Full text of DOI:10.1007/s004100000203

Contrib Mineral Petrol 92001) 140: 588±599
DOI 10.1007/s004100000203
I. S. Puchtel á G. E. Brugmann á A. W. Hofmann
V. S. Kulikov á V. V. Kulikova
Os isotope systematics of komatiitic basalts from the Vetreny belt,
Baltic Shield: evidence for a chondritic source of the 2.45 Ga plume
Received: 13 June 2000 / Accepted: 20 September 2000 / Published online: 7 November 2000
Ó Springer-Verlag 2000
Abstract The Vetreny belt in the southeast Baltic Shield homogenized with respect to the highly siderophile ele-
is a large volcano-sedimentary basin containing a 4- to ments, added during the accretion of a late veneer, on a
8-km-thick sequence of basaltic to komatiitic lavas, time scale of ꢀ1 Ga.
which were erupted ꢀ2.45 Ga ago in a continental rift
setting during the interaction of a mantle plume and the
Archean continental crust of the Karelian granite±
greenstone terrane. Re-Os isotope data for olivine cu-
mulate samples and chromite separates from Victoria's
Introduction
Global models of the origin of radiogenic 9e.g., modern
lava lake and Golets ¯ow 3 de®ne isochrons with ages of
plume sources) and non-radiogenic 9e.g., subcontinental
2387‹57 and 2432‹34 Ma, respectively. These ages are
lithospheric mantle) Os isotope reservoirs in the mantle
in good agreement with the previously reported, and
are based on the assumption of chondritic evolution of
new, Nd-Pb and U-Pb zircon ages. These data, coupled
the upper mantle through time with respect to highly
with the evidence for immobile behavior of the Re and
siderophile elements 9HSE). The extent of deviation of
Os, indicate that the Re-Os system remained closed since
Os isotope composition of certain mantle sources from
the chondritic composition is used in these models to
estimate the amount of ancient oceanic crust or outer
the lava eruption. The weighted average initial c187Os
values range between ±0.43‹0.10 for Golets ¯ow 3 and
±0.07‹0.13 for the lava lake. A single chromite separate
core material added or continental crust subtracted to
account for the Os-enriched or Os-depleted isotope sig-
from Golets ¯ow 1 has a c187Os9T) of ±0.06‹0.15. Nd-
Pb isotope and lithophile trace element data for the
natures of these mantle sources. The hypothesis of
Vetreny belt komatiitic basalts were used to monitor the
chondritic evolution of the upper mantle is in turn
e€ect of crustal contamination on the Os isotope system
mostly based on the Re-Os data derived from modern
and to estimate the initial Os isotope composition of the
Vetreny plume source to be approximately chondritic
MORB and abyssal peridotites 9Roy-Barman and Alle-
gre 1994; Snow and Reisberg 1995; Schiano et al. 1997;
Brandon et al. 2000) and, ultimately if proved, will
provide one of the most compelling arguments behind
with a c187Os9T) of ±0.9. This implies that the mantle
source evolved with a long-term nearly chondritic Re/Os
ratio. The data provide further evidence that, by the end
the hypothesis of the accretion of a late veneer, i.e.,
of the late Archean, the Earth's upper mantle was well
addition of 1±2% material with chondritic relative
abundances of HSE to the accreting Earth after core
segregation was essentially completed 9Kimura et al.
1974; Morgan 1985).
I. S. Puchtel 9&)
Department of the Geophysical Sciences,
The University of Chicago, 5734 S. Ellis Avenue,
Chicago, IL 60637, USA
Though there is a growing consensus about the
chondritic HSE relative abundances in the modern up-
per mantle 9e.g., Meisel et al. 1996), very few data still
exist that would demonstrate the chondritic evolution of
HSE in the upper mantle as far back in Earth history as
the Archean and early Proterozoic. Recent advances in
the negative thermal ionization mass spectrometry
technique made it possible to obtain accurate Re-Os
data on several late Archean, plume-derived mantle
rocks including Munro Township and Kambalda
komatiites 9Foster et al. 1996; Shirey 1997) and early
e-mail: ipuchtel@geosci.uchicago.edu
I. S. Puchtel á G. E. Brugmann á A. W. Hofmann
Max-Planck Institut fur Chemie, Abteilung Geochemie,
Postfach 3060, 55020 Mainz, Germany
V. S. Kulikov á V. V. Kulikova
Institute of Geology, Karelian Research Center RAS,
Pushkinskaya 11, 185610 Petrozavodsk, Russia
Editorial responsibility: J. Hoefs

589
Proterozoic ophiolites 9Shirey and Barnes 1994; Walker
et al. 1996; Tsuru et al. 2000), which provide evidence
that at least part of the early Precambrian upper and,
possibly, lower mantle was indeed chondritic with
respect to the relative abundances of HSE. In this study,
Re-Os data on the well-preserved komatiitic basalts
from the 2.45 Ga Vetreny belt in the Baltic Shield are
presented. We combine these with the previously
reported and new lithophile Nd-Pb isotope and trace
element data to assess the e€ects of crustal contami-
nation on the Os isotope composition of komatiite
magmas, and to reconstruct the initial Os isotope
composition of the Vetreny plume mantle source.
Geological background and previous Pb-Nd
and U-Pb zircon studies
The Karelian Province in the Baltic, or Fennoscandian,
Shield, is an Archean granite±greenstone terrane over-
lain by Paleoproterozoic epicontinental basins and vol-
cano-sedimentary belts 9Fig. 1). The Paleoproterozoic
rocks are subdivided into several informal stratigraphic
units 9from bottom upwards): the Sumian-Sariolian,
Jatulian-Ludikovian, and Vepsian-Kalevian 9Gaal and
Gorbatschev 1987; Sokolov 1987). Most of the Sumian-
Sariolian rocks occur in NW-trending volcano-sedi-
mentary belts, the largest of which, the Vetreny belt in
the southeast part of the shield, can be traced from Lake
Vyg southeastward over a distance of more than 250 km
9Fig. 1). The Paleoproterozoic rocks of the Vetreny belt
comprise six major suites with a total thickness of the
whole succession ranging between 4 and 8 km 9see Fig. 1
in Puchtel et al. 1997). The lower ®ve suites are com-
posed of intercalated basaltic to andesitic lavas and
terrigenous sediments. The uppermost Vetreny suite
consists entirely of komatiitic basalts. The Vetreny belt
is interpreted to have been formed in a continental rift
setting during the interaction of a mantle plume and the
Archean continental crust of the Karelian granite±
greenstone terrane. Impinging of a plume head beneath
Fig. 1 Geological sketch map of the Karelian granite-greenstone
terrane, Baltic Shield 9modi®ed after Sokolov 1987). Study areas:
1 Golets Hills, 2 Lion Hills
the continental lithosphere some 2.45 Ga ago resulted in et al. 91997) reported Sm-Nd primary mineral isochron
its thinning, stretching, and rifting, and eruption of vast ages of 2449‹35 and 2410‹34 Ma, a Pb-Pb bulk rock
volumes of ma®c and ultrama®c magmas over a short isochron age of 2424‹178 Ma for the uppermost
period of time 9Puchtel et al. 1997).
