∼ 3710 and ⪖ 3790 Ma volcanic sequences in the Isua (Greenland) supracrustal belt; structural and Nd isotope implications

∼ 3710 and ⪖ 3790 Ma volcanic sequences in the Isua (Greenland) supracrustal belt; structural and Nd isotope implications

CHEMICAL GEOLOGY INCLUOIICG ELSEVIER ISOTOPE GEOSCIENCE Chemical Geology 141 (1997) 271-287 ~ 3710 and > 3790 Ma volcanic sequences in the Isua (G...

1MB Sizes 1 Downloads 11 Views

CHEMICAL GEOLOGY INCLUOIICG

ELSEVIER

ISOTOPE GEOSCIENCE

Chemical Geology 141 (1997) 271-287

~ 3710 and > 3790 Ma volcanic sequences in the Isua (Greenland) supracrustal belt; structural and Nd isotope implications Allen P. Nutman a,*, Vickie C. Bennett a, Clark R.L. Friend b, Minik T. Rosing c,d a Research School of Earth Sciences, Australian National University, Canberra, A.C.T. 0200, Australia b Department of Geology, Oxford Brookes University, Headington, Oxford, OX30BP, UK c Geological Museum, Oster Voldgade 5-7, DK-1350 Copenhagen, Denmark d Danish Lithosphere Centre, Oster Voldgade 10, DK-1350 Copenhagen, Denmark

Received 3 October 1996; accepted 17 June 1997

Abstract The Isua supracrustal belt of southern West Greenland is the largest body of early Archaean supracrustal rocks, and as such has been the focus of numerous studies to characterise early Earth processes. Based on new geochronological constraints we re-interpret the Isua supracrustal belt stratigraphy and re-evaluate Nd isotopic constraints for the early Earth from Isua supracrustal belt rocks. Part of the belt ( ~ 50%) consists of a package of mafic chloritic schists (the garbenschiefer unit), layered amphibolites and felsic rocks, mica ( ___kyanite) schists and banded iron formation (BIF), and is interpreted as a sequence of mafic to felsic volcanic/volcanosedimentary rocks, turbidites and pelites with minor gabbro. Graded felsic volcanics in this package have yielded a U - P b zircon age of ~ 3710 Ma. Along the southern side of the belt, a package of amphibolites and ultramafic schists with BIF layers, interpreted as a mafic to ultramafic volcanic sequence, is cut by tonalite sheets with U - P b zircon ages of 3790-3800 Ma. Thus the belt contains at least two unrelated supracrustal packages of different ages ( > 3790 Ma and ~ 3710 Ma), that were probably juxtaposed in the early Archaean. Felsic volcanic rocks from a single ~ 3710 Ma unit have consistent initial eNd values of +0.8 to + 1.8, suggesting minimal fractionation of Nd from Sm during later metamorphism. They may have formed from a depleted (but not ultra-depleted) mantle source only a short time before 3710 Ma, with or without a contribution from LREE-enriched crust of 3750-3870 Ma with ~Na(380o) of + 2 to +4. There is debate as to whether the 3806 ___2 Ma 'A6' felsic unit with 6Nd(3S00) of + 1.0 to + 2.9 is derived from volcanic rocks or from granitoid sheets. Thus in the published Isua supracrustal belt Nd database there are no samples which are definitely > 3710 Ma supracrustal or clearly mantle-derived rocks. Given the problems of locally intense alteration, the presence of supracrustal rocks of two or more ages and of older inherited crustal components in some samples, the Isua supracrustal belt is not the ideal locality to find definitive answers on early Archaean mantle evolution. © 1997 Elsevier Science B.V. Keywords: neodymium isotopes; Isua supracrustals; mantle chemistry; Archaean; geochronology; West Greenland

* Corresponding author Fax: + 61 (6) 249-0738; e-mail: [email protected] 0009-2541/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PII S0009-2541 (97)00084-3

272

A.P. Nutman et al. / Chemical Geology 141 (1997) 271-287

I, Introduction Rocks older than 3700 Ma are extremely rare. Aside from abundant exposures in West Greenland and northern Labrador, they are only known from limited outcrops of the Acasta area, Northern Territories (Bowring et al., 1989), the Sino-Korean Craton (Song et al., 1996), the Mt. Sones area, Antarctica (Black et al., 1986) and the Narryer Gneiss Complex of Western Australia (Myers, 1988; Nutman et al., 1991). Supracrustal rocks of this age are even rarer (see table 1 of Nutman et al., 1996). The largest body of early Archaean supracrustal rocks is a diverse package known as the Isua supracrustal belt (Fig. 1), within the early Archaean Itsaq Gneiss Complex of southern West Greenland (Bridgwater and McGregor, 1974; Nutman et al., 1984, 1996; Rosing et al., 1996). Geochemical, isotopic and lithologic data from this belt and other adjacent early Archaean rocks have been used extensively in general models for early Archaean tectonic and crustal growth processes (e.g., Nutman and Collerson, 1991; Maruyama et al., 1992; Rosing et al., 1996). Although well-exposed, the belt has suffered several episodes of deformation, metasomatism and metamorphism at amphibolite and greenschist facies conditions (e.g., Nutman et al., 1984). The dominant lithologies are mafic rocks along with banded iron formations (BIF) and ultramafic rocks, but felsic rocks are also present and have been the focus of geochronologic studies. There is presently debate about the age and nature of the Isua supracrustal belt. For example, opinions have been expressed that the Isua supracrustal belt is not one stratigraphic succession, but is composed of more than one tectonic unit, with the possibility that some components might be older than 3850 Ma (Rosing et al., 1996). Nutman et al. (1996) suggested that successions with ages of > 3790 and 3710 Ma are present, and Hamilton (1993) suggested the supracrustal rocks are late Archaean in age. The age of the units significantly affects the inferences about early crust-mantle evolution that can be made based on Nd isotopic data as well as tectonic models of early Archaean crust formation processes. Using new geochronological data presented in this paper, we re-evaluate the age(s), assembly history and degree of metasomatic alteration of rocks from the belt, in

~'- ~,:

/

~

~

__

,

'

~

~ ~.

'...;

A 3 kyanite

~,I~

L~ 37-07!-6 I

belt

IP'-~a I I~ ~"Idi

i I[~garbenschiefer unit •

~

~

.

_

other supracrusta rocks ~odgin

(3806-5.2 Ma)

I,~'~\'~----~ ~ I ( , ' ~ - ~

L" ~ l ~ - ~ _ k J

I granodlonte g r_a, " "

I >-3570, < 3 6 5 0 " t i iv C tutus econta t

-'P -~'~'~

1 km

Fig. 1. Geological maps of parts of the Isua supracrustal belt. Ages shown with asterisks are from Compston et al. (1986), Nutman and Collerson (1991), Nutman et al. (1995, 1996) and Nutman (unpublished data).

