ReOs isotope systematics of HIMU and EMII oceanic island basalts from the south Pacific Ocean

Earth and Planetary Science Letters, 114 (1993) 353-371 Elsevier Science Publishers B.V., Amsterdam

353

[CL]

Re-Os isotope systematics of HIMU and EMII oceanic island basalts from the south Pacific Ocean Erik H. Hauri and Stanley R. Hart Department of Geology and Geophysics, Woods Hole Oceanographic Institution, Woods Hole, MA 02543 USA Received May 5, 1992; revision accepted November 20, 1992

ABSTRACT The Re-Os and complementary Sr, Nd and Pb systematics of 24 oceanic island basalts from the islands of Savaii, Tahaa, Rarotonga, Rurutu, Tubuai and Mangaia are investigated. Re concentrations range from 100 to 1621 ppt (parts per trillion), while Os concentrations vary from 26 to 750 ppt. The Re and Os concentration variations suggest that fractionation and accumulation of olivine, or a low Re/Os phase in conjunction with olivine, is important in determining the Os concentration and the Re/Os ratio of the erupted basalt. 187Os/186Os in EMIl basalts from Samoa and Tahaa varies from 1.0261 to 1.1275. These ratios are mostly within estimates for depleted upper mantle, and do not constrain the involvement of recycled continental crust in the origin of the EMIl signature. 187Os/186Os ratios in HIMU basalts from Rurutu, Tubuai and Mangaia range from 1.1159 to 1.2473, and provide strong evidence for the role of subducted oceanic crust in the HIMU source. The Pb-Pb systematics constrain the range of possible ages and 238U/2°4pb and T h / U ratios of the subducted crust; this crust is estimated to pass through the subduction zone with Rb/Sr, Sm/Nd, Lu/Hf and T h / U ratios similar to fresh MORB. The homogeneity of the Os isotopic compositions in the Tubuai and Mangaia basalts indicates that interaction of these basalts with low 187Os/186Os mantle had an insignificant effect on the Os isotopic composition of the erupted magmas. This requires a network of channels, veins or cracks capable of delivering melt from the source region (plume) to the surface fast enough to avoid interaction with the depleted upper mantle and the oceanic lithosphere. The possible identification of the HIMU signature (high 2°6pb/2°4pb, low 87Sr/86Sr) with recycled oceanic crust suggests the possible presence of segments of recycled crust, with independent histories, in other oceanic mantle sources, including that of some mid-ocean ridge basalts.

I. Introduction F o l l o w i n g the p i o n e e r i n g w o r k o f L u c k et al. [1] the R e - O s i s o t o p e system has r e c e i v e d conside r a b l e a t t e n t i o n f r o m th e g e o c h e m i c a l c o m m u n i t y over t h e past 10 years. T h e /3 d e c a y of 187Re to 187Os p r o d u c e s m e a s u r a b l e v a r i a t i o n s in the a b u n d a n c e o f 187Os in s a m p l e s o f n a t u r a l osmi um. T h e very d i f f e r e n t b e h a v i o r s of R e and Os d u r i n g basaltic m e l t g e n e r a t i o n , as well as t h e ir s i d e r o p h i l e and c h a l c o p h i l e characteristics, provide i n f o r m a t i o n q u it e d i f f e r e n t f r o m the o t h e r

Correspondence to: E.H. Hauri, Department of Geology and Geophysics, Woods Hole Oceanographic Institution, Woods Hole, MA 02543 USA.

p a r e n t - d a u g h t e r systems (Rb-Sr, Sm-Nd, L u - H f and U - T h - P b ) . R e - O s studies have d e m o n s t r a t e d that b o t h o c e a n i c and c o n t i n e n t a l crustal m a t e r i als have m u c h h i g h e r R e / O s ratios than perid o t i t es [2-4]. R e c o n c e n t r a t i o n s in crustal m a t e r i als are typically 1 0 - 1 0 0 times h i g h e r t h a n in p e r i d o t i t e s , while Os c o n c e n t r a t i o n s are 1 0 10,000 times lower t h a n in peridotites. T h u s R e is t h o u g h t to be m o d e r a t e l y i n c o m p a t i b l e , and Os highly c o m p a t i b l e , d u r i n g m e l t g e n e r a t i o n . T h e high R e c o n c e n t r a t i o n s and high R e / O s ratios of crustal materials, relative to m a n t l e p e r i d o t i t e , m a k e the R e - O s system a p o t en t i al l y p o w e r f u l t r a c e r for e x a m i n i n g crustal recycling in t h e m a n tle. T h e isotopic s i g n a t u r e s o f s o m e o c e a n i c islands have b e e n i n t e r p r e t e d as b ei n g d e r i v e d

0012-821X/93/$06.00 © 1993 - Elsevier Science Publishers B.V. All rights reserved

354

from crustal material mixed into the mantle sources of these islands [5-7]. Specifically, the EMIl signature of high SVSr/S6Sr and high 2°7pb/Z°4pb ratios [8] has been attributed to the recycling of sediments in the mantle [9,10]. In addition, the high Pb isotope ratios which characterize the HIMU mantle signature [8] have been attributed to subducted oceanic crust [5,6]. The trace element characteristics of H I M U have also been attributed to recycled oceanic lithosphere [11,12]. A number of other oceanic island isotope signatures have also been attributed to this process [7]. In order to constrain theories of crustal recycling in the mantle, we have measured the Re-Os and Sr-Nd-Pb isotope systematics in a suite of oceanic island basalts from various islands in the south Pacific Ocean. The combined Os, Sr, Nd and Pb isotope results demonstrate that recycling of oceanic crust is responsible for some of the isotopic heterogeneity observed in the oceanic mantle and place constraints on the composition and age of the recycled crust in the mantle sources of these basalts. In addition, high mantle-derived 1870s/1860s signatures in H I M U basalts provide a tracer for the interaction of these basalts with the depleted mantle. 2. Geological background and sample selection The southern Pacific Ocean is home to many linear island chains of volcanic origin. In the Samoa and Society island chains, as well as in the Cook-Austral Islands east of Mangaia, the age of the volcanism decreases monotonically toward the east-southeast [13-16], supporting a hotspot origin for these features. The basalt samples analyzed in this study were collected in the fall of 1990 specifically for Re-Os analysis. The samples analyzed are from the islands of Savaii (Western Samoa), Rarotonga and Mangaia (Cook Islands), Rurutu and Tubuai (Austral Islands) and Tahaa (Society Islands). Due to the high R e / O s ratios measured, the ages of the various islands examined in this study are important, in that samples from the older islands require corrections for radiogenic ingrowth of ~S7Os since eruption. Although none of the samples have been directly dated, some constraints on their ages exist from field relation-

E.H. HAURI and S.R. HART

ships. The samples from Savaii are all young ( < 1 Ma) shield-capping flows, and two are from historical eruptions. The samples from Rarotonga are from the Te Manga group, dated at 1.2-2.3 Ma [15]. Reported ages from Tahaa range from 1.1 Ma to 3.4 Ma [17]. The ages of the samples from Rurutu are not well constrained; the island is characterized by two separate stages of volcanism from 0.88 to 1.12 Ma and from 11.4 to 14.4 Ma [13,16; H. Barsczus, pers. commun.]. For Tubuai, reported ages range from 5.7 Ma to 10 Ma [14]. The oldest island sampled is Mangaia, with reported ages in the range 16-20 Ma [15,18]. When possible, samples rich in olivine and pyroxene phenocrysts were chosen from each island in order to analyze samples with high Os concentrations and low R e / O s ratios, in an attempt to minimize uncertainties in the age corrections. The analytical procedures are given in the appendix. 3. Overview of Re-Os systematics Due to the geochemical characteristics of the two elements, the Re-Os isotopic system possesses some unique features which distinguish it from other radiogenic isotopic systems. During partial melting of mantle peridotite, Re behaves as a moderately incompatible element, whereas Os is strongly concentrated in the residue [19]. As a result, mantle-derived melts are characterized by high Re concentrations and very high R e / O s ratios relative to the residues of melting. Segregation of the melt leaves a residue with a lower R e / O s ratio than before melting. The behavior of Os contrasts with that of Sr, Nd and Pb, which are strongly partitioned into the magma during melting. The result is a much more extreme pare n t / d a u g h t e r fractionation in the Re-Os system than in the other isotopic systems. Due to its high R e / O s ratio, continental crust develops very high 1870s/1860s with time. In a study of continental river sediments and glacial loess deposits, Esser [3] measured 1870S/1860S ratios of 5.6-13.4 for these sediments, which act as integrators of large areas of continental crust. Although the crust is markedly heterogeneous in R e / O s and 1 8 7 0 S / 1 8 6 0 S ratios, Esser has suggested a lSVOs/186Os ratio of 10.5 for average continental crust, as well as average Re (400 ppt) and Os (50 ppt) concentrations [3], consistent

Re-Os ISOTOPE SYSTEMATICS FROM THE SOUTH PACIFIC

with the existing measurements of continental crustal material [20,21]. The Os isotopic evolution of the earth through geological time is currently an issue accompanied by considerable uncertainty. Due to the possibility of R e / O s fractionation during accretion and core formation, the Os isotope evolution of the silicate earth may be different from that of the whole earth. Analyses of group IIAB and IIIAB iron meteorites [22,23] indicate an initial 187Os/ aS6Os of 0.794 _+ 0.010. Although an early study of osmiridiums of known age [24] suggested that the upper mantle followed a Re-Os evolution with a present day lSVOs/lS6Os of 1.04, subsequent osmiridium studies [25,26] have demonstrated significant Os isotopic heterogeneity within and between localities, precluding the identification of a well-constrained evolution curve. The measured present day 1870S/186Os ratios of chondrites range from 1.00 to 1.16 [27,28], with an average of 1.07. Luck and Allegre [29] have derived a present day bulk silicate earth estimate of 187Os/ 186Os 1.05-1.06 based on analyses of ophiolite peridotites. As an alternative, Martin et al. [30] have suggested a present day value of ~SVOs/~86Os 1.10 for the bulk silicate earth based on analyses of abyssal peridotites and a crust-mantle mass balance. Present day measured 1870s/1860s ratios for peridotite xenoliths from South Africa and Siberia [31,32] and from the Ronda ultramfic complex [33] lie in the range < 0.90-1.10. Major element correlations with ~SVOs/~S6Os in the Ronda peridotites suggests that a 1870s/186Os of 1.06-1.08 may be representative of undifferentiated mantle, similar to the ordinary chondrite average of 1.07. In detail, however, the Os isotope evolution of the earth's mantle is only broadly constrained. 4. Results