Vetreny suite komatiitic basalts, and a U-Pb zircon age
Previous Nd-Pb isotope and geochemical studies, and of 2437‹3 Ma for the lowermost Kirichi suite andesites.
petrological modeling indicate that magmas parental to These ages are identical to the published U-Pb zircon
the Vetreny belt lava sequences were komatiitic in and baddeleyite ages for numerous ma®c±ultrama®c
composition 9>18% MgO) and were derived from a layered intrusions in central and northern Karelia
long-term LREE-depleted mantle source with ꢀ143Nd9T) 9Amelin et al. 1995) and, together with the similarities in
of approximately +2.6 9Puchtel et al. 1997). These chemical and isotope compositions, led Puchtel et al.
workers showed that the chemical evolution of the 91997) to propose the origin of the lavas and the intru-
Vetreny komatiite magmas en route to the surface was sions in a single mantle plume and likely from similar
mainly controlled by fractional crystallization and parental magmas.
4±15% contamination with felsic crustal rocks of the
For the Re-Os isotope studies, we have chosen the
adjacent Archean Vodla and Belomorian Blocks. freshest samples from two localities: Golets Hills in
Assimilation of crustal material resulted in substantial the northwestern part of the belt and Lion Hills in the
changes in incompatible trace element and Pb-Nd central Vetreny belt 9Fig. 1). Samples from the Golets
isotope ratios in the original komatiite magmas. Puchtel Hills area include bulk rock olivine cumulates and

590
Puchtel et al., unpublished). We used the Carius tube digestion
technique introduced by Shirey and Walker 91995). Osmium ex-
traction and puri®cation followed the method of Birck et al. 91997).
Rhenium was recovered from the residual solutions and then fur-
ther puri®ed using the anion exchange separation technique de-
scribed in Morgan and Walker 91989). The total analytical blank,
including the ®lament blank, was ꢀ5 pg for Re and <0.5 pg for Os.
Isotopic compositions of Re and Os were measured in
magnet switching mode via NTIMS on a MAT-262 FinniganÒ
mass spectrometer. The e€ects of fractionation during Os runs
were eliminated by normalizing the Os isotope ratios to
192Os/188Os=3.082678. Measured Re-isotopic ratios were not
corrected for fractionation. Analytical uncertainties on Os isotope
compositions 9typically in the 0.1±0.2% range) are reported for
each sample 9Table 1). Most data represent averages of two inde-
pendent runs. To calculate the age, Re±Os data were regressed
using the ISOPLOT program 9Ludwig 1997) and assuming an error
correlation coecient of 0.9. Error input was determined by mul-
tiple analyses 9n=20 during the period of data collection) of the in-
olivine and spinel separates from ¯ow 3 and a single
chromite separate from ¯ow 1. In the Lion Hills area,
whole rock olivine cumulates and spinel separates from
the 110-m-deep Victoria's lava lake were analyzed 9see
Fig. 2 in Puchtel et al. 91996) for sample location and
detailed information on the geology of the lava lake). All
samples are characterized by a superb state of preser-
vation by Precambrian standards, with the metamorphic
grade barely exceeding the prehnite-pumpellyite facies.
Samples from the lava lake are almost completely
unaltered, and even retain primary volcanic glass.
In this study, additional Pb±Pb isotope data on the
lava lake and Golets ¯ow 3 were also obtained, which
are complementary to those for Golets ¯ow 1 reported
by Puchtel et al. 91997).
house Os and Re standards and were found to be ‹0.5% 92rpop
)
on both the Os isotope compositions and Re/Os ratios. All errors
on age and initial isotopic ratios are quoted at 2r, or 95% con®-
dence level. The initial c187Os value was calculated as the per-
Analytical techniques
The Re-Os isotope studies were performed at the Max-Planck In- centage deviation of the isotopic composition of a rock at the time
stitut 9MPI) fur Chemie in Mainz, and the details of the analytical de®ned by the isochron relative to the average chondritic
procedures employed were speci®ed by Puchtel et al. 91999a; composition at that time 9Eq. 91)):
Fig. 2 Re-Os isochron
diagrams for 9a) Golets and
9b) Lion Hills komatiitic basalts

591
c187Os9T)
Table 1 Re-Os isotopic data
for the Vetreny belt komatiitic
basalts
Sample
Re 9ppb) Os 9ppb)
187Re/188Os 187Os/188Os
Golets Hills, ¯ow 1
Chromite separate
8991 Crt
0.2154
10.09
0.10211
0.11453‹17
±0.06
Golets Hills, ¯ow 3
Olivine cumulates
8950
8954
1.8912
0.5790
0.5252
0.6080
0.9318
0.6853
0.9482
0.7285
9.8541
4.1298
2.6863
4.0786
0.22854‹33
0.28031‹40
0.22064‹23
0.27918‹20
±263
±1.07
±0.93
±0.17
8953
89100
Olivine and chromite separates
89100 Crt
8954 Crt
8950 Ol
0.4393
0.2778
0.0354
9.4317
4.1097
0.0509
0.2229
0.3238
3.3342
0.11856‹54
0.12402‹17
0.1408‹14
±0.94
0.22
±98
Lion Hills, Victoria's lava lake
Olivine cumulates
91103
91104
91105
91106
0.3940
0.3831
0.2934
0.3884
1.2117
1.1529
1.0554
1.1145
1.5675
1.6014
1.3261
1.6810
0.17407‹27
0.17641‹35
0.16521‹17
0.17888‹49
±1.13
±0.28
±0.09
±1.03
Chromite separates
91103 Crt
91105 Crt
2.2041
0.8275
31.904
31.622
0.3309
0.1252
0.12387‹37
0.11652‹14
±0.18
0.88
0.04% per atomic mass unit. All errors on age and initial isotopic
ratios are quoted at 95% con®dence level.