273

A.P. Nutman et al. / Chemical Geology 141 (1997) 271-287

order to contribute., to the understanding of early Archaean tectonic processes and crustal-mantle evolution by Nd isotopic tracing (e.g., Jacobsen and Dymek, 1987; Harper and Jacobsen, 1992; McCulloch and Bennett, 1993, 1994; Moorbath et al., 1997).

2. lsua supracrustal belt 2.1. Lithologies

The Isua supracmstal belt has a 'stratigraphy' of lithological units w]hich was mapped throughout the belt by Nutman et al. (1984), who recognised that the contacts between some units are tectonic, and speculated about how the different packages might be related. However, in the 1980s the available geochronological data were not sufficient to be able to demonstrate whether packages of supracrustal rocks differing in age by < 100-200 Ma are present. Therefore, all the supracrustal rocks in the belt were taken to belong to one chronostratigraphic sequence, but somewhat disturbed by early tectonic imbrication. Widespread metasomatism in the belt has modified many lithologies and in some cases causes difficulties in identifying protoliths (Gill et al., 1981; Nutman et al., 1984; Rosing and Rose, 1993a; Rose et al., 1996). Detailed case histories and investigation of the mechanism of metasomatism have been presented by Rosing et al. (1996). However, by careful sampling, material that has suffered minimal alteration in younger events (hence the best candidates for minimal fractionation of daughter-parent isotopes) can be found (e.g., Jacobsen and Dymek, 1987). Unequivocal supracrustal rocks in the belt are mafic volcanic rocks, gabbros, ultramafic rocks, turbidites, cherts and BIF (Nutman et al., 1984; Jacobsen and Dymek, 1987; Dymek and Klein, 1988; Rosing et al., 1996). Of contentious origin are some siliceous and felsic rocks (discussed in this paper), calc-silicate rocks and marbles (Nutman et al., 1996; Rosing et al., 1996). Some mica schists, and at least some and possibly all calc-silicate rocks and marbles are metasomites and not sediments (Rosing and Rose, 1993b; Rosing et al., 1996). Because of debate over the origin of such lithologies, samples from these units are not discussed further.

2.2. Previous geochronologic constraints

Nd-Sm, Rb-Sr and Pb-Pb errorchrons on rocks from the Isua supracrustal belt initially demonstrated its great antiquity ( > 3700 Ma) but with large uncertainties, on the order of 100 Ma (e.g., Moorbath et al., 1973, 1977, 1986, 1997; Hamilton et al., 1983; Jacobsen and Dymek, 1987). The most precise age, a U - P b zircon age of 3806 ___2 Ma, was obtained by SHRIMP (Compston et al., 1986) on the A6 felsic unit (Fig. 1c). Further reconnaissance SHRIMP U - P b zircon results show that a tonalite sheet cutting amphibolites in the southwest of the belt has an age of 3791 + 4 Ma, and that elsewhere in the belt there are some felsic volcanic rocks as young as ~ 3710 Ma (Fig. 1; Nutman et al., 1996). Thus the widely cited 3806 + 2 Ma date from unit A6 should not be taken as the 'age' of the Isua supracrustal belt, because although some supracrustal rocks in the belt could be this old, it is now evident that some are ~ 100 Ma younger. The new data presented in this paper confirm this finding, and further constrains the rocks belonging to a 'young' ( ~ 3710 Ma) sequence and an 'old' ( > 3790 Ma) sequence.

3. Recognition of supracrustal rocks

~ 3710

and

> 3790

Ma

3.1. S H R I M P U - P b zircon technique and data assessment

U - T h - P b isotopic ratios and concentrations were determined in zircon separates using SHRIMP I and were referenced to the Australian National University standard zircon SL13 (572 Ma; 2°6pb/238U = 0.0928). Further details of the analytical procedure and data assessment are given by Compston et al. (1984), Roddick and van Breemen (1994) and Claou6-Long et al. (1995). Ages presented in this paper are weighted means (2o') derived from 2°7pb/2°6pb ratios of analyses from the isotopically least-disturbed sites (those having close to concordant U / P b ages, with lowest 2°4pb contents) in grains which from optical microscopy were ascertained to belong to single populations. In a few cases, maximum likelihood components analysis, a statistical technique for assessing the number of pop-

274

A.P. Nutman et al. / Chemical Geology 141 (1997) 271-287

ulations present in a set of data (Sambridge and Compston, 1994) was undertaken as an alternative method of assessing the age data. 3.2.

~ 3710

Ma

age

for

the

garbenschiefer-

dominated package

The garbenschiefer unit forms approximately a third of the belt (Fig. 1) and consists of magnesian mafic rocks. The garbenschiefer unit has been used extensively for isotope studies (e.g., Hamilton et al., 1978; Gruau et al., 1996) but much of it is a chlorite + amphibole ___carbonate schist, and thus has probably suffered secondary geochemical alteration. As such it is a problematical candidate for isotope studies to ascertain the age of the belt and constrain evolution of the early Archaean mantle. The garbenschiefer unit was originally interpreted as a large, highly altered, gabbroic sill (e.g., Gill et al., 1981; Nutman et al., 1984). Relict gabbroic textures are locally preserved, but elsewhere weakly deformed pillow structures have been found (S. Maruyama, pers. commun., 1993). Garbenschiefer homogeneous mafic schists are concordantly interlayered with (Nutman et al., 1984), and perhaps grade into (Rosing et al., 1996), sedimentary units with well preserved Bouma sequence structures, and are found with kyanite-rich schists and thin layers of chemical sediment. The garbenschiefer-dominated package is currently interpreted as a pile of volcanic rocks with associated volcaniclastic sediments and minor amounts of gabbro (Rosing and Rose, 1993b; Nutman et al., 1996; Rosing et al., 1996). Zircon geochronology was not attempted on the garbenschiefer unit itself, but on rocks associated with it. Two samples of felsic rocks (from the northwest and northeast ends of the belt) close to the garbenschiefer with graded layering and of intermediate silica content ( ~ 60 wt%) yielded no zircons. The one from the northeast of the belt was from the western margin of the B I f e l s i c f o r m a t i o n . B 1 (Fig. lb) is a layered unit of quartz + plagioclase + biotite + muscovite + garnet rocks interrupted by mica + garnet schist and siliceous layers. Graded layering is quite common in this unit, and indicates that it faces east (Nutman et al., 1984). Towards its eastern margin (top) the felsic rocks are more siliceous ( > 70 wt%) and show graded units up to 8 m thick, all

fining upwards with their bases crowded with finegrained felsic clasts (Fig. 2). Graded felsic units follow each other directly or are separated by thin mica schist layers and are associated with tourmalinites (Appel, 1983). These rocks are interpreted as volcanic or volcanosedimentary in origin, perhaps pyroclastic flow deposits. Zircon data from a fine-grained, homogeneous, flow-top sample were reported by Nutman et al. (1996). The zircons were small ( < 150 ~m) stubby prismatic zircons with a pale brown colour and in some cases distinct euhedral zoning. They have neither obvious inherited cores nor overgrowths. Most of the analyses are concordant within uncertainty. Assuming only minor amounts of ancient Pb-loss a weighted mean 2°7pb/~°6pb age of 3708-t-3 Ma was obtained by rejecting a small proportion of the analyses with lowest 2°7pb/2°6Pb. In addition, zircon geochronology was undertaken on a clast-rich sample (G93-75) from the base of the overlying graded unit, to examine the age of inherited zircons in the clasts. Zircons in this sample have the same morphology as those in the homogeneous sample. One notable characteristic of these zircons is that large, branching fluid inclusions are common (Fig. 3). Such inclusions are most common in magmatic zircons from volcanic rocks. Most analyses of zircons in G93-75 are close to concordant, but display a larger range in amet-rnica chsite schist

,/..

i i

!iiL:i

• . 1 . H I ~ ~ !t.

