake, Geochemical constraints on core
Nature 383 (1996) 517^520. formation in the Earth, Nature 322 (1986) 221^228.
[11] E. Widom, S.B. Shirey, Os isotope systematics in the [27] M. Roy-Barman, G.J. Wasserburg, D.A. Papanastassiou,
Azores Implications for mantle plume sources, Earth M. Chaussidon, Osmium isotopic compositions and Re^
Planet. Sci. Lett. 142 (1996) 451^466. Os concentrations in sul¢de globules from basaltic glass,
[12] J. Hertogen, M.-J. Janssens, H. Palme, Trace elements in Earth Planet. Sci. Lett. 154 (1998) 331^347.
ocean ridge basalt glasses: Implications for fractionations [28] H. Becker, K.P. Jochum, R.W. Carlson, Constraints from
during mantle evolution and petrogenesis, Geochim. Cos- high-pressure veins in eclogites on the composition of £u-
mochim. Acta 44 (1980) 2125^2143. ids in subduction zones, Chem. Geol. 160 (1999) 291^
[13] P. Schiano, J.-L. Birck, C.J. Allegre, Osmium-strontium- 308.
neodymium-lead isotopic covariations in mid-ocean ridge [29] A.W. Hofmann, W.M. White, Mantle plumes from an-
basalt glasses, Earth Planet. Sci. Lett. 150 (1997) 363^379. cient oceanic crust, Earth Planet. Sci. Lett. 57 (1982)
[14] G. Ravizza, C.E. Martin, C.R. German, G. Thompson, 421^436.
Os isotopes as tracers in sea£oor hydrothermal systems [30] C. Chauvel, A.W. Hofmann, P. Vidal, HIMU-EM: The
metalliferous deposits from the TAG hydrothermal area, French Polynesian connection, Earth Planet Sci. Lett. 110
26³N Mid-Atlantic Ridge, Earth Planet. Sci. Lett. 138 (1992) 99^119.
(1996) 105^119. [31] S.R. Hart, E.H. Hauri, L.A. Oschmann, J.A. Whitehead,
[15] M. Roy-Barman, C.J. Alle©gre, 187 Os/186 Os ratios of mid- Mantle plumes and entrainment isotopic evidence, Science
ocean ridge basalts and abyssal peridotites, Geochim Cos- 256 (1992) 517^520.
mochim. Acta 58 (1994) 5043^5054. [32] A.D. Brandon, R.J. Walker, J.W. Morgan, M.D. Nor-
[16] A.D. Brandon, H. Becker, R.W. Carlson, S.B. Shirey, man, H.D. Prichard, Coupled 186 Os-187 Os evidence for
Isotopic constraints on time scales and mechanisms of core^mantle interaction, Science 280 (1998) 1570^1573.
slab material transport in the mantle wedge evidence [33] V.C. Bennett, T.M. Esat, M.D. Norman, Two mantle-
from the Simcoe mantle xenoliths, Washington, USA, plume components in Hawaiian picrites inferred from cor-
Chem. Geol. 160 (1999) 387^407. related Os^Pb isotopes, Nature 381 (1996) 221^224.
[17] H. Becker, K.P. Jochum, R.W. Carlson, Trace element [34] E.H. Hauri, M.D. Kurz, Melt migration and mantle chro-
fractionation during dehydration of eclogites from high- matography, 2: a time-series Os isotope study of Mauna
pressure terranes and the implications for element £uxes Loa volcano, Hawaii, Earth Planet. Sci. Lett. 153 (1997)
in subduction zones, Chem. Geol. 163 (2000) 65^99. 21^36.
[18] D. Rubatto, D. Gebauer, M. Fanning, Jurassic formation [35] W. White, 238 U/204 Pb in MORB and open system evolu-