2
3
7
ꢀ
ꢀ
ꢁsample
187
188
Os
Os
6
4
c187OsꢁT† ˆ 100 Â
À 1
ꢁ1†
i
5
ꢁCHON
187
188
Os
Os
Results
i
The initial 187Os/188Os of a particular sample was calculated
from its measured Os isotopic composition and the Re/Os ratio
using Eq. 92):
Re±Os isotope data
Re±Os isotopic compositions and Re and Os concentra-
tions were measured in bulk rock samples and spinel and
olivine separates from Golets ¯ows 1 and 3 and Victoria's
lava lake in Lion Hills. The data obtained are listed in
Table 1 and plotted on the Re±Os isochron diagrams in
Fig. 2. In the Golets Hills area, two spinel separates and
olivine cumulates from the lower part of ¯ow 3 de®ne a
®ve-point isochron with an age of 2432 ‹ 34 Ma and an
initial 187Os/188Os of 0.1099‹0.0016, corresponding to a
c187Os9T) value of ±0.42‹1.5. One sample 9#8950) has an
anomalously high Re/Os ratio not supported by the
respective increase in 187Os/188Os, and thus plots well o€
the regression line. As this sample, collected from the
lowermost part of the ¯ow, has a Re concentration about
three times higher than an average cumulate sample from
ꢂ
ꢃsample
ꢂ
ꢃ
ꢂ
ꢃsample
187Os
188Os
187Os
188Os
187Re
188Os
À
Á
ˆ
sampleÀ
 ekT À 1
ꢁ2†
i
measur
measur
The average chondritic Os isotope composition at the time T
was calculated from Eq. 93) using the 187Re decay constant 9k)
of 1.666 ´ 10±11 year±1 and the present-day average chondritic
composition adopted from Shirey and Walker 91998):
187Re/188Os=0.40186, 187Os/188Os=0.1270:
ꢂ
ꢃCHON
187Os
188Os
À
Á
ˆ 0:09531 ‡ 0:40186  ek4:558 À ekT
ꢁ3†
i
Pb isotope data were obtained at the MPI fur Chemie using a
MAT-261 FinniganÒ mass spectrometer and following the methods
described by Dupre and Arndt 91990). Small chips of rock were
leached in hot 2.5 to 6 N HCl for 30 to 60 min to remove Pb that
may have been introduced during sample preparation or secondary
processes. Clinopyroxene separates were double-washed in ethanol the ¯ow, the phenomenon is likely due to a recent Re gain.
and 18MW deionized water and then leached in warm 2.5 N HCl
for 20±30 min. Samples were dissolved in a mixture of concentrated
HF + HNO3 in closed Te¯on vials. Pb was separated using the
method of Manhes et al. 91978). The total procedural blank was
<100 pg. The Pb isotopic ratios were corrected for isotopic frac-
tionation, which was determined by multiple analyses of the stan-
dard SRM 982 to be 1.00083 per atomic mass unit. The ``true'' Pb
isotopic ratios for the SRM 982 standard were adopted from Arndt
and Todt 91994) to be 36.738, 17.159, and 36.744. The l1 value was
calculated using a single-stage model, assuming 4.50 Ga as the age
of the Earth and Canyon Diablo values 9Tatsumoto et al. 1973)
for the starting isotopic composition. To calculate the age,
207Pb/204Pb and 206Pb/204Pb data were regressed using the
York 91969) procedure, the ISOPLOT program 9Ludwig 1992), and
assuming an error correlation coecient of 0.9. Error input was
The olivine separate from this sample also plots below the
regression line. The olivine shows some evidence of
secondary alteration, and is also likely to have gained a
certain amount of excess Re during this process. These
two samples were not included in the regression calcula-
tions. The single measured chromite separate 98991 Crt)
from ¯ow 1 has an initial c187Os of ±0.06‹0.15.
In Victoria's lava lake, samples of four olivine
cumulates and two spinel separates yield an isochron
with an age of 2384‹57 Ma. The calculated initial
187Os/188Os of 0.1110‹0.0013 is approximately chon-
dritic 9c187Os9T)=0.28‹1.1). The signi®cant uncertain-
determined by multiple analyses of the standard SRM 982 to be ty on the initial ratios for both ¯ow 3 and the lava lake

592
are attributed to the large extrapolation from the cu- Pb±Pb systematics
mulate samples with high Re/Os. In order to minimize
the error propagation from the extrapolation towards The new Pb isotope data for Golets ¯ow 3 and Victoria's
Re/Os=0 in the isochron diagram, we calculated the lava lake together with the previously published data for
weighted averages of initial 187Os/188Os. The calcula- Golets ¯ow 1 9Puchtel et al. 1997) are presented in
tions yield more precise initial 187Os/188Os ratios of Table 2 and plotted in Fig. 3.
0.10989‹0.00011 9c187Os9T)=±0.43‹0.10) and 0.11029
For the lava lake, residues of spinifex-textured and
‹0.00014 9c187Os9T)=±0.07‹0.13) for ¯ow 3 and the olivine cumulate samples and clinopyroxene separates
lava lake, respectively.
de®ned an isochron with an age of 2437‹140 Ma. This
Table 2 Pb±Nd isotope and
trace element data. Pb data for
Golets ¯ow 1 and Nd and trace
element data from Puchtel et al.