..'./.. . ' "l~l~J",." i":'.• ' ." ".'ll~t/~" 1' • :;:": ES",:'

!:i '.,'i~.,;:'.. .~. . . ; ~ , , ~ , " %'v "/. "'lll~l d L " ;." J~/~n,. "

I "/"

:: i;,!;,:,i,

';r :'; 4m,;,~!r .~. i'.~l~l'~ I. . f ° ..: :'. ~ t ~ : I . • ,'. -,. W/A,i ~. ! •~'.7.WJ/~t" .. ~ . i ~ .'~ "ml~,,~ i •,...,.. 71fabian. • .'1. ' " ' i W ' ,?

fine-grained felsic schist felsic clasts fine eastwards

• "/" ~ • • • I,~111

/~1/1" •"

N 10 m

Fig. 2. Sketch(after photograph)of graded felsic units in unit B1 of the Isua supracrustalbelt.

A.P. Nutman et al. / Chemical Geology 141 (1997) 271-287

275

Fig. 3. Photomicrographzircons from sample G93-75 of the B 1 felsic unit. Prismatic grain 25 (with one pyramidal termination broken-off) has an age of ~ 3710 Ma and contains a large axial early fluid inclusion. Such inclusions are common in magmatic zircons from volcanic and hypabyssal rocks. /Note the trails of younger small inclusions traversing across the grain. Grain 32 displays a rare metamorphic overgrowth with an age of ~ 3500 Ma.

2°7pb/2°6pb, with a bias to slightly older ages as compared with the clast-poor sample (Table 1; Fig. 4a). Maximum likelihood component estimation suggests a dominant population of 3 7 1 8 _ 4 Ma, not resolvable with confidence from the age of 3710 _ 4 Ma obtained on the homogeneous flow-top sample by the same method (previously reported with a weighted mean age., of 3 7 0 8 _ 3 Ma). Secondary populations at ~ 3690 Ma present in both these samples are interpreted to reflect a minor amount of radiogenic Pb-loss in a thermal event at that time, which involved emplacement of voluminous tonalites north of the Isua supracrustal belt (Nutman et al., 1996). In addition a single grain of ~ 3750 Ma was detected, but none older. This suggests that the clasts in the flow base were derived from (volcanic?) rocks the same age as, or only a few million years older than, the flows themselves. The ages for magmatic-looking zircons obtained on two samples of BI felsic formation rocks are thus ~ 100 m.y. younger than the ages on similar zircons in the A6 felsic formation rocks measured with SHRIMP (Compston et al., 1986) and by isotope dilution thermal ionisation analysis (Baadsgaard et al., 1984). Although the radiogenic 2°7pb/2°6pb ratio for 3700 Ma is ~ 6% different from that for 3800 Ma, the most sceptical analyst might contemplate that the ~ 100 Ma 2°Tpb/2°6pb apparent age difference between Isua samples run in analytical sessions years apart could be due to a bias arising out of different operating conditions of the SHRIMP

instrument. To quell any doubts that this unlikely situation could be the case, a few zircons from a sample of A6 felsic formation rocks on which ages of 3806 + 2 Ma (SHRIMP; Compston et al., 1986) and 3813 _ 9 Ma (isotope dilution thermal ionisation analysis; Baadsgaard et al., 1984) had previously been obtained were inserted in a hole drilled in the mount containing the unit B1 felsic formation sample G93-75 zircons with a SHRIMP concordant U - P b age of ~ 3710 Ma. When these A6 felsic formation zircons were analysed in rotation with the B1 felsic formation zircons during the same analytical session, a ~ 100 m.y. age difference was confirmed (Table 1; Fig. 4b). Furthermore, in this new analytical session a few anomalous turbid grains (t) and a rare core and narrow rim (r) were analysed (Table 1; Fig. 3c). The core gave an age of ~ 3700 Ma, whereas the rim gave concordant ages at ~ 3500 Ma, the latter being evidence for an early Archaean metamorphism affecting these rocks. The few turbid grains gave ages of < 3700 Ma, showing that they are disturbed damaged grains, rather than being a demonstrably older inherited component. In the southwest of the belt, the southern side of the garbenschiefer unit is interlayered with diverse mafic-felsic rocks, kyanite schists and amphibole + garnet-bearing BIF. In this paper we refer to this sample as a dirty BIF, because we interpret it as a chemical sediment contaminated at its time of deposition by the influx of a terrigenous (volcanogenic?) component. Dirty BIF sample (GGU288626) gave a

A.P. Nutman et al. / Chemical Geology 141 (1997) 271-287

276 Table 1 SHRIMP U - P b zircon analyses Spot

U

Th/U

f206°~

(ppm)

206P b / 2 3 8 U

207P b / 2 0 6 Pb

207P b / 2 0 6 Pb

%

ratio

ratio

age

disc.

0.3522 + 19 0.3536 _ 15 0.3538 + 13 0.3462 _ 14 0.3524 + 16 0.3494 + 15 0.3520 5 : 4 4 0 . 3 5 1 0 + 17 0~3475 ± 21 0.3465 + 17 0.3461 + 18 0.3532 + 22 0.3540 + 14 0.3572 + 22 0.3518 ± 18 0.3537 + 15 0.3519 + 18 0.3513 + 15 0.3485 _+ 15 0.3258 + 14 0.3529 + 26 0.3508 ± 14 0.3432 + 16 0.3394 ± 37 0.3456 ± 18 0.3518 + 19 0.3507 ± 17 0.3526 + 17 0.3506 ± 31 0.3521 + 16 0.3603 + 14 0.3544 ± 17 0.3556 + 23 0.3277 _ 12 0.3304 ± 27 0.3442 ± 55 0.3488 + 11 0.3065 + 23

3717 + 8 3723 _ 6 3723 + 6 3690 + 6 3717 + 7 3705 -I- 7 3716 ::1:19 3711 -I- 8 3696 + 9 3692 + 7 3690 + 8 3721 + 10 3724 + 6 3738 + 9 3715 + 8 3723 _ 7 3715 + 8 3713 -[- 6 3700+ 7 3597 + 7 3719 + 11 3711 ± 6 3677 + 7 3660 +_ 17 3688 _ 8 3715 ± 8 3710 + 7 3718 ± 7 3710 ± 14 3716 + 7 3751 _ 6 3726 ± 7 3731 ± 10 3606 ± 6 3619 ± 13 3682 ± 25 3702 ± 5 3504 _ 12