The Sr, Nd, U, Th and Pb concentrations and the Sr, Nd and Pb isotope results are given in Table 1. The concentrations of these elements are similar to those typically reported for alkali basalts [e.g. 11]. 238u/Z°4pb (/~) ratios vary from 18 to 61 and T h / U ratios range from 3.14 to 4.98. Both ratios are high relative to mid-ocean ridge basalts [e.g. 34]; differences in source composition and melting processes are both likely to account for this. The Sr-Nd-Pb isotope ratios

355 given in Table 1 agree well with previously reported data for these islands [9-11,35,36]. The islands of Savaii and Tahaa have high 878r/86Sr and 2°7pb/2°4pb ratios, and the islands of Rurutu, Tubuai and Mangaia are characterized by very radiogenic Pb isotope signatures. The two samples from Rurutu delineate a large fraction of the range of previously reported data from this island [36,37]. The Re-Os data are given in Table 2 and plotted in Figs. 1 and 2. Also plotted in Fig. 1 are Re and Os concentration data for basalts and komatiites from Gorgona Island [38], komatiites from Munro township [39], basalts from Hawaii [4] and other oceanic and continental basalts [40]. Re concentrations in the south Pacific basalts range from 100 to 1621 ppt (parts per trillion, 10 i2 g/g), and Os concentrations vary from 26 to 750 ppt. The Os concentrations are systematically lower than Os concentrations in komatiites [38,39]. 187Re/186Os varies from 5.34 to 650, and exhibits a clear negative correlation with Os concentration (Fig. 1B). Several other aphyric basalts from Savaii, Rurutu, Tubuai, and Mangaia have also been analyzed and found to have low Os concentrations (4-15 ppt); due to the high Os blank of the analytical technique, the 1870s/1860s ratios of these basalts are unreliable, and so these data are not reported here. 1870s/1860s ratios for the various islands are also reported in Table 2. Measured 187Re/~S6Os ratios in some of the samples were high enough to require age corrections on ~SVOs/186Os. Initial 1870s/1860s ratios were calculated using the middle of the age ranges listed in Table 2, and the age-corrected 1870S/1860S ratios are plotted with the STSr/86Sr, 143Nd/144Nd and 2°6pb/Z°4pb data in Fig. 2, along with ~SVOs/186Os data from Hawaii and Iceland reported by Martin [4]. Initial 187Os/ 186Os ratios for Samoa vary from 1.0261 to 1.0739, while t870S/1860S ranges from 1.0762 to 1.1275 for Tahaa and from 1.0418 to 1.0725 for Rarotonga. Most of these values are between the two lSTOs/lS6Os estimates for the bulk silicate earth [29,30]. In order to evaluate the data relative to depleted upper mantle, also displayed in Fig. 2 are whole rock 1 8 7 0 S / 1 8 6 O s analyses of peridotites with 87Sr/86Sr, 143Nd/144Nd and 2°6pb/ 2°4pb similar to MORB, as determined from analyses of clinopyroxene separates [31,33; Hauri,

356

E . H . H A U R I and S.R. H A R T

unpublished data]. Excluding the peridotite with 187Os/186Os o f 1.10, w h i c h is a g a r n e t - r i c h s a m p l e

l a y e r [33], t h e M O R B - l i k e peridotites have 187Os/186Os a r o u n d 1 . 0 4 - 1 . 0 5 , w h i c h is l o w r e l a -

f r o m R o n d a t h a t is p r o b a b l y a d i s a g g r e g a t e d m a f i c

tive t o b o t h b u l k s i l i c a t e e a r t h

estimates. Thus

TABLE 1 Sr, Nd and Pb isotopic data for oceanic island basalts from the south Pacific Ocean Sample

8 7 S r / 8 6 S r 143Nd/143Nd 206pb/204I~ 207pb/2041~ 208pb/204Fb

Sr

Nd

U

Th

II

Savaii SAV-B-5 SAV-B-6 SAV-B-7 SAV-B-8 SAV-B-15

.705566 .705786 .706592 .706342 .706004

.512732 .512716 .512689 .512685 .512690

18.693 18.785 18.734 18.889 18.797

15.598 15.610 15.604 15.616 15.606

38.938 39.121 39.045 39.144 39.038

392 468 551 711 570

25.32 29.86 49.53 49.72 52.53

.628 .693 1.55 1.54 1.38

2.75 3.27 4.81 6.24 5.18

2.10 2.48 2.86 4.90 3.86

.705958 .706226

.512639 .512623

19.158 19.241

15.630 15.657

38.922 39.010

592 700

43.95 51.98

1.71 1.90

6.51 7.25

4.53 2.02

.704295 .704243 .704873 .704344

.512729 .512760 .512690 .512683

18.383 18.724 18.553 18.842

15.494 15.551 15.556 15.543

38.763 38.892 38.967 38.804

753 456 644 580

42.41 23.35 47.63 79.69

1.42 .557 1.37 2.81

5.37 2.12 5.51 14.0

4.12 1.62 4.31 8.53

.703283 .703366

.512903 .512925

20.847 20.050

15.744 15.656

40.161 38.804

818 820

22.65 52.75

2.21 1.30

8.12 4.88

2.25 2.48

.702817

.512914

21.115

15.746

40.343

683

51.60

.512910 .512894 .512893 .512904 .512927

20.818 20.960 21.195 20.956 21.118

15.750 15.758 15.779 15.756 15.758

40.203 40.396 40.498 40.234 40.330

494 427 427 434 1056

32.01 31.04 33.69 22.86 68.01

7.96 7.89 4.72 3.83 4.06 3.41 12.2 12.2

2.87

.702822 .702834 .702791 .702996 .702800

1.78 1.76 1.23 .894 1.14 .915 3.15 3.14

.702941

.512842

21.529

15.809

40.547

407

32.35

.702912 .703081 .702827

.512864 .512864 .512842

21.678 21.508 21.784

18.807 15.804 15.813

40.512 40.405 40.734

347 244 287

23.22 21.11 19.97

1.02 1.02 .730 .673 .635 .636

3.57 3.54 2.45 2.25 2.18 2.17

<150

<30

<5

<1

Tahaa TAA-B-7 TAA-B-26

Rarotonga RAR-B-1 RAR-B-9 RAR-B-12 RAR-B-16

Rurutu RRT-B-21 RRT-B-30 Tubuai TBA-B-3 replicate TBA-B-15 TBA-B-16 TBA-B-19 TBA-B-22 TBA-B-23 replicate

2.73 2.02 1.56 1.54 6.14

Mansaia MGA-B-21 replicate MGA-B-25 MGA-B-26 MGA-B-47

replicate Blanks(]~

1.96 1.40 1.33 1.54

<50

Pb was run using the silica gel-phosphoric acid technique on single Re filaments in the static mode using five Faraday cups, and the data were corrected for fractionation relative to the NBS 981 values given by Todt et al. [89]. Reproducibility of the Pb runs is 0.02%/amu based on repeat runs of NBS 981. Sr was run on single W filaments with a Ta oxide activator in a three-step dynamic mode with five Faraday cups and fractionation corrected relative to 86Sr/S8Sr = 0.1194. The 87Sr/86Sr data are adjusted to a value of 0.71022 for NBS 987, and repeat analyses of this standard indicate a reproducibility of 0.0045%. Nd was loaded on the Ta side filament of a Ta-Re double filament assembly, and was run in a three-step dynamic mode with five Faraday cups, and fractionation corrected relative to 146Nd/144Nd = 0.7219. 143Nd/144Nd was adjusted to a value of 0.511860 for the La Jolla Nd standard, and repeat analysis of this standard indicate a reproducibility of 0.0040%. U and Th were loaded together on a single Re filament with colloidal graphite and analyzed using the electron multiplier, and were corrected for fractionation based on repeat analyses of SRM 500 U. Trace element concentrations are accurate to better than 1%. Analytical blanks are insignificant for all analyses reported here.

Re-Os ISOTOPE SYSTEMATICS FROM THE SOUTH PACIFIC

357

The initial

t h e r e s u l t s f r o m S a m o a , T a h a a a n d R a r o t o n g a lie i n t h e r a n g e o f 187Os/186Os v a l u e s c h a r a c t e r i s t i c of depleted and undifferentiated

187Os/186Os r a t i o s

v a r y f r o m 1.2334

t o 1.2473 a t M a n g a i a , f r o m 1.1927 t o 1.2147 a t T u b u a i , a n d f r o m 1.1159 t o 1.2449 a t R u r u t u .

mantle.

TABLE 2 Re-Os isotope data for oceanic island basalts from the south Pacific Ocean Sample

Ages

187Os/186Os measured

Re(ppt)

Os(ppt)

187Re/186Os

187Os/186Os initial

Phenocyrst %

1911 1907 <1 Ma <1 Ma <1 Ma

1.0261 1.0510 1.0444 1.0739 1.0629

198

132 170 213 26.2 219

37.1

1.0261 1.0510 1.0438 1.0739 1.0629

1.3-3.4 Ma 1.3-3.4 Ma

1.0789 1.1333

64.5 48.0

73.3 155

1.0762 + 0.0049 1.1275 ± 0.0068

5% OL, 5% CPX 5% OL, 10% CPX

1.2-2.3 Ma 1.2-2.3 Ma

1.0609 1.0910

212 98.3

650

1.0725 + 0.0108

15% OL, 25% CPX 5% OL

1.2-2.3 Ma

1.0424

556

20.1

1.0418 ± 0.0033

30% OL, 30% CPX

1.2-2.3 Ma

1.1128

.88-12.8 Ma .88-12.8 Ma

1.2493 1.1214

309 388

313 315

39.5 49.3

1.2449 ± 0.0084 1.1159 ± 0.0089

10% OL, 15% CPX 5% CPX

6-10 Ma

1.2179

646

24.9

1.2147 ± 0.0047

5% OL, 10% CPX

6-10 Ma 6-10 Ma

1.2134 1.2039 1.2069 1.2001 1.2165

407 397 359 191

254 221 213 256 421

56.6 35.3

1.2060 ± 0.0061 1.2008 ± 0.0052

trace OL trace OL

56.5 54.9

1.1927 :t 0.0061 1.2093 ± 0.0061

trace OL 20% OL, 20% CPX

589 554

16.0

1.2097 ± 0.0043

10% OL

265 272

51.7

1.2473 + 0.0066

10% OL, 25% CPX

348 273 259 751

55.2 30.1

1.2354 ± 0.0068 1.2444 + 0.0054

20% OL, 20% cpx 15% OL, 20% CPX

5.34

1.2334 ± 0.0040

60% OL, 10% CPX

Savaii SAV-B-5 SAV-B-6 SAV-B-7 SAV-B-8 SAV-B-15

± 0.0031 ± 0.0032 ± 0.0031 ± 0.0032 ± 0.0032

trace OL 5% OL aphyric trace OL,CPX 5% OL

Tahaa TAA-B-7 TAA-B-26 replicate

118 186 186

Rarotonga RAR-B-1 RAR-B-9 replicate RAR-B-12 replicate RAR-B-16

1569 1621 283 276

50.5

trace CPX

Rurutu RRT-B-21 RRT-B-30

Tubuai TBA-B-3 replicate TBA-B-15 TBA-B-16 replicate TBA-B-19 TBA-B-22 replicate TBA-B-23 replicate

6-10 Ma 6-10 Ma 6-10 Ma

1.2116 1.2120

16-20 Ma

1.2605 1.2644

16-20 Ma 16-20 Ma

1.2516 1.2523 1.2541 1.2350

361 586 569 231 226

Mangaia MGA-B-21 replicate1 replicate2 MGA-B-25 MGA-B-26 replicate MGA-B47

16-20 Ma

339 347 354 480 200 100

(lS7Os/186Os)measured is the measured 187Os/186Os after correction for analytical blank and spike contributions. Re and Os concentrations are reproducible to 2.6% and 4.6% respectively. Errors on the initial 187Os/186Os ratios represent the combined uncertainties on sample ages, 187Re//186Os(7%), and analytical reproducibility on the m e a s u r e d 187Os//186Os (0.3%). Data averages are taken for samples with replicate analyses. Ages are estimated from sample locations and published ages for these islands [13-15,17]. Phenocryst abundances are estimated from visual inspection of the hand specimens.