91997)
Sample
206Pb/204Pb 207Pb/204Pb 208Pb/204Pb l1
ꢀ143Nd9T) 9La/Sm)
Zr/Y
N
Golets Hills, ¯ow 1
Flowtop breccia
8978
17.789
17.882
Spinifex±textured komatiitic basalt
15.751
15.759
38.263
38.137
9.29 )1.42
9.28
2.44
±
4.46
±
8978a
±
8982
8984
18.876
18.247
18.253
18.407
17.794
17.552
18.398
18.740
18.487
15.902
15.818
15.824
15.857
15.757
15.701
15.828
15.879
15.844
40.038
39.362
39.387
39.557
38.714
38.358
39.510
40.015
39.632
9.25
±
±
2.64
±
±
2.47
±
±
2.52
±
±
4.46
±
±
4.43
±
±
4.21
±
9.28 )1.41
9.29
9.31
8984a
8986
±
±
8987
9.30 )1.44
8987a
8989
9.26
9.25
9.24
9.25
±
±
±
±
8990
8990a
Olivine cumulate
8991
18.045
17.944
18.172
17.367
17.913
15.763
15.765
15.799
15.653
15.745
39.065
38.937
39.178
38.154
38.754
9.23 )1.37
9.27
9.26 )1.64
9.22
9.24
2.47
±
2.50
±
4.29
±
4.48
±
8991a
8993
8994
8995
±
±
±
±
±
Golets Hills, ¯ow 3
Spinifex±textured komatiitic basalt
8996
8997
8998
16.662
16.833
16.598
17.760
17.780
15.465
15.497
15.469
15.657
15.656
37.056
37.397
37.069
38.502
38.570
9.05
9.06
9.08
9.09
9.08
0.55
±
±
0.03
±
1.86
±
±
1.93
±
3.62
±
±
3.68
±
8999
8999a
Olivine cumulate
8950
17.673
17.459
17.621
17.646
17.417
16.974
15.612
15.594
15.609
15.612
15.593
15.508
38.380
38.105
38.315
38.380
38.001
37.704
9.02
9.06
9.03
9.03
9.07
9.04
±
±
0.17
±
0.32
0.27
±
±
1.91
±
1.89
1.84
±
±
3.56
±
3.31
3.47
8953
8954
8956
8957
89100
Lion Hills, Victoria's lava lake
Spinifex±textured komatiitic basalt
91100
91101
91107
91117
91119
91121
91122
91124
16.761
17.696
17.326
17.837
16.702
16.995
17.166
17.195
15.547
15.673
15.643
15.713
15.525
15.588
15.623
15.635
37.351
38.522
38.007
38.707
37.282
37.565
37.874
37.949
9.20
±
±
2.14
±
2.14
±
±
±
±
±
3.94
±
3.94
±
±
±
±
9.15 )0.87
9.21
9.19 )0.76
9.17
9.21
9.23
9.24
±
±
±
±
±
Olivine cumulate
91103
91104
91105
91106
17.160
17.333
17.258
17.812
15.607
15.618
15.606
15.702
37.685
37.934
37.988
38.746
9.19 )0.57
9.16 )1.12
9.15 )0.82
9.17 )0.90
2.11
2.13
2.14
2.12
3.83
3.91
3.91
3.96
Clinopyroxene separates
91103 Cpx
91104 Cpx
91106 Cpx
16.064
16.535
16.087
15.431
15.511
15.427
36.444
37.058
36.504
9.19 )0.95
9.20 )1.24
9.17 )0.92
±
±
±
±
±
±
a Separate chips from the same sample

593
Fig. 3 Pb-Pb isotope data for the Vetreny belt komatiitic basalts.
The mantle evolution curves were drawn assuming a single-stage Discussion
evolution model, a l1 mantle value of 8.75, 4.50 Ga as the age of
the Earth, and Canyon Diablo values 9Tatsumoto et al. 1973) for
Pb±Pb isotope data
the starting Pb isotopic composition
The obtained Pb±Pb isochron ages of 2437‹140 and
2434‹190 Ma for the lava lake and Golets ¯ow 3 coincide
isochron plots above a hypothetical primary mantle with the previously reported eruption ages of the Vetreny
evolution curve implying that the erupted magmas had belt komatiitic basalt magmas 9see review in introduction)
a more radiogenic Pb isotopic composition than the and, therefore, are an additional argument for a simul-
contemporary mantle. This is also re¯ected by a high l1 taneous and very rapid, areally extensive, catastrophic
value of the rocks 99.19‹0.16 vs. 8.75 in the mantle). eruption of ma®c±ultrama®c magmas in the region at
Residues of spinifex-textured and olivine cumulate ꢀ2.45 Ga. They are perhaps part of a global, plume-re-
komatiitic basalt samples from Golets ¯ow 3 gave a lated ma®c magmatism at that time, which resulted in the
nearly identical isochron age of 2434‹190 Ma. Golets formation of one of the oldest known large igneous
¯ow 3 had a slightly less radiogenic Pb isotopic com- provinces on Earth 9Heaman 1997). The fact that the
position than the lava lake, as manifested by a lower samplesfrom the threelocalities form regressionlineswith
calculated l1 value of the rocks 99.05‹0.21). Finally, correct ages 9i.e., isochrons) also implies a closed-system
the samples from Golets ¯ow 1 had the most radiogenic behavior for Pb isotopes since the lava emplacement. This
Pb isotopic composition 9l1=9.28‹0.22) with the in turn suggests that the calculated time-integrated
isochron extending well above the mantle evolution 238U/204Pb 9i.e., l1 values) and Th/U ratios re¯ect those of
curve. In the 206Pb/204Pb vs. 208Pb/204Pb space, the data the corresponding erupted lavas and are not an alteration
from all three localities also de®ne linear trends 9Fig. 3). artifact. Therefore, the U±Th±Pb isotope data were fur-
Calculated from these data, 232Th/238 U ratios are 4.8 ther used for geochemical modeling.
for Golets ¯ow 1 9samples of ¯ow-top breccia 8978,
The obtained Pb isotope compositions correlate well
which plot well below the regression line, not included), with the Nd isotope and trace element parameters of the
4.4 for Golets ¯ow 3, and 4.6 for the lava lake. These lavas from the three localities reported by Puchtel et al.