- 3 - 1 -2 -6 -2 + 1 0 -4 - 1 + 1 - 1 -4 -4 -2 -3 -4 - 5 -3 -4 -5 - 3 -3 - I + 1 + 1 -3 -4 - 1 -9 -4 -3 0 -2 - 3 -7 -2 + 2 0

0.3762 0.3615 0.3571 0.3792 0.3756 0.3733

± 43 + 39 + 22 ± 63 ± 70 _+ 46

3816 3756 3738 3828 3814 3805

-4 + 1 - 10 + 1 - 10 -2

0.3506 0.3501 0.3497 0.3526

+ ± + ±

3710 + 3707 _ 3706+ 3718 +

Clast-rich flow base G93-75 of felsic volcanic unit - B1 felsic schist formation 1-1 114 0.84 2-1 196 0.68 3-1 208 0.57 4-1 194 0.48 5-1 152 0.63 6-1 171 0.76 6-2 * 160 0.70 7-1 161 0.75 7-2 * 109 0.51 8-1 143 0.65 9-1 148 0.46 10-1 84 0.52 11-1 196 0.77 11-2 * 194 0.66 12-1 147 0.80 13-1 192 0.79 14-1 115 0.43 15-1 164 0.67 16-1 195 0.46 17-1 214 0.27 18-1 77 0.80 19-1 231 0.81 20-1 181 0.69 21-1 44 0.69 22-1 150 0.40 23-1 187 0.68 24-1 207 0.73 25-1 187 0.73 25-2 * 87 0.72 25-2 161 0.68 26-1 224 0.84 27-1 152 0.60 28-1 103 0.65 29-1t * 296 0.50 30-1t * 155 0.64 31-It ~ 114 0.93 32-1c * 658 0.99 32-2r * 604 0.09 A6 unit zircons run interspersed with G93-75 1-1 60 0.85 2-1 96 1.30 3-1 141 1.12 4-1 87 0.78 5-1 79 1.06 6-1 75 0.87

< 0.01 0.754 + 16 0.02 0.767 + 15 < 0.01 0.757 + 15 0.02 0.714 ± 14 0.09 0.763 + 15 0.04 0.786 _+ 16 0.11 0.784 ___30 0.22 0.741 + 15 0.08 0.764 _+ 27 0.10 0.784 _+ 16 < 0.01 0.759 ± 16 0.15 0.741 ± 16 0.07 0.742 + 15 0.08 0.761 _ 32 0.09 0.749 _ 15 0.06 0.738 + 15 < 0.01 0.734 + 15 < 0.01 0.746 ___15 0.19 0.735 + 15 0.11 0.696 -t- 14 0.67 0.755 + 16 0.24 0.746 + 15 0.22 0.762 + 15 0.14 0.771 ± 23 < 0.01 0.777 ± 20 0.19 0.747 + 20 0.02 0.740 ± 19 0.09 0.772 ± 20 0.13 0.686 + 42 < 0.01 0.743 ± 19 0.03 0.762 ± 20 < 0.01 0.781 ± 20 0.11 0.766 ± 20 0.10 0.717 + 41 0.17 0.683 +_ 35 0.42 0.746 ___30 0.09 0.791 ± 24 0.03 0.725 ± 35 zircons marked by asterisk 0.29 0.761 + 27 0.22 0.801 ± 28 0.20 0.690 ± 20 0.56 0.823 ± 31 0.45 0.701 __+41 0.16 0.781 ± 32

+ 17 ___ 17 + 9 ± 25 ± 28 ± 19

'Dirty' banded iron formation GGU288626 - A3 variegated schist formation 1-1 1-2 2-1 2-2

89 116 69 170

0.70 0.60 0.55 0.77

0.14 < 0.01 0.21 0.04

0.810 0.795 0.741 0.762

+ + + +

13 21 12 20

18 17 16 14

8 7 7 6

+3 +2 -4 -2

A.P. Nutman et al. / Chemical Geology 141 (1997) 271-287

277

Table 1 (continued) Spot

U

Th/U

f206%

2°6pb/238U

2°7pb/2°6pb

2°7pb/2°6pb

%

ratio

ratio

age

disc.

'Dirty' banded iron formation GGU288626 - A3 variegated schist formation 3-1 144 0,42 < 0.01 0.770 5:12 4-1 80 0.37 0.43 0.677 + 11 5-1 77 0.39 < 0.01 0.779 5:13 6-1 136 0.41 <0.01 0.821 + 13 6-2 131 0.50 0.27 0.753 5:20 7-1 46 0.51 < 0.01 0.795 + 14 8-1 171 0.69 0.05 0.748 + 19 9-1 106 0.65 0.05 0.743 + 20 10-1 111 0.49 0.08 0.692± 18 11-1 224 0.53 0.67 0.747 5:19 12-1 240 0.40 0.13 0.722 ± 19

0.3494 5:10 0.2934 5:t5 0.3036 5:12 0.32875:11 0.3296 5:15 0.3284 + 16 0.3431 5:14 0.3469 5:20 0.3161 5:16 0.3237 ± 16 0.3161 5:12

3704 3436 3489 3611 3615 3609 3677 3693 3551 3587 3551

+ 4 5:8 5:6 5:5 + 7 5:7 + 6 5: 9 5: 8 5: 7 5: 6

- 1 - 3 +6 +7 0 +4 -2 -3 -5 0 - 1

Kyanite schist G93-22 - A3 variegated schist formation 1-1 61 1.01 9.94 1-2 421 1.17 1.44 2-1 1109 0.93 0.05 2-2 952 0.90 0.17 3-1 172 0.69 0.07 3-2 116 0.82 0.15 3-3 246 0.66 0.27 3-4 124 0.77 0.20 4-1 163 1.29 2.76 4-2 245 1.41 0.54

0.1566 5:65 0.1627 5:11 0.1420 + 04 0.1538 5:04 0.3509 ± 11 0.35175:16 0.3513 5:12 0.3487 5:16 0.1772 + 22 0.1813 + 11

2419 5:70 2484 5:11 2252 ___ 5 2388 5 : 5 3711 5 : 5 37155:7 3712 5 : 5 3701 ± 7 2627 ± 21 2665 5:10

+4 - 16 - 14 - 16 - 4 -1 - 17 - 1 -3 + 1

(ppm)

0.479 5:08 0.383 ± 05 0.351 __+05 0.364 + 05 0.734 + 10 0.764+ 11 0.613 ___08 0.763 5:11 0.485 ± 07 0.519 ± 07