E.H. H A U R I and S.R. H A R T

358 I0000

gala samples are the highest yet reported for young oceanic island basalts. The homogeneity in ~87Os/186Os observed for the Tubuai and Man-

A

Gorgona Tholeiites

Komatiites

MAT iooo

Hawaii" I

&

~

~

Adk

I 30

ePUM StH loo

~ 0

• •

•

•

[]

•

angaia Tubuai Rurutu Tahaa Samoa Rarotonga ......

[] Mangaia 0 Tubuai /k Rurutu • Tahaa

HIMU

~

Samoa

• Rarotonga • Peridotites

120

-

;'o

. . . . .

i&

. . . . .

~o'oo

.....

~

,oooo

Icelan

R~nion ~EMII

110

Os (ppt)

AP • T

B

~Gorgona Tholeiites

}oo

I~

A

• I

DMM • 7013

702

7O4

70'5

706

707

87Sr ! 86Sr

IOOO

I

MAT

Ha

_ R

Ioo

Io

A

130-

.ZX~ /

Tubuai O Rurutu • Tahaa • Samoa • Rarotonga

(~

[]

[] Mangaia 0 Tubuai A Rurutu • Tahaa • Samoa • Rarotonga 120" • Peridotites

HIMU

e

PUM 1000

I00

Reunion[]

I0000

Os (ppt)

H~i

[cetatad

EMII AP

100 ¸

DMM

C

I00-

5126

/x

2 O • • •

130"--

C

[] O A •

Mangaia Tubuai Rurutu Tahaa Samoa Rarotong~ 2 Io

5130

•~9

[]

1

512g

143Nd / 144Nd

O

[]

10

z

5128

5127

Mangaia Tubuai Rurutu Tahaa • Samoa • Rarotonga • Peridotites

120

oO ~ HIMU

~ 410

6 '0

8 '0

eland

]00

n"

N d (ppm) Fig, 1. (A) Re versus Os concentrations and (B) lS7Re/lS6Os versus Os concentrations for OIB from the Pacific Ocean; (B) shows a negative correlation, suggesting the influence of a phase with low R e / O s in which Os is compatible. Other Re-Os data are from [4,38-40] St. StH~ St. Helena; T = Tubuai; F = FAMOUS; I = Iceland; R = Reunion; L = Loihi. PUM is the primitive mantle estimate of Morgan [19]. (C) R e / N d versus Nd concentration. The negative trend suggests that Re is more compatible than Nd.

These results are much higher than both depleted mantle values and bulk silicate earth estimates. The 1870s/1860s ratios measured for the Man-

Ar

~ DMM

100

17

18

19

20

21

22

2°6pb / 2°4pb

Fig. 2. Sr-Nd-Pb-Os isotope ratio diagrams for south Pacific oceanic island basalts. Also shown are data from Hawaii, Iceland and Reunion [4,40], the range of data for abyssal peridotites (AP) [4,42], and peridotites with Sr, Nd and Pb isotope ratios similar to MORB [31,33; Hauri, unpublished data]. Error bars on the lS7Os/lS6Os ratios represent the combined uncertainties on sample ages, 187Re/186Os (7%), and analytical reproducibility on t87Os/186Os (0.3%), and for most samples are smaller than the symbols.

Re-Os ISOTOPE SYSTEMATICS FROM THE SOUTH PACIFIC

gala samples is consistent with the relative homogeneity in 87Sr/S6Sr, 143Nd/144Nd and Pb isotopes from these islands [35,36]. The data from Rurutu, Tubuai and Mangaia display a positive correlation between lSTOs/lS6Os and 2°6pb/ 204 Pb. 5. Discussion

5.1 Re and Os concentrations From Table 2, it is clear that samples with high abundances of olivine + clinopyroxene phenocrysts have higher Os concentrations and lower 187Re/186Os ratios than aphyric samples. The Re-Os concentration systematics in komatiites also suggest the involvement of a low R e / O s phase in which Os is compatible [38,39]. Preliminary Re and Os partitioning data reported by Watson et al. [41] suggest that clinopyroxene should have low Re/Os; however both elements are incompatible in this phase. Fractional crystallization of an early liquidus phase, such as olivine, chromite, or trace sulfides associated with olivine, may be capable of lowering the Os concentration of a crystallizing magma, and raising the R e / O s ratio. Differentiated magmas would be expected to have low Os concentrations and high R e / O s ratios. Cumulative basalts, such as the ankaramite samples TBA-B-22, RAR-B-12, MGA-B-21, MGA-B-25 and MGA-B-26 and the picrite MGA-B-47 (Table 2) have high Os concentrations and low R e / O s ratios. However, some samples with only moderate amounts of phenocrysts show both high Os concentrations (TBA-B-3, 646 ppt) and low Os concentrations (Tahaa samples, 48-65 ppt). This suggests that olivine may not be directly involved in controlling the Os concentrations in these basalts, although the extent to which individual samples may have dissolved cumulate olivine cannot be evaluated here. In this context, the Os concentrations of 100-200 ppt in Samoa basalts SAV-B-5, SAV-B-6 and SAV-B-7, which are largely free of phenocrysts, may be representative of the Os concentration of a nearprimary melt of mantle peridotite. Relative to Morgan's estimate of Os = 3.1 ppb for the bulk silicate earth [19], this would suggest an apparent peridotite/melt partition coefficient for Os of 10-30.

359

With one exception, Re concentrations in these basalts are systematically lower than Re concentrations from Hawaiian tholeiites [4] and tholeiites from Gorgona Island [38], but similar to the basalts analyzed by Pegram and Allegre [40]. Some of the lowest Re concentrations are probably due to dilution of the melt with phenocrysts (e.g. MGA-B-47). The systematically lower Re contents in the basalts studied here may reflect either differences in the generation of tholeiites and alkali basalts, or differences in source composition. However, if Re is incompatible during partial melting of peridotite, Re concentrations in alkali basalts should be higher than in tholeiites. The higher Re concentrations of tholeiites might indicate the involvement of a mantle phase which remains in the residue at small degrees of melting, yet might be exhausted during the relatively higher degrees of melting involved in tholeiite generation. With the exception of sample RARB-9 from Rarotonga, the Re concentrations in Table 2 show little variation (100-586 ppt, average 282 ppt). Compared to Morgan's estimate for the Re concentration of primitive mantle (250 ppt, [19]), the average Re concentrations of the basalts in this study suggest an apparent partition coefficient (peridotite/melt) of 0.5-1.0 during the generation of alkali basalts. This is consistent with the negative trend in Fig. 1C, which indicates that Re was more compatible than Nd during melting and crystal fractionation/accumulation.

5.2 1870s//1860s systematics of EMIl islands Despite their extreme 87Sr/S6Sr and elevated 2°7pb/2°4pb, the Samoa and Tahaa samples are not distinctive in t h e i r 1 8 7 0 s / 1 8 6 0 s ratios, falling within the range of lSVOs/lSgos for abyssal peridotites [4,42], "normal" ophiolite peridotites [29] and MORB-like peridotites [31,33; Hauri, unpublished data]. The high SVSr/S6Sr and elevated 2°7pb/Z°4pb of the Samoa and Tahaa samples may have their origins in a source containing a component similar to modern sediments, as advocated previously for Samoa [10]. If sediment is indeed involved in the origin of the Samoan mantle source, the amount of sediment must be small, not more than a few percent, in order to fit the Nd and Hf isotope data for Samoa [43]. Mixing of

360

E.H. HAURI and S.R. HART

such a small amount of sediment into the mantle would have only a small effect on the R e / O s and lSTOs/~S6Os ratios of the mixture, due to the low Re and Os concentrations of continental crustal material [3]. Mixing of a depleted mantle (878r/ 86Sr = 0.703, 143Nd/144Nd = 0.51290, MORB-like Pb, 1870S/1860S = 1.02-1.05), with 1-2% modern sediment, could reproduce the Samoan data. The relative concentrations of Pb, Sr, Nd and Os in most clastic sediments and peridotite (ratio of ppm s e d i m e n t / p p m peridotite) decrease in the order P b - S r - N d - O s . Thus, in a sediment/peridotite mixture with a small fraction of sediment, the Pb isotopes would be closest to the sediment component, followed by SVSr/86Sr, then 143Nd/ 144Nd and 176Hf/C177Hf, and finally 1870S/1860S would be only slightly changed from the peridotite value, if at all. This is essentially the order of decreasing deviation of the EMII isotope ratios from depleted mantle values. Some type of metasomatism, which has not strongly effected the Re-Os system, could also be invoked to explain the Samoa and Tahaa data. 5.3