Th/U ratios are basically identical to those calculated 91997) and shown in Fig. 4 and Table 3. The relationships
from the trace element data 9Puchtel et al. 1997) and are between trace element abundances and Nd isotopes in
signi®cantly higher than the Th/U ratio of 3.79 in the these komatiitic basalts were proposed by these authors to
contemporary mantle estimated using the evolution be the result of mixing of material of the upper continental
curve of Collerson and Kamber 91999). The large errors crust and an incompatible element-depleted komatiitic
on ages and Pb-initial isotopic compositions are magma with ꢀ18% MgO. Upper continental crust has a
attributed to a very limited range in Pb isotopic ratios substantially higher U/Pb ratio compared to the mantle,
obtained.
and with time develops a much more radiogenic Pb

594
Fig. 4 Variation diagrams of
selected lithophile trace element
and isotope ratios. Pb isotope
data combined from this study
and from Puchtel et al. 91997);
other data and the calculated
original melt composition from
Puchtel et al. 91997)
isotope composition, manifested by higher l1 values. As trace element composition of the original komatiite melt
can be seen from the diagrams, there is a positive corre- parental to the Vetreny belt komatiitic basalts, which was
lation between l1 and La/Sm, Zr/Y 9both of which are evaluated by Puchtel et al. 91997) and also in this study on
high in the upper crust) and a negative correlation the basis of the correlations between trace elements and
between l1 and the initial ꢀ143Nd 9the latter is low in the Nd±Pb isotopes. Several ratios, including Eu/Eu*=1
upper crust). These relationships indicate a coupled 9uncontaminated komatiite magma should not have a
behavior of the Pb-Nd isotopes and trace elements in the Eu-anomaly), l1=8.75 9a mantle value) and Th/U=3.79
Vetreny belt komatiitic basalts. Also plotted in Fig. 4 is the 9the ratio in the contemporary mantle), were adopted to
Table 3 Parameters and esti-
Parameter
ꢀNd9T) l1
Th/U
Eu/Eu* 9La/Sm)N Zr/Y
c187Os9T)
j 9%)
mates used in modeling, and
model results. Trace element
and Nd isotope data from
Puchtel et al. 91997). Degree of
contamination with upper
continental crust 9j) from
Puchtel et al. 91997)
Golets, ¯ow 1
Golets, ¯ow 3
Lava lake
Original komatiite 2.6
melt
±1.46
0.27
±0.91
9.28
4.8
4.4
4.6
3.8
0.934
0.957
0.947
1.00
2.51
1.89
2.13
0.90
4.39
3.53
3.92
2.22
±0.06
±0.43
±0.07
±0.9
10
5
8
9.05
9.19
8.75
0

595
represent a primary komatiite composition. These ratios Re/Os ratio and with time develops radiogenic
were then used to calculate other incompatible element 187Os/188Os. Therefore, in a suite of crustally contami-
and Nd isotope ratios in the primitive komatiite magma, nated rocks, a correlation between the degree of Os
which were compared with the published data on pre- isotope enrichment and the degree of crustal contami-
sumably uncontaminated komatiites. Comparisons re- nation is to be expected.
vealed a close similarity, and the estimates were further
Plotted in Fig. 6 are variations of Nd±Pb isotope and
used to address the issue of the initial Os isotope com- trace element parameters of the lavas versus Os isotope
position of the Vetreny mantle source.
compositions. A positive correlation between c187Os and
l1, Zr/Y, La/Sm, and Th/U, and an inverse correlation
between c187Os and ꢀ143Nd and Eu/Eu* indicate that the
variations in Os isotope compositions between the dif-
ferent localities in the Vetreny belt are likely due to the
di€erences in the degrees of crustal contamination.
Knowing the degrees of contamination from lithophile
element systematics and the ratios of lithophile elements
in the initial komatiite magma, it is possible to estimate
the Os isotope composition of the original non-con-
taminated komatiite melt from the regressions in Fig. 6,
using the parameters mentioned above. Calculations
give consistent results averaging at c187Os9T) of ±0.9‹
0.1. It must be noted that this estimate depends on the
Os isotope composition of the end member with the
highest degree of contamination, i.e., Golets ¯ow 1,
which is in turn based entirely on a single analysis of a
chromite separate from this ¯ow. However, a good
agreement between the results in Fig. 6 indicates that
this estimate of c187Os9T) is fairly robust.
In order to further test the proposed model, we
consider the potential theoretical e€ect of crustal con-
tamination on the Os isotope composition of mantle
magmas. Estimates of Re and Os abundances in crustal
rocks have become available only recently; for the
purposes of modeling, the compositions of the upper
continental crust from Schmidt et al. 91997) 9referred
to as UCC-1) and Peucker-Ehrenbrink and Jahn
91999) 9referred to as UCC-2) were used 9Table 4). The
upper continental crust in this part of the Karelian
granite±greenstone terrane consists of the Vodla Block
tonalite±trondhjemite±granodiorite gneisses with ages of
3.1±3.2 Ga 9Lobach-Zhuchenko et al. 1993). This means
that the upper crust had ꢀ700 Ma to evolve with the
suprachondritic Re/Os ratio before being digested by the
komatiite magma of the Vetreny plume. As a possible
estimate of Re and Os abundances in a primitive kom-
atiite melt giving rise to the Vetreny belt lavas, we chose
those in the Munro Township komatiite, known to have
a chondritic initial Os isotopic composition 9Walker
et al. 1988; Shirey 1997). The results of these calculations
are summarized in Fig. 7. It has been recognized earlier
that crustal contamination had a very substantial e€ect
on the lithophile element abundances in the Vetreny belt
magmas even at the relatively low degrees of contami-
nation, estimated to be ꢀ5% for Golets ¯ow 3, ꢀ8% for
the lava lake, and ꢀ10% for Golets ¯ow 1 9Puchtel et al.
1997). At the same time, as is evident from Fig. 7, the
addition of 10% upper continental crust will have a
minor e€ect on the Os isotope composition in the initial
komatiite magma, increasing it by only 0.8±1.2 c187Os
units. The data obtained in this study agree well with the
Re±Os age of lavas and initial Os isotopic composition
of the Vetreny plume source
The Re±Os ages of the Vetreny belt lavas are in good
agreement with the data reported from the lithophile
element isotope systems. This implies that either the Re±
Os system remained closed after the lava eruption or it
was disturbed during or shortly after the lava emplace-
ment and stayed intact ever since. The fact that the Re
data 9except for one sample from Golets ¯ow 3) from
both localities plot on the olivine control lines as illus-
trated in Fig. 5, and, thus, exhibit the magmatic be-
havior, strongly favors the former interpretation. The
relative enrichment of the ¯ow 3 melt in Re compared to
that of the lava lake correlates well with the elevated
concentrations of several other incompatible elements,
such as TiO2 and CaO, in this ¯ow, which was inter-
preted by Puchtel et al. 91997) to re¯ect slight di€erences
in the degrees of partial melting of the respective mantle
source regions. Thus, the Os isotopic compositions of
the lavas can potentially be used to constrain the initial
Os isotope composition of the Vetreny plume source.