Granodiorite sheet GGUT'25917 (100 m from GGU288626) 1-1 651 1.97 9.25 0.668 1-2 1206 1.42 0.55 0.693 1-3 393 0.62 1.95 0.704 2-1 897 1.41 < 0.01 0.740 2-2 683 1.09 0.17 0.724 2-3 1001 1.33 0.08 0.695 2-4 655 1.05 0.55 0,731 3-1 2681 1.22 0.15 0.723 4-1 1333 0.34 0.20 0.427 4-2 611 1.65 0.87 0.705

5:12 5:15 5:16 + 13 5:16 ± 15 + 16 + 13 + 08 + 16

0.3258 0.3191 0.3222 0.3380 0.3200 0.3140 0.3220 0.2924 0.2933 0.3370

5:19 + 05 5:13 5:05 + 06 + 04 5:08 + 04 ± 04 + 08

3598 3565 3581 3654 3570 3541 3579 3430 3435 3649

5: 9 5: 2 + 6 5: 2 5:3 5:2 + 4 5: 2 + 2 + 4

-8 -5 -4 -2 -2 -4 - 1 +2 -33 -6

Tonalite sheet MR81-261 1-1 106 2-1 41 3-1 88 4-1 66 5-1 411 6-1 172 7-1 142 8-1 182 10-1 215 11-1 263 12-1 187 13-1 237 14-1 96 15-1 112 17-1 196

+ 34 + 33 + 31 + 25 + 16 + 17 + 21 + 20 5:20 + 19 ___27 5:16 5:20 5:20 5:20

0.3720 + 55 0.3728 + 72 0.3737 5:37 0.3687 5:67 0.3715 5:11 0.3726 _+ 14 0.3659 _+ 45 0.3721 5:25 0,3711 + 30 0,3720_ 19 0,3699 5:56 0.3728 -1- 33 0.3729 5:20 0.3710 + 25 0.3700 + 13

3800 3803 3807 3786 3998 3802 3774 3800 3796 3800 3791 3803 3803 3795 3791

5:22 + 29 5:15 5:27 5:5 5:6 5:19 5:10 5:12 + 08 + 23 5:13 + 8 5:10 5:5

-4 0 -3 -5 0 -3 - 1 -2 - 2 -1 0 -6 -5 - 1 - 1

0.37 0.35 0.38 0.40 0.72 0.35 0.83 0.58 0.73 0.49 0.50 0.53 0.64 0.40 0.99

0.26 0.96 0.27 0.71 0.10 0.16 0.32 0.35 0.56 0.26 0.17 0.21 0.41 0.59 0.32

0.765 0.808 0.778 0.744 0.802 0.771 0.786 0.780 0.778 0.791 0.797 0.745 0.753 0.787 0.795

278

A.P. Nutman et al. / Chemical Geology 141 (1997) 271-287

Table 1 (continued) Spot

U

Th/U

f206%

(ppm)

206Pb/238 U

207Pb/206 Pb

207Pb/2o6 Pb

%

ratio

ratio

age

disc.

Tonalite sheet MR81-261 18-1 155 19-1 296 20-1 212 21-1 177

0.58 0.82 0.62 0.40

0.11 0.14 0.18 0.98

0.780 0.764 0.759 0.749

-I- 32 ± 21 + 17 _+ 18

0.3722 0.3711 0.3721 0.3663

+ 68 -t- 41 + 13 + 21

3800 3796 3800 3776

_+ 28 ± 17 ± 5 _ 9

-2 -4 -4 -5

Quartzite ? G93-25 1-1 37 2-1 63 3-1 78 4-1 96 5-1 80 6-1 161 7-1 50 8-1 188 9-1 81 10-1 122 11-1 84 12-1 41 13-1 122 14-1 128 15-1 78 15-2 86 16-1 99 17-1 218 17-2 189 18-1 132 19-1 286 20-1 62 21-1 78 22-1 50 23-1 84 24-1 45 25-1 275 26-1 54 27-1 228 27-2 364 28-1 123 29-1 238 30-1 116 31-1 124 32-1 66 32-2 54

0.48 0.76 0.34 0.70 0.65 0.45 0.76 0.59 0.58 0.63 0.48 0.51 0.64 0.71 0.54 0.55 0.70 0.55 0.52 0.46 0.60 0.68 0.50 0.59 0.35 0.65 0.59 0.66 0.46 0.58 0.66 0.72 0.42 0.66 0.41 0.40

0.57 0.99 0.04 0.18 0.28 0.06 0.31 0.10 < 0.01 0.01 0.89 0.86 0.06 0.17 0.14 0.05 0.21 0.18 0.23 0.16 0.37 0.69 0.83 0.49 3.59 1.00 0.68 0.80 0.30 1.08 1.33 1.83 0.29 19.91 6.29 0.30

0.841 0.861 0.823 0.826 0.837 0.812 0.793 0.794 0.834 0.740 0.778 0.773 0.814 0.788 0.779 0.793 0.778 0.805 0.834 0.755 0.770 0.784 0.828 0.757 0.746 0.833 0.777 0.799 0.749 0.752 0.750 0.771 0.798 0.787 0.754 0.793

+ 41 + 30 +_ 33 _+ 27 _+ 27 + 24 + 32 + 24 + 33 _+ 16 +_ 25 + 37 ± 18 + 16 +__23 + 26 + 20 _____13 + 19 + 21 + 29 + 18 + 24 5:28 __ 21 + 77 ± 22 + 25 + 24 + 24 + 19 + 22 + 22 + 32 _____49 ± 21

0.3902 0.3863 0.3886 0.3849 0.3844 0.3842 0.3800 0.3836 0.3865 0.3865 0.3827 0.3913 0.3854 0.3835 0.3770 0.3756 0.3830 0.3967 0.3967 0.3852 0.3810 0.3864 0.3738 0.3837 0.3886 0.3877 0.3867 0.3814 0.3747 0.3730 0.3815 0.3854 0.3866 0.3898 0.3914 0.3844

___49 + 32 _ 26 _+ 30 ± 31 +_ 17 + 40 ± 15 _ 31 _+ 36 +_ 28 _+ 82 _ 31 +_ 31 __+30 + 31 + 22 __ 23 _____17 + 17 ± 27 + 22 + 47 +___56 _____31 + 48 __+16 + 22 + 30 + 10 + 30 + 20 ± 15 + 64 + 84 + 19

3872 3856 3865 3851 3849 3848 3832 3846 3857 3857 3842 3876 3853 3846 3820 3814 3843 3897 3897 3852 3836 3857 3807 3847 3866 3862 3858 3837 3811 3804 3838 3853 3858 3870 3876 3849

+ 19 + 13 _+ 10 + 12 ± 12 + 7 +_ 16 + 6 +_ 12 + 14 ± 11 + 32 -i- 12 ± 12 + 12 + 13 +___ 9 ± 9 ± 6 _____ 7 + 11 + 9 __+19 + 22 + 12 + 19 + 6 + 9 + 12 + 4 + 12 ± 8 __ 6 + 25 + 32 + 8