1870S//1860Ssystematics

of HIMU islands

The 187Os/~86Os ratios of the Rurutu, Tubuai and Mangaia basalts are distinctly higher than measurements for mantle peridotites and other oceanic basalts. Since both Os and Pb are chalcophile elements, it is possible that the high lSVOs/l~6Os, high 2°6pb/2°4pb signature is the result of removal of Os and Pb relative to Re and U by the separation of a sulfide phase from some portion of the earth's mantle [44,45]. Strong arguments against this hypothesis of continuous extraction of siderophile and chalcophile elements from the mantle to the core through geological time have been presented by Newsom et al. [46]. However, several other possibilites remain to account for the HIMU Os isotope signature. 5.3.1 Crustal contamination In attempting to identify the source of the radiogenic ~SVOs/18('Os in the H I M U basalts, crustal contamination of the magmas must be evaluated as a possibility. The islands of Rurutu, Tubuai and Mangaia rest on oceanic lithosphere which is 60-80 Ma old. Given the range of R e / O s ratios measured in oceanic basalts, it is quite

possible that the crustal part of the lithosphere could evolve to 187Os/186Os in excess of 1.25 over this time period, and assimilation of large amounts of this material into a magma with low Os concentration could raise the 1870s/186Os in the erupted magmas. However, if the high ~SVOs/ 186Os ratios are characteristic of the total volcanic volumes of these islands, to account for this signature by assimilation would require several times the amount of MORB which lies below these islands. In addition, the amount of crust necessary to account for the elevated 1870S/1860S ratios in the H I M U basalts would shift the Pb isotopic composition of the magmas toward MORB values, since U / P b and T h / P b ratios measured in fresh MORB [34,47,48] are not high enough to generate the H I M U Pb signature in the required time. Altered MORB crust has higher U / P b ratios, but assimilation of 60-80 Ma old MORB crust cannot account for the H I M U 2°7pb/Z°4pb signature, irrespective of the U / P b ratio. Strong contamination by altered crust would also result in higher 87Sr/S6Sr values characteristic of seawater alteration. Variable assimilation of MORB crust into variably Os-depleted basalts would also destroy the relative Os isotopic homogeneity observed at Tubuai and Mangaia. In addition, if this process were important, it would be expected to have an even greater influence on the ~SVOs/~6Os ratios of the lavas from Savaii, which rests on seafloor with an age in excess of 100 Ma. This is in conflict with the l o w 1 8 7 0 s / 1 8 6 0 s ratios thus far observed in the Samoa data. Thus the high 187Os/i86Os ratios are indicative of the Os isotopic composition of the mantle sources of the Rurutu, Tubuai and Mangaia basalts. 5.3.2 Mantle metasomatism In order to evaluate mantle metasomatism as a process for raising the JSVOs/186Os of the H I M U basalts, it is instructive to compare the HIMU Os data with 187Os/18~Os measurements of metasomatized peridotites. The measurements of Walker et al. [31] on xenoliths from the Kaapvaal craton and Pearson et .al. [32] on South African and Siberian xenoliths from kimberlites all have 1870s/1860s less than 1.08. These xenoliths include some extensively metasomatized and LREE-enriched samples, some with incredibly high S7Sr/S6Sr and low 143Nd/144Nd, yet the

Re-Os ISOTOPESYSTEMATICSFROMTHE SOUTHPACIFIC

metasomatism apparently did not result in increased R e / O s and 187Os/186Os ratios (notwithstanding Re contamination from the kimberlite host [31]). The Ronda plagioclase facies peridotites analyzed by Reisberg et al. [33] are also L R E E enriched, with low 143Nd/144Nd and very radiogenic 2°6pb/Z°4pb [49]. These peridotites had the lowest R e / O s and 187Os/IS6Os of all the Ronda samples, again suggesting that the R e / O s system was resistant to the metasomatism which resulted in enriched signatures in other isotope systems [33]. Although a Re-Os characterization of mantle metasomatism is far from complete, it appears that the metasomatic enrichment processes reflected in these peridotite samples are incapable of generating high lSTOs/lS6Os signatures. Given that the measured 187Os/lS6Os ratios of the H I M U basalts are higher than any peridotite values, we can tentatively rule out metasomatism as a mechanism for producing the high 1870S/1860S in these basalts. Layers of mafic bulk compositions found in the Ronda ultramafic complex, with 1870S/1860S ratios of 1.71 to 47.9 [33], may provide a high lSTOs/~S6Os source in the mantle. However, since nearly all of the Pb in a melting mafic layer-peridotite mixture will be derived from the layers, such layers must also have high 2°6pb/2°4pb. In general, Ronda mafic layers do not have such high 2°6pb/Z°4pb [Hauri, unpublished data]. In addition, the elevated 1870S/1860S ratios of the Ronda mafic layers were probably created while the massif was isolated in the subcontinental lithosphere over the last 1.3 Ga [33,50,51]. These layers, ranging in thickness from several centimeters to several meters, would probably not survive the mixing processes operating in the convecting mantle, where the H I M U source must reside.

5.3.3 Crustal recycling It is clear that the Rurutu, Tubuai and Mangaia lSVOs/186Os systematics require a source with a high time-integrated R e / O s ratio. The only geological materials thus far recognized to have high R e / O s are magmas, insofar as oceanic and continental crust are the products of melting. Ancient, subducted oceanic crust may provide a suitable high R e / O s , high 1870S/1860S reservoir in the convecting mantle. The oceanic crust, with its several kilometers of thickness, would be more

361

resistant to diffusive and convective mixing than meter-scale veins or layers. It is unlikely that the Os in the H I M U magmas was derived entirely from ancient oceanic crust, as this material would probably have much higher 1870S/1860S ratios than observed. Only a moderately high mantle R e / O s ratio (about 10-20) is needed to raise the lS7Os/IS6Os from depleted mantle values (about 1.05) to the H I M U 1870s/1860s values (about 1.25) in a few billion years. This suggests that the time-integrated R e / O s ratio of the H I M U source is the result of mixing high R e / O s crust with low R e / O s peridotite. We can quantitatively model this mixing if we have constraints on the timing of the recycling process. If the H I M U source is a mixture of recycled oceanic crust and mantle peridotite, the U-Th-Pb evolution of the mixture is totally dominated by the recycled crust, even at low fractions of crust, due to the very low concentrations of these elements in peridotites [52,53; Hauri unpublished data]. As a result, the Pb isotopic composition of the H I M U source is essentially that of the crustal component, and constraints on the age, U / P b and T h / U ratios emerge from inversion of the Pb-Pb systematics. The working assumptions are as follows: (1) The depleted M O R B mantle (DMM) is assumed to be a second stage reservoir, with a maximum age of 3.0 Ga, derived from the bulk silicate earth (present day bulk silicate earth values of 87Sr/86Sr --- 0.7047, 143Nd/144Nd = 0.512638, 176Hf/177Hf = 0.282880, 1870S/1860S = 1.07). (2) The M O R B crust is assumed to be a subsequent stage derived from the DMM at any time < 3.0 Ga ago, with initial isotope ratios corresponding to either DMM or the bulk silicate earth (BSE) (present day DMM taken from most depleted M O R B points for 87Sr/S6Sr (0.70220), 143Nd/144Nd(0.51330) and 176Hf/lVVHf(0.28355), and 1870s/1860s = 1.00). (3) The evolution of the M O R B crust from the time of its generation to the present day isotopic composition of H I M U (87Sr/S6Sr = 0.7028, 143Nd/la4Nd = 0.512850, 176Hf/177Hf = 0.282900, 1870S/1860S ~ 1.25) is approximated by a single evolution curve from either the DMM or BSE initial isotope ratios. For the Pb-Pb system, these assumptions imply

E.H.HAURIand S.R.HART

362

that the intersection of any third stage isochron (representing HIMU) with any second stage isochron (representing DMM) must lie within the present day M O R B field. These assumptions result in the boundary conditions displayed in Fig. 3. N-MORB data are MORB analyses with 878r/ 86Sr < 0.70300 (data sources summarized in [54]) and Mangaia data are from [11,36] and this study. Given the above assumptions, the bounding isochrons shown in Fig. 3A are absolute limits on the age of the crustal component in the H I M U source, and indicate an age of 770-2100 Ma. Calculated third stage 23Su/2°4pb ratios range from 20 to 40. Using these ages, the T h / U ratios calculated from the bounds in Fig. 3B range from 1.42 to 3.56.

15.9

4.55Ga / Geochron~ ~

158-

7

7

0

2100~, ~ /

A Ma

~,5~. ~o ,~G_~o~ ~ 156"

155"

q

, ~,

~54"

1531-6

"

17

"

lB

I

I

Ig

I

20

"

21

I

22

"

25

204pb

206pb /

42¸ 41

J

40

,~ 36

/ / ~

36 3515

N-MORB

Th/U=3.54 16

17

18

19

206pb /

20

21

22

23

204pb

Fig. 3. Pb isotope ratio plots with data for N-MORB (defined as MORB with S7Sr/S6Sr < 0.70300, from [56,57], and data sources summarized in [54]) and Mangaia [11,36; and this study]. (A) The bounding isochrons delineate the upper and lower limits on the age of the recycled oceanic crust in the HIMU source (see text). (B) Bounding lines showing the limiting T h / U ratios of the recycled oceanic crust in the HIMU source.

Working from the above assumptions and age constraints of 770-2100 Ma, calculated limits on the tSTRe/tS6Os of the H I M U source are 8.9022.80 (Fig. 4). These ratios are low compared with measured basalt lS7Re/~S6Os ratios (Table 2 and [4,38]). Assuming a peridotite Os concentration of 3000 ppt and crustal Re concentrations of 500-2000 ppt, the calculated lSVRe/tS6Os ratios constrain the amount of crust in the H I M U source to be 16-75%. The lower age limit (770 Ma) requires the highest 1 8 7 R e / 1 8 6 O s and largest amount of crust, whereas the upper age limit (2100 Ma) requires the l o w e r 1 8 7 R e / 1 8 6 O s and a smaller amount of crust in the H I M U source. Even at 16% crust, the Sr, Nd and H f evolution of the H I M U source is determined largely by the crustal component, and, in a similar fashion, we can place limits on the range of possible R b / S r , S m / N d and L u / H f ratios in the crustal component (Fig. 4). Table 3 summarizes the results of the inversion along with the ranges for these trace element ratios measured by isotope dilution for fresh MORB glasses [34,47,48,55-65]; a comparison is made with estimates for average N-MORB [66] and altered MORB [67]. The limits on these ratios calculated from the inversion are notable for their similarity to the range of values measured in fresh mid-ocean ridge basalts. If the mantle sources of Rurutu, Tubuai and Mangaia contain subducted oceanic crust, this is a surprising result, considering the large additions of U, Rb and seawater Sr to the upper crust during alteration [68]. In detail, the U / P b and T h / P b ratios are higher than the averages of the MORB results. The deviation of the U / P b ratio is in the direction of altered M O R B crust [67] (Table 3), but T h / P b and T h / U are higher than expected for altered crust. The similarity of the measured and calculated T h / U ratios suggests that the high U / P b ratio of H I M U may not be due to the addition of U, but to the depletion of Pb relative to U and Th, although some U addition is allowable if the recycled crust has a high T h / U ratio. Chauvel et al. [69] have proposed that Pb is leached from M O R B crust during hydrothermal alteration and deposited in metalliferous sediments. Given that the MORB crust is obviously affected by low temperature alteration, the low 878r/S6Sr of HIMU and the similarity of the

Re-Os ISOTOPESYSTEMATICSFROMTHE SOUTHPACIFIC

363

model R b / S r and T h / U ratios to flesh M O R B may be used to suggest that the alteration signature may be partially removed during passage through the subduction zone. This is a plausible scenario, since the U, Rb, seawater Sr, and leached M O R B Pb are all resident in low temperature alteration phases in the oceanic lithosphere [69]. Alternatively, the altered portion of the M O R B crust which has been added to the lithosphere may comprise only a small part of the total crustal budget for these elements. The constraints provided by the inversion results are consistent with all of these possibilities, and allow for some fraction of the U and Rb contributed by alteration to remain in the crust after subduction.