The main obstacle here to overcome is the need to ac-
count for the Os component introduced with the conti-
nental crustal material.
Due to more incompatible behavior of Re compared
to Os during partial melting, continental crust has a high
Fig. 5 MgO vs. Re variation diagram for Golets ¯ow 3 and lava
lake komatiitic basalts. The linear trends correspond to olivine
control lines for respective lava suites. Major element composition
of olivines from Puchtel et al. 91997). Olivine from the lava lake is
assumed to have Re abundances similar to those in olivine from
¯ow 3 9Table 1)

596
Fig. 6 Variation diagrams of
c187Os9T) vs. selected averaged
lithophile trace element and
isotope ratios in the three lava
suites used to estimate the Os
isotope composition of the
original komatiite melt
Table 4 Re-Os abundances in upper continental crust and the original komatiite. Re±Os abundances for UCC-1 and UCC-2 from
Schmidt et al. 91997) and Peucker-Ehrenbrink and Jahn 91999)
Parameter
Re 9ppb)
Os 9ppb)
187Re/188Os
187Os/188Os
c187Osa
700 Ma old UCC-1
700 Ma old UCC-2
Original komatiite melt
0.400
0.237
1.06
0.030
0.031
1.51
62.8
36.0
3.30
0.8585
0.5444
0.1258
576
329
±0.9
a Os isotopic composition at the time of mixing, after allowing the upper continental crust to evolve for 700 Ma
model calculations, as the hybrid melts from the three Depletion history of the Vetreny mantle source
localities plot within the envelope de®ned by the
two upper crustal estimates in Fig. 7. We conclude, Puchtel et al. 91997) have established that the primary
therefore, that the Vetreny plume came from a mantle komatiite magma that gave rise to the Vetreny belt ko-
source with a slightly subchondritic Os isotopic com- matiitic basalts was depleted in LREE relative to HREE,
position 9c187Os9T)=)0.9), which thus evolved with a and in Zr relative to Y, indicating that the Vetreny
long-term slightly subchondritic Re/Os ratio.
mantle source was depleted in more incompatible ele-

597
150 Ma. Vetreny lavas contained between 0.5 and
0.8 ppb of Re during eruption 9Fig. 5), which scales
down to 0.4±0.65 ppb of Re in a primary komatiite
magma after accounting for assimilation±fractional
crystallization processes en route to the surface. This is
approximately two times lower than the Re abundance
this primary komatiite melt would have had if it were
derived from a source with a primitive mantle Re con-
tent of 0.26 ppb 9Morgan 1986) assuming a degree of
melting of ꢀ30% 9Puchtel et al. 1997). This in turn
implies that the Vetreny source retained ꢀ50% of its
Re-inventory after the previous depletion events, and,
thus, the time of Re depletion must be older than the
estimate from Eq. 94). It is very likely that it occurred
during the major late Archean crust-forming events in
the region that took place between 3.1 and 2.9 Ga
9Kulikov et al. 1990; Lobach-Zhuchenko et al. 1993;
Puchtel et al. 1999b).
Fig. 7 Diagram illustrating the theoretical e€ect of addition of
upper continental crust material 9UCC) on the Os isotope
composition of the komatiitic magma parental to the Vetreny
komatiitic basalt. Composition of UCC-1 and UCC-2 from
Schmidt et al. 91997) and Peuker-Ehrenbrink and Jahn 91999),
respectively. Percentage of contamination from Puchtel et al. 91997)
Implications for the early accretion history of the Earth
Several lines of evidence suggest that the source of the
Vetreny komatiitic basalt resided in the upper mantle.
These include a relatively low potential temperature of
the source coupled with a moderately high MgO content
in the original komatiite magmas and their Al-unde-
pleted nature 9Puchtel et al. 1997), implying that the
magma separated from the mantle residue in the spinel
peridotite stability ®eld, e.g., at depths <200 km 9e.g.,
Herzberg 1995). This source was similar in Os isotope
composition to several late Archean plume sources, in-
cluding those that gave rise to Munro Township
9Walker et al. 1988; Shirey 1997) and Kambalda kom-
atiites 9Foster et al. 1996), which also evolved with long-
term near-chondritic Re/Os ratios. Because Re is one of
the most incompatible highly siderophile elements dur-
ing mantle melting, a near-chondritic Re/Os ratio pro-
vides strong evidence for the chondritic relative
abundances of other highly siderophile elements in the
late Archean upper mantle in this region.
Chondritic relative abundances of HSE in the mantle
are apparently not what would be expected from the
available data on the metal±silicate partitioning, which
would predict substantially lower absolute abundances
and suprachondritic Re/Os and Pt/Os ratios as a result
of segregation of the iron metal into the core during the
Earth's accretion 9e.g., Borisov et al. 1994; Borisov and
Palme 1995; O'Neill et al. 1995). This discrepancy led
some workers to propose the hypothesis of heteroge-
neous accretion 9Kimura et al. 1974; Morgan 1985),
whereby the excess highly siderophile elements were
added to the Earth's upper mantle after core segregation
was completed and in the amount that would entirely
overwhelm any earlier core±mantle equilibrium signa-
ture. The new Re±Os data provide further evidence that,
by the turn of the Paleoproterozoic, the Earth's mantle
ments relative to less incompatible ones as a result of
previous episodes of melt extraction. A positive ꢀNd9T)
value of +2.6 implies that the depletion event had oc-
curred at least a few hundreds of millions of years prior
to the formation of the Vetreny belt. The Os isotope
data provide new evidence to support this observation.
It has been long established that during mantle melt-
ing, Os is compatible with the mantle residue, while Re is
highly incompatible. For instance, Walker et al. 91988)
estimated a mantle residue/komatiitic liquid bulk distri-
bution coecient for Os and Re of 1.9 and 0.001, re-
spectively. This implies that during mantle melting, Re is
e€ectively removed from the mantle and is concentrated
in crustal materials, while Os is accumulated in mantle
residues. This is best illustrated by the Os isotope syste-
matics of the subcontinental lithospheric mantle
9SCLM), which is thought to be the residual part of the
upper mantle, underlying continents, that lost its basaltic
component during multiple episodes of melt extraction.