+2 +4 0 + 1 +2 0 - 2 -2 + 1 -7 -3 -5 0 -3 - 3 - 1 -4 -2 0 -6 -4 -3 +2 -6 -7 + 1 -4 - 1 -5 - 5 -6 -4 -2 -3 -7 -2

f206% is the proportion of 2°6pb in percent that is not radiogenic. % disc. is the degree of discordance in percent between the 2°7pb/2°6pb and 2°6pb/238U ages. All errors are quoted at the 1 tr level. For G93-75, t = turbid grain, c = core and r = rim. The six-digit sample numbers refer to the files of the Geological Survey of Denmark and Greenland. Other samples were collected on expeditions funded by other agencies.

low yield (~

10 z i r c o n s k g - ~ ) o f s t u b b y p r i s m a t i c ,

slightly rounded

zircons, which show a spread of

early Archaean 2°7pb/2°6pb

ages between

~ 3715

a n d 3 4 0 0 M a ( T a b l e 1; F i g . 4c). T h e o l d e s t a n a l y s e s (from three grains) with a 2°7pb/Z°6pb

weighted

m e a n a g e o f 3 7 0 7 + 6 M a , are i n t e r p r e t e d as g i v i n g

279

A.P. Nutman et al. / Chemical Geology 141 (1997) 271-287

0.9 G93-75 B1 felsic unit

G93o75 B1 felsic unit zircons run

3 8 0 0 y

DO.8

interspersed with unit ~ A6 felsic unit ~.:"~00 zircons 3650 ~"/d/.7

'' /

r \ t "xv~

O.

~ 0.7

0.6 0.9

~

I

I

i

I

G93-22 kyanite schist

288626 BIF

3 8 0 ~

~0.8 O3 Q.

35oo

0.7

0.6 0.9

~

I

I

i

i

..

,

,v~

metamorphic zircons

~)/,

with late Archaean ages not shown

I

I

=

I

MR810261 tonalitic sheet

225917 granodioritesheet 3800

~ 0.8

3

3700/

%

7

0

~

Q.

0.7

0.6 0.9

i

I

I

I

25.0

i

i

35.0

I

I

45.0

G93-25 quartzite ~

37

-~0.8

.jf ;900

6-

D.

~ 0.7

O >

0.6 ~ 25.0

I

I

35.0 207pb/235U

i

I

45.0

0 3700

3800 3900 4000 2°7pb/2°6pb age (Ma)

Fig. 4. U-Pb zircon geochronology.Errors on the U-Pb concordiadiagramsare depicted at the I tr level.

the original age of a disturbed population of zircons. It could be argued that the younger, close to concordant zircons in this sample are either the product of ancient loss of radiogenic Pb from ~ 3710 Ma zircons, or are possibly detrital in origin, giving an age of deposition perhaps as young as ~ 3550 Ma. However, granodiorite sheet GGU225919 cutting

supracrustal rocks ~ 100 m to the south of the BIF sample contains prismatic (high Th + U) zircons which yielded a°7pb/2°6pb ages of predominately ~ 3570 Ma, but with two analyses yielding ages up to ~ 3650 Ma (Table 1; Fig. 4d). This intrusive sheet either formed at ~ 3650 Ma and the high U + Th magmatic zircons were badly disturbed in a

280

A.P. Nutman et aL / Chemical Geology 141 (1997) 271-287

~ 3570 Ma thermal event, or the sheet formed at ~ 3570 Ma and contains some slightly older inherited zircon. Either way the granodiorite sheet has a minimum age of ~ 3570 Ma, requiring the BIF sample to be > 3570 Ma, and most likely deposited at ~ 3710 Ma. An adjacent kyanite schist (sample G93-22; Fig. ld) yielded very few zircons (Table 1; Fig. 4e). Four analyses on a single small, pale brown, prismatic grain similar to those in the dated B 1 felsic formation volcanic samples yielded a weighted mean 2°Tpb/2°6pb age of 3711 + 6 Ma, indistinguishable from the least disturbed zircons in the nearby dirty BIF. Three other deep red-brown stubby prismatic to equant grains yielded concordant late Archaean or discordant younger ages, and are interpreted to have either grown or to have been strongly disturbed during metamorphism. Although not definitive alone, these data combined with the useful results from the B1 felsic formation volcanic rocks indicate that diverse supracrustal rocks and probably the associated garbenschiefer unit are ~ 3710 Ma old.

3.3. > 3790 Ma age for amphibolite-dominated package, southern side of the belt This package is dominated by amphibolites containing layers of ultramafic rocks, banded iron formation and possible metachert. Ultramafic rocks and adjacent lithologies in this package are overall the most highly metasomatised rocks of the belt (Rosing et al., 1996). Deformed pillow structures are in rare instances preserved in very fine-grained amphibolites. This package is cut by numerous discordant tonalite sheets. Owing to the lack of suitable targets for zircon geochronology in the amphibolites themselves, these sheets have been used to define a minimum age of this package. Sample G93-13 from the southwest margin of the belt has a simple magmatic population of euhedral, prismatic zircons, devoid of either obvious inherited structural cores or metamorphic overgrowths and with a 2°7pb/2°6pb weighted mean age of 3791 _+ 4 Ma, giving the age of the tonalite protolith. (Fig. ld; Nutman et al., 1996). In the central-southern segment of the belt, tonalite sheet MR810261 (Fig. lc) discordantly cuts its host layered amphibolites. The zircons in this sheet are similar to those in G93-13. All analyses are close to concordant (Table 1; Fig. 4f) and yield a

2°7pb/2°6pb weighted mean age of 3798 +_ 4 Ma, indistinguishable from the age obtained on G93-13. This shows that along the southern part of the belt there are > 3790 Ma supracrustal rocks.

3.4. Polymodal ages of zircons in siliceous rocks (qua rtzites ?) The search in detrital zircon populations for old components can show whether 'pre-volcanic' sialic crust contributed to the supracrustal sequences in the Isua supracrustal belt. Possible detrital quartzites in the belt are extremely rare. A sample from a unit of biotite _+ garnet-bearing siliceous rock (quartzite?; MR81-318) from the southern central part of the belt (Fig. lc) yielded approximately 100 zircons. All the grains are very small (typically 30 to 50 txm across) and are both equant and stubby prismatic in habit (Nutman and Collerson, 1991). These zircons have suffered variable loss of radiogenic Pb and also show a few 2696 ___6 Ma metamorphic overgrowths. Nonetheless, there are clearly two > 3800 Ma groups present, which yielded weighted mean ages of 3808 _+ 5 Ma and 3847 _ 10 Ma. Another possible biotite-quartzite unit (sample G93-25) has been found in the southwest of the belt (quartzite?; Fig. ld). This unit is ~ 1 m thick, and is bounded to the south by a thick, partly carbonated, ultramafic schist unit and to the north by garnetiferous mafic schist. This mafic schist is separated from the A6 felsic formation to the north by a seam of quartz-fuchsite rock; possibly a recrystallised mylonite. Sample G93-25 also gave a low yield of zircons ( ~ 35 grains), only slightly larger in size than those in quartzite MR81-318. Most analyses are close to concordant within error (Table 1; Fig. 4g) with 2°7pb/2°6pb ages between 3800 and 3900 Ma. A predominant group of ~ 3850 Ma is present. Grains 17 (two analyses), 21 and 27 (two analyses) yielded 2°7pb/2°6pb ages of 3800-3820 Ma. Two analyses of grain 5 both yielded 2°7pb/2°6pb ages of 3897 Ma (Table 1). These results, particularly with the duplicate analyses, suggest that although a ~ 3850 Ma group is dominant in the population, a few younger (3800-3820 Ma) and older (up to 3900 Ma) grains are also present. The presence of the oldest zircons require contribution of at least limited amounts of older crustal components to the Isua supracrustal belt.