The Re-Os isotope systematics allow for a peridotite/basaltic crust mixture with up to 75% crust in the H I M U source. Such a source would have a high F e / M g ratio, and would presumably generate high F e / M g melts• In this context, it should be noted that basalts from Mangaia, Tubuai and Rurutu have higher total FeO for a given MgO than all other OIB. This feature was also pointed out for Tubuai by Chauvel et al. [69]. Alternatively, seawater alteration may affect the Re-Os systematics of the oceanic crust. The residence time of Re in seawater has been estimated at 750,000 yrs [70], and like U, Re concentrations in seawater covary with salinity [71]. Compared with seawater (about 44 pmol/kg), Re concentra-

030/ \~.o

i

2000ppt Re (16%-41%)

t,7\

i I •

i

o,o°2° "

i

"

i

o.o o

"

Age of Recycled Crust (Ga)

i

"

½l

-,BSE .--

4

Age of Recycled Crust (Ga)

o.o4-I \

.

BSE

0.03

~ . 2 o t /

~

1.150 DMM

II / D M M

] O.Ol]

~

~

oo

.0016V

o

~

~

Age of Recycled Crust (Ga) Age of Recycled Crust (Ga) Fig. 4. Results of the inversion of the HIMU isotope data for parent/daughter ratios in the recycled crustal component. The curves represent the parent/daughter ratios calculated for recycled crust of various ages, assuming that the crust had initial isotopic ratios corresponding to either depleted MORB mantle (DMM) or bulk silicate earth (BSE). The vertical lines are limits on the age of the recycled crust from the Pb-Pb system (see text, Table 3 and Fig. 3). Percentages in the Re/Os figure refer to the amount of recycled crust in the HIMU source for a given Re concentration.

364

E.H. H A U R I a n d S.R. H A R T

tions in hydrothermal fluids are low ( < 0 . 1 p m o l / k g ) [70], suggesting that any Re which may be added to the crust is not removed by high temperature alteration. In addition, seawater has 187Os/lS6Os around 8.5 [72], the influence of which has been observed in Os isotope analyses of M O R B whole rocks [73]. Thus, seawater alteration would serve to increase both the Re concentration and 1870S/1860S of the M O R B crust. The result would be to decrease the amount of crust needed in the H I M U source, and reduce the curvature of the mixing hyperbolae in Fig. 5. These calculations serve to demonstrate the suitability of subducted oceanic crust as a component in the generation of the H I M U isotopic signature.

5.3.4 Constrains on melt percolation The high 1870s/1860s of the Tubuai and Mangaia basalts also serves as a tracer for evaluating the interaction of these magmas with the depleted upper mantle, which has 1870S/1860S in the range 1.00-1.08. Due to the high Os concentration of mantle peridotite relative to basalts, the 187Os/IS6Os ratio of a given m a g m a should be sensitive to assimilation of peridotite with a different lSYOs/186Os ratio. For example, assimilation of 10% of a peridotite (3000 ppt Os) with ~SVOs/~S6Os of 1.05 into a magma (300 ppt Os) with 1870S/1860S of 1.25 will lower the 187Os/ ~86Os of the m a g m a to 1.15. Magma-wallrock reaction [74] or melt percolation [75,76] through

2000 ppt Re 500 pt Re 2100 Ma 77~Ma (29) (3.58) 130

3~ OT

07

Ds

O

120 ¸

F P

-a ~0 1 I0 AP

• D

I ,OC - 17

18

19

20

21

22

2°6pb / 2°4pb Fig. 5 . 1870S/1860S versus 2°6pb/2°4pb for results from this study. The two mixing curves delineate the locus of mixtures of peridotite (Os = 3000 ppt, 1 8 7 0 S / 1 8 6 0 S = 1 . 0 5 , Pb = 10 ppb, e°6pb/2°4pb = 18.50, Pb = 10 ppb) with two possible oceanic crustal compositions (2000 ppt Re and 500 ppt Re, both with common Os = 100 ppt, 2°6pb/2°4pb= 21.80, P b = 100 ppb). The ages represent the time at which the crust was formed, and the numbers in parentheses are the putative present day tS7Os/lS6Os ratios of the crust as if it evolved as a closed system.

peridotite with low 187Os/tS6Os will have a similar effect on the 187Os/186Os of the magma. In the case of percolation of melt through peridotite, it is well known that each element moves at some fraction of the melt velocity which is inversely proportional to the element's bulk solid/melt partition coefficient [75,76]. The bulk p e r i d o t i t e / m e l t partition coefficient for Os is much higher than for Sr, Nd, H f and Pb. As a

TABLE 3 Calculated p a r e n t / d a u g h t e r ratio results in subducted oceanic crust compared with isotope dilution measurements in fresh MORBs, average N - M O R B of H o f m a n n [66], and an altered M O R B composite [67]

87Rb/86Sr 147Sm / 144Nd

176Lu/177Hf 238U/204pb 232Th/204pb Th/U

Calculated Subducted Crust (HIMU Source)

Measured Fresh M O R B (data averages)

Avg. N-MORB Hofmann (1988)

Altered MORB Hart&Staudigel (1989)

0-.095 .150-.248 .002 - .036 19.7 - 40 25 - 142 1.42 - 3.54

.006-.26 (.03) .15-.23 (.20) .023 -.040 (.027) 4.8 - 38 (12.3) 7.7 - 85 (38.7) 1.8 - 4.2 (2.80)

.032 .203 .0281 9.07 23.9 2.63

.223 .200 30 7.0 .23

Averages of the M O R B data are given in parentheses. The calculated p a r e n t / d a u g h t e r ratios are those needed in order for subducted oceanic crust to evolve from either depleted M O R B mantle or bulk silicate earth initial isotope ratios to the composition of the H I M U e n d m e m b e r in 770-2100 Ma (see text). M O R B data sources: R b / S r and S m / N d [55-57], L u / H f [58-60], U-Pb [34,47,48], T h / U [34,47,48,62-65], Y h / P b [61].

Re-Os ISOTOPE SYSTEMATICS

FROM THE SOUTH PACIFIC

result, for any reasonable melt velocity, these elements will move much faster than Os, and the fronts of these elements will be almost entirely separated from that of Os after only a few hundred meters of percolation. The effect on the isotopic composition of the melt is such that the St, Nd, Hf and Pb isotopic compositions will reflect the source of the melt, whereas the Os isotopic composition will reflect that of the matrix through which the melt is moving. The only way to prevent this fractionation of the compatible elements from the incompatible elements is to move the magma through the matrix fast enough to prevent significant exchange of compatible elements with the matrix peridotite. This essentially requires the existence of a channel or fracture network which can transport the magma from the source region within the plume to the surface, where it is ultimately erupted. The high 1870s/186Os ratios for the HIMU basalts, and the consistent covariation of the isotope ratios of the two elements with the largest difference in peridotite/melt partition coefficients (Os and Pb), requires that interaction with the depleted mantle had a negligible effect on the isotopic composition of the basalts erupted at Rurutu, Tubuai and Mangaia. If melt segregation mechanisms are similar beneath other oceanic islands, this would suggest that chromatographic trace element fractionation processes [e.g. 77] do not have a significant influence on the compositions of magmas erupted at hotspots.

5.4 Global isotope systematics With the presently available Re-Os data, it is now possible to delineate much of the global Sr-Nd-Pb-Os isotopic variability in the oceanic mantle (Fig. 2). The composition of the depleted mantle is estimated from 1870S/186Os measurements of abyssal peridotites [4,42] and other peridotites with MORB-like Sr, Nd and Pb isotopes [31,33]. Estimates of the positions of the various end members proposed by Zindler and Hart [8] are shown in Figs. 2A-C. Notwithstanding the lack of MORB 1870S/1860S data, it is likely that the MORB source is characterized by 187Os/186Os of ~< 1.08, and DMM may have 1870S/1860S as low as the lowest measured lSVOs/lS6Os for abyssal peridotites of 1.0032 [4]. The position of

365 EMI is unknown due to the lack of Re-Os data for this end member, although Sr, Nd and Pb isotope affinities with the subcontinental mantle may suggest a low 1870s/186Os signature for EMI ( < 1.00?). The 1870s/1860s data for Samoa and Tahaa suggests that the EMII end member is characterized by 1870s/1860s ratios which lie between estimates of depleted mantle (about 1.00) and bulk silicate earth (about 1.08), consistent with-the EMII 143Nd/144Nd and 176Hf/lV7Hfsignatures [9,10,58]. The homogeneity of the Mangaia 1870s/1860s data suggests that a value of about 1.25 is representative of the Os isotopic composition of HIMU. Martin [4] has proposed that a mantle source exists with high 3He/4He and 1870S/1860S of around 1.10, based on Re-Os analyses of basalts from Hawaii. The only island chain in the present study with documented high 3He/4He is Samoa [78], with 1870S/1860S of 1.0261--1.0739. These results suggest that the proposed high 3He/4He end member, recently recognized in the Sr-Nd-Pb isotope data and termed F O Z O by Hart et al. [79], may be characterized by a range of 1870s/186Os ratios. The possible identification of the HIMU isotopic signature (high 1870S/1860S, high 2°6pb/ 2°4pb, low 878r/86Sr) with recycled oceanic crust has important implications for mixing relationships in the earth's mantle. The earth presently has 37,000 km of subduction zones, and if we extrapolate the present subduction velocity of 80 k m / M a [80] back through time, this suggests that the total volume of oceanic crust subducted in 3 Ga is roughly 7% of the total volume of the mantle (or 10% since 4.55 Ga). This contrasts with the relative rarity of the extreme HIMU signature, which has only been observed in the Cook-Austral Islands, St. Helena [81] and two seamounts in the western Pacific Ocean [82]. This suggests that the extremely high Pb isotopic compositions may represent only the most well preserved (i.e. least well mixed), and possibly the oldest, subducted crust. The slightly higher 2°7pb/Z°4pb signature of St. Helena, relative to the Cook-Austral Islands trend, also suggests that St. Helena represents a source that is separate from that for Tubuai and Mangaia, with a slightly different evolution. Realistically, there is likely to be a continuum of Pb isotopic compositions between DMM and HIMU which would