During the melt extraction, the SCLM becomes in-
creasingly depleted in Re and enriched in Os and,
thus, with time evolves with nonradiogenic Os isotope
composition 9e.g., Walker et al. 1989). To characterize
the minimum time of Re-depletion, these workers intro-
duced the TRD value, or the minimum time of Re-de-
pletion, which is calculated assuming that all the Re was
removed from the mantle during the melting event and,
thus, the Re/Os of the residual mantle approached zero:
2
3
ꢀ
ꢁ
ꢀ
ꢁsample
CH À
187
188
187
188
Os
Os
Os
Os
1
6
4
7
5
i
i
TRD
ˆ
Â
‡ 1
ꢁ4†
ꢀ
ꢁCH
187
188
k
Re
Os
Using this equation, the minimum time of the Re was already well homogenized 9as evidenced by uni-
depletion of the Vetreny mantle source is estimated at form chondritic Re/Os in several localities worldwide

598
Collerson KD, Kamber BZ 91999) Evolution of the continents and
the atmosphere inferred from Th±U±Nb systematics of the
depleted mantle. Science 283: 1519±1522
Dupre B, Arndt NT 91990) Pb isotopic composition of Archean
komatiites and sul®des. Chem Geol 85: 35±56
Foster JG, Lambert DD, Frick LR, Maas R 91996) Re±Os isotopic
evidence for genesis of Archaean nickel ores from uncontami-
nated komatiites. Nature 382: 703±706
Gaal G, Gorbatschev R 91987) An outline of the Precambrian
evolution of the Baltic Shield. Precambrian Res 35: 15±52
Hartmann WK 91970) Lunar cratering chronology. Icarus 13:
299±301
Heaman LM 91997) Global ma®c magmatism at 2.45 Ga: remnants
of an ancient large igneous province? Geology 25: 299±302
Herzberg C 91995) Generation of plume magmas through time: an
experimental perspective. Chem Geol 126: 1±16
Kimura K, Lewis RS, Anders S 91974) Distribution of gold and
rhenium between nickel±iron and silicate melts; implications for
abundance of siderophile elements on the earth and moon.
Geochim Cosmochim Acta 38: 683±701
Kulikov VS, Simon AK, Kulikova VV, Samsonov AV, Kajrjak AI,
Ganin VA, Sudin AI 91990) Evolution of Archaean magmatism
in the Vodla Block, Karelian granite±greenstone terrane. In:
Lobach-Zhuchenko SB, Bibikova EV9eds) Precambrian geol-
ogy and geochronology of the East European Platform. Nauka,
Leningrad, pp 92±100
Lobach-Zhuchenko SB, Chekulaev VP, Sergeev SA, Levchenkov
OA, Krylov IN 91993) Archaean rocks from southeastern
Karelia 9Karelian granite±greenstone terrain). Precambrian Res
62: 375±397
mentioned above) with respect to the highly siderophile
elements, which were likely to be added during the ac-
cretion of the late veneer. It is noteworthy that the lunar
cratering record indicates that the last phase of heavy
bombardment ended between 3.8 and 3.5 Ga 9Hart-
mann 1970; Chyba 1991; ), i.e., a billion years prior to
the formation of the Abitibi±Kambalda and Vetreny
belt lavas. This implies that the late veneer was ho-
mogenized throughout the upper mantle on a time scale
of ꢀ1 Ga.
Conclusions
1. The e€ect of continental crustal contamination on
Os isotopic composition of komatiite magmas is limited
and can be monitored and corrected for by combining
Re±Os studies with Pb±Nd isotope and lithophile trace
element data.
2. Vetreny belt parental komatiite magmas were de-
rived from an upper mantle plume source with a slightly
subchondritic initial Os isotopic composition 9c187Os-
9T)=±0.9). The data indicate that the mantle source was
depleted in Re at least 150 Ma before the formation of
the Vetreny belt as a result of previous melt-extraction
episodes. The results provide further evidence that, by
the end of the late Archean, the Earth's upper mantle
was well homogenized with respect to highly siderophile
elements, added during the accretion of the late veneer,
on a time scale of ꢀ1 Ga.
Ludwig KR 91992) ISOPLOT ± a plotting and regression program
for radiogenic isotope data, version 2.57. US Geol Surv Open-
File Rep 91±445
Ludwig KR 91997) ISOPLOT ± a plotting and regression program
for radiogenic isotope data, version 2.95. US Geol Surv Open
File Rep 91±445
Manhes G, Minster JF, Allegre CJ 91978) Comparative uranium±
thorium±lead and rubidium±strontium of Saint Severine
amphoterite: consequences for early solar system chronology.
Earth Planet Sci Lett 39: 14±24
Acknowledgements We thank V. Kevlich for obtaining mineral
separates and handpicking. The manuscript greatly bene®ted from
thorough and constructive reviews by Eero Hanski and Al Bran-
don, and from comments by Munir Humayun and Andy Campbell.
This study was supported by the Alexander von Humboldt and
Max-Planck Institut postdoctoral fellowships to IP and by the
funds from the MPI fur Chemie and The University of Chicago.