A.P. Nutman et al. / Chemical Geology 141 (1997) 271-287

4. Discussion 4.1. Packages o f unrelated supracrustal rocks

The zircon age determinations demonstrate that the Isua supracrustal belt comprises at least two temporally distinct packages of supracrustal rocks. Along the northern side of the Isua supracrustal belt, sedimentary and volcanic rocks found with the dominant garbenschiefer unit, and probably interlayered with it, have yielded ~ 3710 Ma zircon populations. The zircon and fielld data are interpreted here as showing that the garbenschiefer and associated rocks are part of a volcanosedimentary sequence formed at ~ 3710 Ma. Additionally, Rosing et al. (1996) noted that the garbenschiefer package had experienced a different metasomatiic history from the rest of the belt, and that it contains markedly fewer felsic intrusive sheets. On this basis they suggested that it was tectonically assembled with the other components of the Isua supracrustal belt after the main metasomatic event. Along the southern side of the Isua supracrustal belt, the package of amphibolites (locally with preserved pillow structure) interlayered with ultramafic rocks and BIF is cut by discordant 3790-3800 Ma tonalite sheets. Therefore this package is >_ 3790 Ma and must be considerably older than the package containing the garbenschiefer. Rosing et al. (1996) suggested the possibility that quartzite sample MR810318 from the _>3790 Ma amphibolitedominated package was a metasomatised intrusive sheet with an age of 3847 _+ 10 Ma. In this interpretation, however, no attempt was made to account for an equally significant population of 3808 + 5 Ma zircons. Both populations of zircons are very small and somewhat rounded and are atypical of those in definite intrusive sheets in the belt. Morphologically, they more closely resemble detrital zircons. Similarly, possible quartzite G93-25 is dominated by ~ 3850 Ma zircons, but both younger ( ~ 3800-3820 Ma) and older ( ~ 3900 Ma) grains are present. Rosing et al. (1996) may be correct in their interpretation of these contentious rocks as altered intrusive sheets (hypabyssal because of the small zircon size), requiring the host rocks to be >_ 3800 Ma. However, with the varied age and small size of zircons, it must also be considered possible that they are intra-flow

281

sediments with a likely age of ,-, 3800 Ma, containing older detrital zircons. Such rocks have been found in some late Archaean komatiitic-tholeiitic sequences (e.g., Kambalda district of Western Australia; Claout-Long et al., 1988). Aside from ages, there are other significant differences between the > 3790 Ma and ~ 3710 Ma packages. The > 3790 Ma package is intruded by abundant, locally discordant tonalite sheets, whereas such sheets are rare in the ~ 3710 Ma package. Secondly, excluding the 3806 ± 2 Ma 'A6' felsic unit as being of unclear volcanic or intrusive origin (according to Rosing et al., 1996), the > 3790 Ma package is dominated by mafic and ultramafic rocks for which reconnaissance geochemical studies indicate low Z r / T i values. In contrast, the ~ 3710 Ma package is lithologically much more diverse and many fresh rocks show high, variable Z r / T i values and generally show LREE enrichment (Nutman et al., 1996). We speculate that the high, variable Z r / T i package along the north of the Isua supracrustal belt, dominated by the garbenschiefer and containing intermediate to felsic volcanic rocks and sediments might have formed in an arc environment. Some Archaean volcanosedimentary sequences containing komatiites are argued to be plume-related (e.g., McDonough and Ireland, 1993, and references therein). A similar origin is possible for the older (low Zr/Ti?) package in the Isua supracrustal belt. The question then arises as to the nature of the boundary between these two lithologically and chronologically distinct sequences. Mylonites have been detected within the Isua supracrustal belt and at its northern margin, giving rise to truncation of lithological units (Fig. 1). Some of these mylonites must be Archaean, because they are cut by Ameralik (mafic) dykes which were recrystallised during regional late Archaean ductile deformation and amphibolite facies metamorphism (e.g., on the northern margin of the belt; Fig. lc). However, it is not established that any of the mylonites marked on Fig. 1 represent large-scale tectonic breaks formed in the early Archaean. Sutures between the different sequences might be very hard to find, because they are likely to be in areas of highest strain where discordances would have been obliterated, and additionally might be masked by some of the locally intense metasomatic alteration. Therefore, the full extent of

282

A.P. Nutmanet al. / Chemical Geology 141 (1997)271-287

the two packages of different age is not known. Still unresolved is a precise and accurate age of the large package of banded iron formation and chert that dominates the eastern extremity of the belt (Fig. 1). The U - P b zircon data suggests that the Isua supracrustal belt comprises unrelated ~ 3710 and >_ 3790 Ma packages of supracrustal rocks. Within each package there might remain a semblance of a stratigraphy. The belt resides in a terrane of 38103650 Ma dioritic to granitic gneisses which locally contain units of supracrustal rocks smaller than the Isua supracrustal belt (Nutman et al., 1993, 1996). Although the tectonic setting of the Isua supracrustal belt is not clear, the events outlined here give rise to a picture very similar to early Archaean crustal evolution in southern Africa, where the Barberton greenstone belt containing supracrustal rocks of different ages resides in a surrounding sea of older gneiss and younger granitoids, and developed over 200 to 300 Ma (e.g., De Wit et al., 1992; KriSner et al., 1996; R.A. Armstrong, pers. commun., 1996). 4.2. Implications for Nd isotopic compositions in the early Earth

147Sm-143Ndsystem Initial 143Nd/144Nd ratios

4.2.1.