366

correspond to either younger, or more well mixed, recycled oceanic crust. Depending on the individual histories, there could also be a variety of lSVOs/lS6Os isotopic compositions in these parts of the mantle. There may also be a recycled oceanic crustal signature in the data for N-MORB. From Fig. 3, it is obvious that there are many MORBs with 878r/S6Sr less than 0.70300 that also have radiogenic Pb isotopic compositions [57,83]. If the MORB mantle is the depleted mantle complement of the continental crust, it should be very depleted in highly incompatible elements such as U, Th and Pb. As a result, it will be more susceptible to having its Pb isotopic signature dominated by a recycled component. In this context, "less depleted" M O R B mantle might lie closer to the geochron, whereas "more depleted" M O R B mantle, prone to contamination by a recycled oceanic crustal component, would tend to lie farther to the right of the geochron. If in fact a large fraction of the mantle exists with Pb isotopic compositions near the geochron, but susceptible to contamination with recycled oceanic crust due to very low Pb concentrations, this may provide an explanation for the enigmatic "Pb paradox", especially in MORBs. More lSYOs/ lS6Os data are needed to test these ideas, on both MORBs and OIBs with a variety of Pb isotopic compositions. 6. Conclusions The Re-Os isotope systematics in a group of oceanic island basalts from the south Pacific Ocean have been examined in order to constrain the role of crustal recycling in the mantle. The highly variable Os concentrations in the lavas are controlled by a low R e / O s phase (or phases) in which Os is compatible, possibly olivine, chromite a n d / o r sulfides. These Os concentration variations determine the R e / O s ratios of the erupted basalts. The 187Os/186Os ratios of EMIl basalts from Samoa and Tahaa range from 1.0261 to 1.0861, while lSTOs/186Os in H I M U basalts from Rurutu, Tubuai and Mangaia varies from 1.1169 to 1.2483. Sediment involvement in the origin of the EMII mantle component is not ruled out by the 187Os/ lS6Os data for the Samoa and Tahaa lavas. How-

E.H. HAURI and S.R. HART

ever, the Os isotope systematics of peridotitecrust mixtures indicates that the 1 8 7 0 S / 1 8 6 0 S r a t i o of these mixtures are only very weakly influenced by the small amounts of sediment required to account for the Sr and Pb isotopic signatures. The high lS7Os/lS6Os ratios in basalts from Rurutu, Tubuai and Mangaia provide strong evidence for the role of ancient subducted oceanic crust in the H I M U mantle. The H I M U Pb isotope systematics constrain the age of the recycled crust to 770-2100 Ma, with a 23Su/Z°4pb of 20--40 and a T h / U ratio of 1.42-3.65. Calculated R b / S r , S m / N d , L u / H f and T h / U ratios in the subducted crust are similar to those for fresh MORB, suggesting that the p a r e n t / d a u g h t e r fractionations which occur during M O R B genesis are sufficient to generate the Sr-Nd-Hf-z°8pb/2°6pb signature of HIMU. The high calculated U / P b and T h / P b ratios indicate a depletion of Pb in the recycled crust. The possible identification of the HIMU isotopic signature (high 1870S/1860S, high 2°6pb/2°4pb, low 87Sr/86Sr)with recycled oceanic crust suggests that other oceanic mantle sources, including that of some mid-ocean ridge basalts, may contain a component of subducted basaltic crust. Acknowledgements Thanks go to Greg Ravizza for his tutelage in the finer arts of fire assay and Os distillation, and for duplicate Re and Os analyses during the research and develepment N-TIMS. We also thank Alan Zindler and Laurie Reisberg for many discussions about the oxygen leak during the evolution of the N-TIMS technique. Many thanks to Hans Barsczus for help in planning the 1990 sampling tour of the South Pacific, and Mr. and Mrs. Henri Mallarde for their hospitality on Tubuai. Ken Burrhus was instrumental in allowing us to keep up the rapid pace of N-TIMS development. We also appreciate the assistance of Mark Kurz for access to the VG354, Jon Snow for software development, Dave Kammer for VG354 assistance, and Jurek Blusztajn for cleanlaboratory support. The reviews of Laurie Reisberg, Rick Carlson and Keith O'Nions improved the manuscript significantly. This research was supported by NSF grant EAR8708372 to S.R. Hart.

Re-Os ISOTOPE

SYSTEMATICS

367

FROM THE SOUTH PACIFIC

Appendix

187Os/186Os of 6.60. The Re and Os data in Table 3 are corrected for these analytical blanks.

Analysis of Re and Os isotopes by oxygen-enhanced N-TIMS Re and Os separation

Mass spectrometry

For analysis of Re, 1-5 g of rock powder were weighed out, spiked with 185Re, and dissolved in a 10:1 mixture of H F and H N O 3 in a 60 ml closed Savillex P F A Teflon screw cap vessel. Re was separated on a column of 1 ml of A G 1 X 8 (200-400 mesh) in 0.2 N HC1 and 0.2 N H N O 3, and eluted with 8 N H N O 3. This procedure was then repeated with a smaller column (20 p.1 resin). For Os analysis, 30-100 g of rock powder were weighed into a Coors glazed ceramic crucible, spiked with 19°Os, and Os was preconcentrated from the rock matrix by nickel sulfide fire assay [84]. The flux consisted of a mixture of 20 g borax, 10 g Na2CO 3, 4 g Ni powder, and 2 g sulfur. The sample was fused at 1000°C for 1.5 to 3 h (blanks were fused for 1 h). During the fusion, the platinum-group elements are concentrated in immiscible NiS beads, which segregate at the bottom of the crucible [84]. After cooling, the NiS beads were extracted by crushing the crucible and its contents and picking out the beads. The NiS beads were dissolved in 6.2 N HC1, and the solution was filtered through a 0.45 tzm Millipore cellulose filter. The Os was concentrated in the residue left on the filter paper. The distillation and collection of Os as the volatile OsO 4 (b.p. 100°C) is similar to the procedure described by Luck [85]. The sample Os is distilled from a solution of CrO 3 (VI) in 4 N H 2 S O 4 and collected in concentrated HBr chilled in an ice water bath. The H B r solution was then transferred to a 15 ml Savillex P F A Teflon screw cap beaker, capped tightly, and heated in an oven at 100°C overnight to allow the reduction of OsO 4 by HBr to go to completion. The beaker was then opened and evaporated on a hotplate at 75°C to a volume of less than 1 /xl. The sample Os was purified by ion exchange using single Chelex 20 resin beads [33,85]. For the overall separation and purification, the Re yield is about 80-90%, while the Os yield varied from 50 to 90%. Measured blanks averaged 35 pg for Re (including filament blank) and 40 pg Os for the full procedures. The analytical blank had a measured

The negatively charged oxides of Re ( R e O 4 ) and Os (OsO 3) are produced by the oxidation of the sample Re and Os by barium oxide on the surface of a Pt filament [86,87]. Filaments for mass spectrometry consist of 0.025 × 0.508 m m zone refined 99.95% Pt ribbon (H. Cross). This ribbon contains small amounts of Re, presumably as contamination from the manufacturing of Re products. This is a potential problem due to the possible isobaric interference of 187Re1603 on 187Os1603 during mass spectrometry; however, the isobaric interference of 187Re16Of on 187Os1603 was not observed during our measurements (within the measured precision). The sample Re is dissolved in 1 Ixl of 0.1 N H N O 3 and is mixed with 1-3 /xg of Ba(NO3) 2 and then loaded onto the filament and dried at 0.5 amps. R e p e a t e d analyses of Re spike indicate an average filament Re blank of 0.38 pg for two different batches of Pt filament, with a range from 0.2 to 0.52 pg. After purification, the Os in < 1 ml HBr is loaded onto the center of the filament and dried down gently at 0.5 amps. The Os is then reduced to the metal on the filament by heating to 600°C in a vacuum of < 10 -6 torr for 3-12 h. Then, 1-3 mg of Ba(NO3) 2 is loaded directly on top of the Os and dried at 0.5 amps. At this stage, the Os sample is ready for mass spectrometry. Re and Os isotope ratio measurements were made on the N I M A - B 9 inch, 60 ° sector thermal ionization mass spectrometer at the Woods Hole Oceanographic Institution. For Re analysis the filament was immediately raised to about 700°C, which is just below running temperature. As soon as R e O f peaks were observed, the filament temperature was raised to 750-800°C, as measured by optical pyrometry. Ion currents for several hundred picograms of Re were of the order of 10 -13 to 10 -12 A, and 251/249 was usually measured to better than 0.1%. The sample Re, loaded from nitric acid, is already in its most oxidized form (Re207), and it is likely that the formation of ReO~- occurs at different temperatures and with different efficiencies for the sample Re

368

E.H. HAURI and S.R. HART

T A B L E A1 Isotope data for osmium standard solution, Re standard solution, and IS7Re spike measured by N-TIMS

187Re/185Re Sample S i z e (Os)

187Os/186Os

100 ng 1 ng 1 ng 100 pg 100 pg 100 pg I0 pg I pg 1 pg 800 f[~ Average

1.4476 + 0.0006 1.4452 + 0.0008 1.4470 + 0.0005 1.4476 + 0.0011 1.4467 + 0.0017 1.4444 + 0.0011 1.4456 + 0.0089 1.4448 + 0.0062 1.4448 + 0.0099 1.450 + 0.015 1.4460 _+0.0025

189Os/188Os 1.2198 + 0.0002 1.2198 + 0.0002 1.2196 + 0.0002 1.2200 + 0.0005 1.2194 + 0.0007 1.2195 + 0.0014 1.2194 + 0.0040 1.2190 + 0.0033 1.2206 + 0.0026 1.223 + 0.013 1.2197 _+0.0009

190Os/188Os 1.9824 + 0.0004 1.9822 + 0.0003 1.9848 + 0.0002 1.9838 + 0.0006 1.9837 + 0.0016 1.9841 + 0.0013 1.9847 + 0.0036 1.9838 + 0.0037 1.9866 + 0.0031 1.9890 + 0.0170 1.9840 _+0.0026

187Re/18$Re

2OO pg

250 pg

(standard)

(spike)

1.6742 + 1.6741 + 1.6730 + 1.6748 + 1.6753 + 1.6735 + 1.6725 + 1.6753 + 1.6744 + 1.6737 + 1.6741 +

0.0009 0.0007 0.0009 0.0003 0.0002 0.0003 0.0002 0.0003 0.0002 0.0011 0.0019

.06015 + .05954 + .06019 + .05995 + .06006 + .05964 + .05922 + .05964 + .05987 + .05955 + .05978 +