Meisel T, Walker RJ, Morgan JW 91996) The osmium isotopic
composition of the Earth's primitive upper mantle. Nature 383:
517±520
Morgan JW 91985) Osmium isotope constraints on Earth's late
accretionary history. Nature 317: 703±705
Morgan JW 91986) Ultrama®c xenoliths: clues to Earth's late
accretionary history. J Geophys Res 91: 12375±12387
Morgan JW, Walker RJ 91989) Isotopic determinations of rhenium
and osmium in meteorites by using fusion, distillation and
ionexchange separations. Anal Chim Acta 222: 291±300
O'Neill HSC, Dingwell DB, Borisov A, Spettel B, Palme H 91995)
Experimental petrochemistry of some highly siderophile ele-
ments at high temperatures, and some implications for core for-
mation and the mantle's early history. Chem Geol 120: 255±273
Peucker-Ehrenbrink B, Jahn B-M 91999) PGE-concentrations
and Os-isotopic ratios in loess: a proxy for the eroding upper
continental crust? In: 9th V.M. Goldschmidt Conf, Abstr,
Cambridge, MA, pp 224±225
Puchtel IS, Hofmann AW, Mezger K, Shchipansky AA, Kulikov
VS, Kulikova VV 91996) Petrology of a 2.41 Ga remarkably
fresh komatiitic basalt lava lake in Lion Hills, Central Vetreny
Belt, Baltic Shield. Contrib Mineral Petrol 124: 273±290
Puchtel IS, Haase KM, Hofmann AW, Chauvel C, Kulikov VS,
Garbe-Schonberg C-D, Nemchin AA 91997) Petrology and
geochemistry of crustally contaminated komatiitic basalts from
the Vetreny Belt, southeastern Baltic Shield: evidence for an
early Proterozoic mantle plume beneath rifted Archean conti-
nental lithosphere. Geochim Cosmochim Acta 61: 1205±1222
Puchtel IS, Brugmann GE, Hofmann AW 91999a) Precise Re±Os
mineral isochron and Pb±Nd-Os isotope systematics of a ma®c±
ultrama®c sill in the 2.0 Ga Onega plateau 9Baltic Shield). Earth
Planet Sci Lett 170: 447±461
References
Amelin YV, Heaman LM, Semenov VS 91995) U±Pb geochronol-
ogy of layered ma®c intrusions in the eastern Baltic Shield:
implications for the timing and duration of Paleoproterozoic
continental rifting. Precambrian Res 75: 31±46
Arndt NT, Todt W 91994) Formation of 1.9-Ga-old Trans-Hudson
continental crust: Pb isotopic data. Chem Geol 118: 9±26
Birck JL, Roy-Barman M, Capman F 91997) Re±Os isotopic
measurements at the femtomole level in natural samples.
Geostandards News 20: 19±27
Borisov A, Palme H 91995) The solubility of iridium in silicate
melts: new data from experiments with Ir10Pt90 alloys. Geochim
Cosmochim Acta 59: 481±485
Borisov A, Palme H, Spettel B 91994) Solubility of Pd in silicate
melts: implications for core formation in the Earth. Geochim
Cosmochim Acta 58: 705±716
Brandon AD, Snow JE, Walker RJ, Morgan JW, Mock TD 92000)
190Pt±186Os and 187Re±187Os systematics of abyssal peridotites.
Earth Planet Sci Lett 177: 319±335
Chyba CF 91991) Terrestrial mantle siderophiles and the Lunar
impact record. Icarus 92: 217±233

599
Puchtel IS, Hofmann AW, Amelin YV, Garbe-Schonberg C-D, Shirey SB, Walker RJ 91998) The Re±Os isotope system in cos-
Samsonov AV, Shchipansky AA 91999b) Combined mantle mochemistry and high-temperature geochemistry. Annu Rev
plume ± island arc model for the formation of the 2.9 Ga Earth Planet Sci 26: 423±500
Sumozero±Kenozero greenstone belt, SE Baltic Shield: isotope Snow JE, Reisberg L 91995) Os isotopic systematics of the MORB
and trace element constraints. Geochim Cosmochim Acta 63:
3579±3595
mantle: results from altered abyssal peridotites. Earth Planet Sci
Lett 136: 723±733
Roy-Barman M, Allegre CJ 91994) 187Os/186Os ratios of mid-ocean Sokolov VA 91987) Geology of Karelia. Nauka, Leningrad
ridge basalts and abyssal peridotites. Geochim Cosmochim Tatsumoto M, Knight RJ, Allegre CJ 91973) Time di€erences in the
Acta 58: 5043±5054 formation of meteorites as determined from the ratio of lead
Schiano P, Birck J-L, Allegre CJ 91997) Osmium±strontium±neo- 207 to lead 206. Science 180: 1279±1283
dymium±lead isotopic covariations in mid-ocean ridge basalt Tsuru A, Walker RJ, Kontinen A, Peltonen P, Hanski E 92000)
glasses and the heterogeneity of the upper mantle. Earth Planet Re±Os isotopic systematics of the 1.95 Ga Jormua Ophiolite
Sci Lett 150: 363±379 Complex, northeastern Finland. Chem Geol 164: 123±141
Schmidt G, Palme H, Kratz K-L 91997) Highly siderophile ele- Walker RJ, Shirey SB, Stecher O 91988) Comparative Re±Os, Sm±
ments 9Re, Os, Ir, Ru, Rh, Pd, Au) in impact melts from three
European impact craters 9Saaksjarvi, Mien, and Dellen): clues
to the nature of the impacting bodies. Geochim Cosmochim
Acta 61: 2977±2987
Nd and Rb±Sr isotope and trace element systematics for
Archean komatiite ¯ows from Munro Township, Abitibi belt,
Ontario. Earth Planet Sci Lett 87: 1±12
Walker RJ, Carlson RW, Shirey SB, Boyd FR 91989) Os, Sr, Nd,
and Pb isotope systematics of southern African peridotite
xenoliths: implications for the chemical evolution of subconti-
nental mantle. Geochim Cosmochim Acta 53: 1583±1595
Shirey SB 91997) Initial Os isotopic composition of Munro
Township, Ontario komatiites revisited: additional evidence for
near-chondritic, late-Archaean convecting mantle beneath the
Superior Province. In: 7th V.M. Goldschmidt Conf, Abstr, Walker RJ, Hanski E, Vuollo J, Liipo J 91996) The Os isotopic
Tucson, AZ, p 193
Shirey SB, Barnes S-J 91994) Re±Os and Sm±Nd isotopic con-
straints on basaltic komatiitic volcanism and magmatic sul-
composition of Proterozoic upper mantle: evidence for chon-
dritic upper mantle from the Outokumpu ophiolite, Finland.
Earth Planet Sci Lett 141: 161±173
phide formation in the Cape Smith Foldbelt, Quebec. Mineral York D 91969) Least squares ®tting of a straight line with
correlated errors. Earth Planet Sci Lett 5: 320±324
Mag 58A: 535±536
Shirey SB, Walker RJ 91995) Carius tube digestion for low-blank
rhenium±osmium analysis. Anal Chem 67: 2136±2141