EPSL 5412 4-4-00


300 H. Becker / Earth and Planetary Science Letters 177 (2000) 287^300

tion of the depleted mantle, Earth Planet. Sci. Lett. 115 [45] A. Halliday, D.-C. Lee, S. Tommasini, G. Davies, C.
(1993) 211^226. Paslick, J. Fitton, D. James, Incompatible trace elements
[36] M.M. Hirschmann, E.M. Stolper, A possible role for gar- in OIB and MORB and source enrichment in the sub-
net pyroxenite in the origin of the `garnet signature' in oceanic mantle, Earth Planet. Sci. Lett. 133 (1995) 379^
MORB, Contrib. Mineral. Petrol. 124 (1996) 185^208. 395.
[37] R.P. Rapp, N. Shimizu, M.D. Norman, G.S. Applegate, [46] M.R. Handler, V.C. Bennett, T.M. Esat, The persistence
Reaction between slab-derived melts and peridotite in the of o¡-cratonic lithospheric mantle Os isotopic systematics
mantle wedge experimental constraints at 3.8 GPa, Chem. of variably metasomatised southeast Australian xenoliths,
Geol. 160 (1999) 335^356. Earth Planet. Sci. Lett. 151 (1997) 61^75.
[38] M. Caro¡, M.R.C., G. Guille and J. Cotten, Partial melt- [47] A.W. Hofmann, S.R. Hart, An assessment of local and
ing below Tubuai (Austral Islands, French Polynesia), regional isotopic equilibrium in the mantle, Earth Planet.
Contrib. Mineral. Petrol. 127, 369^382. Sci. Lett. 38 (1978) 44^62.
[39] D.H. Eggler, The e¡ect of CO2 upon partial melting of [48] M. Brearley, C. Scarfe, H. Brenner, Dissolution rates of
peridotite in the system Na2 O-CaO-Al2 O3 -MgO-SiO2 - upper mantle minerals in an alkali basalt melt at high
CO2 to 35 kb, with an analysis of melting in a perido- pressure an experimental study and implications for ultra-
tite-H2 O-CO2 system, Am. J. Sci. 278 (1978) 305^343. ma¢c xenolith survival, J. Petrol. 27 (1986) 1157^1182.
[40] G.M. Yaxley, D.H. Green, Experimental demonstration [49] F.J. Spera, Carbon dioxide in petrogenesis III role of
of refractory carbonate-bearing eclogite and siliceous melt volatiles in the ascent of alkaline magma with special
in the subduction regime, Earth Planet. Sci. Lett. 128 reference to xenolith-bearing ma¢c lavas, Contrib. Miner-
(1994) 313^325. al. Petrol. 88 (1984) 217^232.
[41] D. BenOthman, W. White, J. Patchett, The geochemistry [50] S.R. Hart and G.E. Ravizza, Os partitioning between
of marine sediments, island arc magma genesis, and crust- phases in lherzolite and basalt, in: A. Basu and S.R.
mantle recycling, Earth Planet. Sci. Lett. 94 (1989) 1^21. Hart (Eds.), Earth Processes: Reading the Isotopic
[42] B. Peucker-Ehrenbrink, G. Ravizza, A. Hofmann, The Code, American Geophysical Union, 1996.
marine 187 Os/186 Os record of the past 80 million years, [51] A. Borisov, H. Palme, Experimental determination of Os
Earth Planet. Sci. Lett. 130 (1995) 155^167. metal/silicate partitioning, Neues Jahrb. Mineral. Abh.
[43] R. Walker, J. Morgan, M. Horan, Osmium-187 enrich- 172 (1998) 347^356.
ments in some plumes evidence for core-mantle interac- [52] D. McKenzie, The extraction of magma from the crust
tion?, Science 269 (1995) 819^822. and mantle, Earth Planet. Sci. Lett. 74 (1985) 81^91.
[44] R.J. Walker, J.W. Morgan, E. Hanski, V.F. Smolkin, Re^ [53] A.W. Hofmann, Chemical di¡erentiation of the earth: the
Os systematics of early Proterozoic ferropicrites, Pechenga relationship between mantle, continental crust and ocean-
Complex, NW Russia: evidence for ancient 187 Os-en- ic crust, Earth Planet. Sci. Lett. 90 (1988) 297^314.
riched plumes, Geochim. Cosmochim. Acta 61 (1997) [54] A. Zindler, S.R. Hart, Chemical Geodynamics, Annu.
3145^3160. Rev. Earth Planet. Sci. 14 (1986) 493^571.

EPSL 5412 4-4-00

You might also like