0.00007 0.00004 0.00006 0.00002 0.00003 0.00008 0.00003 0.00002 0.00003 0.00(~ 0.00063

Errors on the average values are 2o-, and within-run errors are reported as 2¢r of the mean. Isotopic ratio data are corrected for instrumental mass fractionation and oxide contribution

( R e 2 0 7) and the filament Re (Re metal). This procedure for measuring Re is that which minimizes the contribution of Re from the filament, as determined by repeated measurements of Re spike. For Os analysis, an oxygen pressure of 6 - 8 × 10 7 torr is established in the source by leaking filtered oxygen through an inlet valve. The filament is gradually brought to a running temperature of 600-800°C over 30 min. Ion currents were typically 2-10x10-13 A at mass 240 for nanogram-sized basalt Os loads. The data were corrected for mass-dependent isotope fractionation using an exponential law, and are normalized to 2 4 0 / 2 3 6 = 3 . 0 9 2 1 9 calculated from 1920S/ lSaos = 3.08271 [27] and 170/160 = 0.0003708 and 180/160 = 0.002045 [88]. Our data reduction accounts for the contribution of the spike to 192Os/lSSOs, tg°os/188Os and lS7Os/lSSOs, and the variable oxide contributions on 240/236 resulting from variable amounts of 19°Os (spike) and radiogenic lSVOs. The data are reported as 1870s/1860s, which is calculated from the measured lS7Os/ISSOs and 1S6Os/1SSOs=0.12035 [27]. In-run precisions were always better than 0.1% on lSVOs/lSaos and 19°Os/18SOs. No corrections f o r 1 8 7 R e 1 6 0 3 o n 1 8 7 0 s 1 6 0 3 w e r e n e c e s s a r y for these measurements. The presence of oxygen gas has two effects. The first is to enhance the oxide speciation of Re

and Os to R e O 4 and OsO 3 by reducing the level of the subordinate oxides to background levels. The second is to improve the overall ionization efficiency for R e O 4 and OsO3, resulting in higher signal intensities and higher ion yields. Under optimum conditions of temperature and oxygen pressure, the potentially serious interference of 1 8 7 R e 1 6 0 3 o n 1 8 7 0 s 1 6 0 3 - is eliminated within the resolution of our measurements. However, this interference can become substantial at higher temperatures a n d / o r non-optimal oxygen pressures. Table A1 shows the results of Os standard runs during the course of this work. The measured 187Os/1S6Os ratios for the standard runs are reproducible to better than 0.15% over five orders of magnitude in sample size (excluding the 800 fg run). Os ion yields ranged from < 4% to 15%. Table A1 also shows several measurements of 250 pg of Re spike and 250 pg of Re standard solution, as a check on the reproducibility of the Re measurements. The Re isotopic compositions are reproducible to 0.1% for standard Re and 1% for spike Re (due to Re from the filament). The reproducibilities of the Re and Os basalt analyses in Table 3 are calculated from replicate analyses of separate powder aliquots subject to separate dissolutions or fusions. The average reproducibilities of Re and Os concentrations are 2.6% and 4.6% respectively. The 1870S/1880S ra-

Re-Os ISOTOPE SYSTEMATICSFROM THE SOUTH PACIFIC

369

tios are reproducible to 0.3% after correction for the analytical blank and spike contributions. This measure

of the reproducibility incorporates

all

aspects of the analytical technique.

References 1 J.M. Luck, J.L. Birck and C.J. Allegre, lS7Re-187Os systematics in meteorites: early chronology of the solar system and the age of the galaxy, Nature 283, 256-259, 1980. 2 J.W. Morgan, G.A. Wandless, R.K. Petrie and A.J. Irving, Composition of the Earth's upper mantle, 1: Siderophile trace elements in ultramafic nodules, Tectonophysics 75, 47-67, 1981. 3 B.K. Esser, Osmium isotope geochemistry of terrigenous and marine sediments, Ph.D. Thesis, Yale Univ., 1990. 4 C.E. Martin, Os isotopic characteristics of mantle derived rocks, Geochim. Cosmochim. Acta. 55, 1421-1434, 1991. 5 A.W. Hofmann and W.M. White, The role of subducted oceanic crust in mantle evolution, Carnegie Inst. Wash. Yearb. 79, 477- 483, 1980. 6 A.W. Hofmann and W.M. White, Mantle plumes from ancient oceanic crust, Earth Planet. Sci. Lett. 57, 421-436, 1982. 7 C.G. Chase, Oceanic island lead: two-stage histories and mantle evolution, Earth Planet. Sci. Len. 52, 277-284, 1981. 8 A. Zindler and S.R. Hart, Chemical geodynamics, Annu. Rev. Earth Planet. Sci. 14, 493-571, 1986. 9 W.M. White and A.W. Hofmann, Sr and Nd isotope geochemistry of oceanic basalts and mantle evolution, Nature 296, 821-825, 1982. 10 E. Wright and W.M. White, The origin of Samoa: new evidence from Sr, Nd and Pb isotopes, Earth Planet. Sci. Lett. 81, 151- 162, 1987. 11 Z.A. Palacz and A.D. Saunders, Coupled trace element and isotope enrichment in the Cook-Austral-Samoa islands, southwest Pacific, Earth Planet. Sci. Lett. 79, 270280, 1986. 12 C. Dupuy, H.G. Barsczus, J. Dostal, P. Vidal and J.-M. Liotard, Subducted and recycled lithosphere as the mantle source of ocean island basalts from southern Polynesia, central Pacific, Chem. Geol. 77, 1-18, 1989. 13 R.A. Duncan and I. McDougall, Linear volcanism in French Polynesia, J. Volcanol. Geotherm. Res. 1, 197-227, 1976. 14 H. Bellon, R. Brousse and A. Pantaloni, Ages de l'ile de Tubuai: l'alignement des Australes et des Cook, Cah. Pac. 2, 1980. 15 D.L. Turner and R.D. Jarrard, K-Ar dating of the CookAustral chain: a test of the hot-spot hypothesis, J. Volcanol. Geotherm. Res. 12, 187-220, 1982. 16 C. Diraison, Le volcanisme aerien des Archipels Polynesiens de la Societe, des Marquises et des Australes-Cook, Ph.D. Thesis, Univ. Bretagne Occidentale, 1991. 17 T. Gisbert, Volcanolog!e de l'ile de Tahaa, Archipel de la Societe, Doct. Sci. Thesis, Univ. Paris-Sud, Orsay, Paris, 1989.

18 G.B. Dalrymple, R.D. Jarrard and D.A. Clague, K-Ar ages of some volcanic rocks from the Cook and Austral Islands, Geol. Soc. Am. Bull. 86, 1463-1467, 1975. 19 J.W. Morgan, Ultramafic xenoliths: clues to Earth's late accretionary history, J. Geophys. Res. 91, 12,375-12,387, 1986. 20 J.W. Morgan and J.F. Lovering, Rhenium and osmium abundances in some igneous and metamorphic rocks, Earth Planet. Sci. Lett. 3, 219-224, 1967. 21 R.J. Walker, S.B. Shirey, G.N. Hanson, V. Rajamani and M.F. Horan, Re-Os, Rb-Sr, and O isotope systematics of the Archean Kolar schist belt, Karnataka, India, Geochim. Cosmochim. Acta 53, 3005-3013, 1989. 22 M.F. Horan, J.W. Morgan, R.J. Walker and J.N. Grossman, Rhenium-osmium isotope constraints on the age of iron meteorites, Science 255, 1118-1121, 1992. 23 J.W. Morgan, R.J. Walker and J.N. Grossman, Rheniumosmium isotope systematics in meteorites I: magmatic iron meteorite groups IIAB and IIIAB, Earth Planet. Sci. Lett. 108, 191-202, 1992. 24 C.J. Allegre and J.M. Luck, Osmium isotopes as petrogenetic and geological tracers, Earth Planet. Sci. Lett. 48, 148-154, 1980. 25 S.R. Hart and E.D. Kinloch, Osmium isotope systematics in Witwatersrand and Bushveld ore deposits, Econ. Geol. 84, 1651- 1655, 1989. 26 K. Hattori and S.R. Hart, Osmium isotope ratios of platinum- group minerals associated with ultramafic intrusions: Os isotopic evolution of the mantle, Earth Planet. Sci. Lett. 107, 499-514, 1991. 27 J.M. Luck and C.J. Allegre, 187Re-lS7Os systematics in meteorites and cosmochemical consequences, Nature 302, 130-132, 1983. 28 R.J. Walker and J.W. Morgan, Rhenium-osmium isotope systematics of carbonaceous chondrites, Science 243, 519522, 1989. 29 J.-M. Luck and C.J. Allegre, Osmium isotopes in ophiolites, Earth Planet. Sci. Lett. 107, 406-415, 1992. 30 C.E. Martin, B.K. Esser and K.K. Turekian, Re-Os isotopic constraints on the formation of mantle and crustal reservoirs, Aust. J. Earth Sci. 38, 1992. 31 R.J. Walker, R.W. Carlson, S.B. Shirey and F.R. Boyd, Os, St, Nd, and Pb isotope systematics of southern African peridotite xenoliths: Implications for the chemical evolution of subcontinental mantle, Geochim. Cosmochim. Acta. 53, 1583-1595, 1989. 32 D.G. Pearson, S.B. Shirey, R.W. Carlson, F.R. Boyd, P.H. Nixon and N.P. Pokhilenko, Re-Os and Sm-Nd isotopic constraints on the chronology of lithospheric mantle formation and enrichment events, EOS 72, 527, 1991. 33 L.C. Reisberg, C.J. Allegre and J.M. Luck, The Re-Os systematics of the Ronda Ultramafic Complex of southern Spain, Earth Planet. Sci. Lett. 105, 196-213, 1991. 34 M. Tatsumoto, Isotopic composition of lead in oceanic basalts and its implication for mantle evolution, Earth Planet. Sci. Lett. 38, 63-87, 1978. 35 P. Vidal, C. Chauvel and R. Brousse, Large mantle heterogeneity beneath French Polynesia, Nature 307, 536-538, 1984. 36 Y. Nakamura and M. Tatsumoto, Pb, Nd, and Sr isotopic

370

37

38

39

40

41

42 43

44 45

46

47

48

49

50

51

52

53

evidence for a multicomponent source for rocks of CookAustral Islands and heterogeneities of mantle plumes, Geochim. Cosmochim. Acta 52, 2909-2924, 1988. W.F. McDonough and C. Chauvel, Sample contamination explains the Pb isotopic composition of some Rurutu Island and Sasha seamount basalts, Earth Planet. Sci. Lett. 106, 397-404, 1991. R.J. Walker, L.M. Echeverria, S.B. Shirey and M.F. Horan, Re-Os isotopic constraints on the origin of volcanic rocks, Gorgona Island, Columbia: Os isotopic evidence for ancient heterogeneities in the mantle, Contrib. Mineral. Petrol. 107, 150-162, 1991. R.J. Walker, S.B. Shirey and O. Stecher, Comparative Re-Os, Sm-Nd and Rb-Sr isotope and trace element systematics for Archean komatiite flows from Munro Township, Abitibi Belt, Ontario, Earth. Planet. Sci. Lett. 87, 1-12, 1988. W.J. Pegram and C.J. Allegre, Osmium isotopic compositions from oceanic basalts, Earth Planet. Sci. Lett. 111, 59-68, 1992. E.B. Watson, D. Ben Othman, J.M. Luck and A.W. Hofmann, Partitioning of U, Pb, Hf, Yb, Cs, Re and Os between chromian diopsidic pyroxene and haplobasaltic liquid, Chem. Geol. 62, 191-208, 1987. J.E. Snow, Os isotopic composition of the oceanic upper mantle, EOS 72, 527, 1991. W.M. White and P.J. Patchett, Hf-Nd-Sr isotopes and incompatible element abundances in island arcs: Implications for magma origins and crust-mantle evolution, Earth Planet. Sci. Lett. 67, 167-185, 1984. R. Vollmer, Terrestrial lead isotopic evolution and formation of the earth's core, Nature 270, 144-147, 1977. P. Vidal and L. Dosso, Core formation: catastrophic or continuous? Sr and Pb isotope geochemistry constraints, Geophys. Res. Lett. 5, 169-171, 1978. H.E. Newsom, W.M. White, K.P. Jochum and A.W. Hofmann, Siderophile and chalcophile element abundances in oceanic basalts, Pb isotope evolution and growth of the Earth's core, Earth Planet. Sci. Lett. 80, 299-313, 1986. R.S. Cohen, N.M. Evensen, P.J. Hamilton and R.K. O'Nions, U-Pb, Sm-Nd and Rb-Sr systematics of mid-ocean ridge basalt glasses, Nature 283, 149-153, 1980. R.S. Cohen and R.K. O'Nions, The lead, neodymium and strontium isotopic structure of ocean ridge basalts, J. Petrol. 23, 299-324, 1982. B. Bourdon, A. Zindler and L. Reisberg, Pb isotopic results for the Ronda ultramafic complex, Geol. Soc. Aus. Abstr., 1990. L. Reisberg and A. Zindler, Extreme isotopic variations in the upper mantle: evidence from Ronda, Earth Planet. Sci. Lett. 81, 29 45, 1986. L. Reisberg, A. Zindler and E. Jagoutz, Further Sr and Nd isotopic results from peridotites of the Ronda Ultramafic Complex, Earth Planet. Sci. Lett. 96, 161-180, 1989. B. Hamelin and C.J. Allegre, Lead isotope study of orogenic lherzolite massifs, Earth Planet. Sci. Lett. 91, 117131, 1988. S.J.G. Galer and R.K. O'Nions, Chemical and isotopic studies of ultramafic inclusions from the San Carlos volcanic field, Arizona: a bearing on their petrogenesis, J. Petrol. 30, 1033- 1064, 1989.

E.H. HAUR1and S.R. HART 54 S.R. Hart, Heterogeneous mantle domains: signatures, genesis and mixing chronologies, Earth Planet. Sci. Lett. 90, 273-296, 1988. 55 W.M. White and J.G. Schilling, The nature and origin of geochemical variation in Mid-Atlantic Ridge basalts from central North Atlantic, Geochim. Cosmochim. Acta 42, 1501-1516, 1978. 56 J.J. Mahoney, J.H. Natland, W.M. White, R. Poreda, S.H. Bloomer, R.L. Fisher and A.N. Baxter, Isotopic and geochemical provinces of the western Indian Ocean spreading centers, J. Geophys. Res. 94, 4033-4052, 1989. 57 L. Dosso, B.B. Hanan, H. Bougault, J.-G. Schilling and J.-L. Joron, Sr-Nd-Pb geochemical morphology between 10° and 17° N on the Mid-Atlantic Ridge: a new MORB isotope signature, Earth Planet. Sci. Lett. 106, 29-43, 1991. 58 P.J. Patchett and M. Tatsumoto, Hafnium isotope variations in oceanic basalts, Geophys. Res. Lett. 7, 1077-1080, 1980. 59 P.J. Patchett, Hafnium isotope results from mid-ocean ridges and Kerguelen, Lithos 16, 47-51, 1983. 60 V.J.M. Salters and S.R. Hart, The Hf-paradox, and the role of garnet in the MORB source, Nature 342, 420-422, 1989. 61 S.E. Church and M. Tatsumoto, Lead isotope variations in oceanic ridge basalts from the Juan de Fuca-Gorda Ridge area, N.E. Pacific Ocean, Contrib. Mineral. Petrol. 53, 253-279, 1975. 62 M. Condomines, P. Morand and C.J. Allegre, 23°Th-238U radioactive disequilibria in tholeiites from the FAMOUS zone (Mid-Atlantic Ridge, 36°50'N): Th and Sr isotopic geochemistry, Earth Planet. Sci. Lett. 55, 247-256, 1981. 63 K.P. Jochum, A.W.I. Hofmann E., H.M. Seufert and W.M. White, K, U and Th in mid-ocean ridge basalt glasses and heat production, K / U and K / R b in the mantle, Nature 306, 431-436, 1983. 64 S. Newman, R.C. Finkel and J.D. Macdougall, 23°Th-23Su disequilibrium systematics in oceanic tholeiites from 21°N on the East Pacific Rise, Earth Planet. Sci. Lett. 65, 17-33, 1983. 65 S.J. Goldstein, M.T. Murrell, D.R. Janecky, J.R. Delaney and D.A. Clague, Geochronology and petrogenesis of MORB from the Juan de Fuca and Gorda ridges by 238U/23°Th disequilibrium, Earth Planet. Sci. Lett. 107, 24-41, 1991. 66 A.W. Hofmann, Chemical differentiation of the Earth: the relationship between mantle, continental crust, and oceanic crust, Earth Planet. Sci. Lett. 90, 297-314, 1988. 67 S.R. Hart and H. Staudigel, Isotopic characterization and identification of recycled components, in: Crust/Mantle Recycling at Convergence Zones (NATO ASI Ser. C), Vol. 258, pp. 15-28, Reidel, 1989. 68 S.R. Hart and H. Staudigel, The control of alkalies and uranium in sea water by ocean crust alteration, Earth Planet. Sci. Lett. 58, 202-212, 1982. 69 C. Chauvel, A.W. Hofmann and P. Vidal, HIMU-EM: the French Polynesian connection, Earth Planet. Sci. Lett. 110, 99-119, 1992. 70 D. Colodner, G. Ravizza, J. Sachs, K. Turekian, J. Edmond and E. Boyle, The geochemical cycle of rhenium: a reconnaissance, Earth Planet. Sci. Lett., in prep., 1992.

371

Re-Os I S O T O P E SYSTEMATICS F R O M T H E S O U T H PACIFIC

71 A.D. Anbar, R.A. Creaser, D.A. Papanastassiou and G.J. Wasserburg, Rhenium in seawater: confirmation of generally conservative behavior, Geochim. Cosmochim. Acta, submitted, 1992. 72 G. Ravizza and K.K. Turekian, The osmium isotopic composition of organic-rich marine sediments, Earth Planet. Sci. Lett. 110, 1-6, 1992. 73 M. Roy-Barman, J.L. Birck and C.J. Allegre, Osmium isotopic composition of the mantle, EOS 73, 374-375, 1992. 74 P.B. Kelemen, K.T.M. Johnson, R.J. Kinzler and A.J. Irving, High field strength element depletions in arc basalts due to mantle-magma interaction, Nature 345, 521-524, 1990. 75 D.P. McKenzie, The generation and compaction of partially molten rock, J. Petrol. 25, 713-765, 1984. 76 O. Navon and E. Stolper, Geochemical consequences of melt percolation: the upper mantle as a chromatographic column, J. Geol. 95, 285-307, 1987. 77 D. McKenzie and R.K. O'Nions, Partial melt distributions from inversion of rare earth element concentrations, J. Petrol. 32, 1021-1091, 1991. 78 K.A. Farley, J. Natland, J.D. MacDougall and H. Craig, He, Sr and Nd isotopes in Samoan basalts: evidence for enriched and undepleted mantle components, EOS 71, 1669, 1989. 79 S.R. Hart, E.H. Hauri, L. Oschman and J.A. Whitehead, Mantle plumes and entrainment: the isotopic evidence, Science 256, 517-520, 1992. 80 A. Reymer and G. Schubert, Phanerozoic addition rates to the continental crust and crustal growth, Tectonics 3, 63-77, 1984. 81 DJ. Charley, R.A. Cliff and B.M. Wilson, Characteriza-

82

83

84

85

86

87

88

89

tion of the St. Helena magma source, in: Magmatism in the Ocean Basins, A.D. Saunders and M.J. Norry, eds., Geol. Soc. London Spec. Publ. 42, 257-276, 1988. H. Staudigel, K.-H. Park, M. Pringle, J.L. Rubenstone, W.H.F. Smith and A. Zindler, The longevity of the South Pacific isotopic and thermal anomaly, Earth Planet. Sci. Lett. 102, 24-44, 1991. E. Ito, W.M. White and C. Goepel, The O, Sr, Nd and Pb isotope geochemistry of MORB, Chem. Geol. 62, 157-176, 1987. E.L. Hoffman, A.J. Naldrett, J.C.V. Loon, R.G.V. Hancock and A. Mason, The determination of all the platinum group elements and gold in rocks and ore by neutron activation analysis after preconcentration by a nickel sulfide fire-assay technique of large samples, Anal. Chim. Acta 102, 157-166, 1978. J.M. Luck and C.J. Allegre, The study of molybdenites through the lSYRe-lS7Os chronometer, Earth Planet. Sci. Lett. 61,291-296, 1982. R.A. Creaser, D.A. Papanastassiou and G.J. Wasserburg, Negative thermal ion mass spectrometry of osmium, rhenium, and iridium, Geochim. Cosmochim. Acta 55, 397401, 1991. J. Volkening, T. Walczyk and K.G. Heumann, Osmium isotope ratio determinations by negative thermal ionization mass spectrometry, Int. J. Mass Spectrom. Ion Phys. 105, 147-159, 1991. A.O. Nier, A redetermination of the relative abundances of the isotopes of carbon, nitrogen, oxygen, argon and potassium, Phys. Rev. 77, 789-793, 1950. W. Todt, R.A. Cliff, A. Hanser and A.W. Hofmann, 2o2Pb + 20sPb double spike for lead isotopic analyses, Terra Cognita 4, 209, 1984.