of Isua supracrustal belt rocks and of TI'G gneisses, both components of the Itsaq Gneiss Complex of southern West Greenland, have been extensively used to try and establish early Archaean mantle Nd isotopic compositions and to provide evidence of crustal recycling and fractionation events in the early Earth (e.g., Jacobsen and Dymek, 1987). The results of these studies have been interpreted to indicate that some portions of the early Archaean ( > 3760 Ma) mantle were subjected to extreme LREE fractionation early in Earth's history (Bennett et al., 1993). More recent publications (e.g., Bridgwater and Rosing, 1995; Gruau et al., 1996; Moorbath et al., 1997) have questioned this interpretation and have focused on the possible role of secondary S m / N d fractionation in generating high apparent eNa(3800) values of > + 3, and concluded that eNa values of + 1 to + 2 for the 3800 Ma depleted mantle are more likely. Moorbath et al. (1997) reached this conclusion based on regression treatment of S m - N d isotopic data from mixed suites of mica schists, amphibolites and quartzo-feldspathic

rocks, of diverse age and metasomatic history. They also noted that the end(3800) values of + 1 to + 2 were in accord with a class of depleted mantle evolution models with linear or close to linear trajectories joining eNd(a560) of 0 to eNd(present) of ~ + 10 (e.g., Goldstein et al., 1984). Thus it should be stressed that the 'expected' ey a values of + 1 to + 2 for the depleted mantle at 3800 Ma have only been suggested by one type of Nd evolution model (one of several diverse model types in the literature, see McCulloch and Bennett, 1994); such values have never been determined from isotopic measurements of unmetamorphosed 3800 Ma MORB! We hope there will be more sceptical examination of Nd data on all metamorphosed and deformed Archaean and Proterozoic rocks, even if their apparent initial end values agree with currently favoured models of Precambrian depleted mantle evolution. Additional questioning of Nd whole rock initial compositions has been raised on the basis of Hf isotopes. In a comparative zircon Hf and whole rock Nd isotope study of some of the best-preserved > 3600 Ma rocks, Vervoort et al. (1996) noted the greater scatter in whole rock initial end values compared to initial eHf values determined on zircon separates and that eHf ~ 2 × eNd , as documented in some younger suites. This discrepancy was interpreted to be due to fractionation of Sm from Nd during young ( < 3600 Ma) metamorphism. However, in the Vervoort et al. (1996) sample suite, variation in whole rock initial eNd and enf values might be more related to age and lithology, rather than to severity of later metamorphism and metasomatism (details of samples lodged at h t t p : / / w w wrses.anu.edu.au/earlyHf-Nd), enf/eNa ratios vary from ~ 1 to ~ 2 between the different Itsaq Gneiss Complex suites discussed by Vervoort et al. (1996). However, the same eHf//gNd range is observed in modem MORB (e.g., Salters, 1996) and may be attributed to derivation from different residual assemblages, i.e. garnet versus spinel peridotites, with long mantle residence times. Thus the eHf--eNd relationships in the early Archaean rocks of West Greenland might equally well reflect ancient fractionation processes rather than disturbance of the whole rock S m - N d systems. Prior to the 1990s there was little U - P b zircon data on Isua supracrustal belt rocks (e.g., Michard-

283

A.P. Nutman et al. / Chemical Geology 141 (1997) 271-287

Vitrac et al., 1977; Baadsgaard et al., 1984; Compston et al., 1986), thus appraisals of whole rock initial Nd isotopic ratios (e.g., Hamilton et al., 1978; Moorbath et al., 1!)86; Jacobsen and Dymek, 1987) relied largely on S m - N d isochron age calculations (Table 1). Even neglecting possible fractionation of parent-daughter isotopic systems by later whole rock open system behaviour, inaccurate isochron 'ages' for Archaean suites can be produced by combining isotopically distinct sources (e.g., Chauvel et al., 1985) or samples of different ages (e.g., Collerson et al., 1989). In the ,;tudies of Hamilton et al. (1978) and Moorbath et al. (1986, 1997) samples from different lithological units were often plotted together, presumably to increase the range of S m / N d for the isochrons. Partly based on the geochronological data presented here, we argue that this approach included samples from units differing in age by ,,, 100 Ma (i.e. the 3806 _ 2 Ma ' A 6 ' felsic unit and the ~ 3710 Ma garbenschiefer unit and associated intermediate to felsic volcanic rocks). These S m / N d isochron age determinations have large errors and mostly scatter between 3700 and 3800 Ma, the known protolith ages from the recent U - P b zircon geochronology (Table 2). We interpret these determinations as variously weighted average ages of mixed suites. Notwithstanding, the S m - N d and R b - S r reference lines (none had MSWDs low enough to qualify as genuine isochron ages) were important prior to the more precise U - P b zircon geochronological investi-

gations because they documented the great age of the Isua supracrustal belt. However the large errors obtained by this method precludes the possibility of reconstructing detailed tectonic histories by resolving closely spaced events, as has been done by U - P b zircon geochronology in many Archaean greenstone sequences (e.g., of the Superior Province; Corfu and Davis, 1992). Jacobsen and Dymek (1987) calculated isochron ages using samples from single units, but also got widely scattered ages with large uncertainties (Table 2). For the ' A 6 ' felsic unit, the scatter in the S m - N d data was so great that they resorted to bracketing the data by two 3810 Ma reference lines (the U - P b zircon age of these rocks), with initial end values of + 4.0 to + 1.0. Jacobsen and Dymek interpreted the + 4 . 0 to + 1.0 spread of ~d(3800) values for their whole sample suite to indicate that ,the Isua supracrustal rocks contain material derived from two or more sources, principally depleted mantle (the most positive ~NdO800) component) and older sialic crust (lower 6Sd(3800)). They also noted that many of the samples had more positive eNd(3800) values than predicted by most depleted mantle evolution models at 3800 Ma. This led them to suggest that some of the Isua supracrustal rocks could be younger than 3800 Ma. We have re-calculated the Nd data of Jacobsen and Dymek (1987) on the basis of the new age constraints presented in this paper. Although the older sequence represents perhaps half the areal ex-

Table 2 Summary Sm-Nd isotopic investigations of Isua supracrustal rocks Source

Samplesuite

Regression age (Ma)

1 1

Garbenschiefer amphibolites ( ~ 3710 Ma) Garbenschiefer amphibolites ( ~ 3710 Ma) and 'A6' felsic rocks (3806 4- 2 Ma) Garbenschiefer amphibolites (~ 3710 Ma) Garbenschiefer amphibolites ( ~ 3710 Ma) and 'A6' felsic rocks (3806 + 2 Ma) Assorted Isua supracrustal belt volcanic and sedimentaryrocks Assorted Isua supracrustal belt volcanic and sedimentaryrocks 'A6' felsic unit (3806 4- 2 Ma) Sequence 'B' units (~ 3710 Ma) 58 assorted Isua supracrustal belt volcanic and sedimentary rocks

3772 4- 134 3770 4- 42

2 2 3 4 4 4 5

Initial ~Nd

3730 4- 150 3770 4- 42

+ 1.5 4- 3.8 + 1.4 4- 0.9

3716 4- 73 3780 4- 170 scattered between 3810 reference lines 3600 + 290 3776+ 52

+ 2.0 4- 0.8 +2.2 + 1.7 + 0.1 to + 4.0 +0.3 4- 2.8 + 2.0 4- 0.3

1 = Hamilton et al. (1978); 2 = Hamilton et al. (1983); 3 = Moorbath et al. (1986); 4 = Jacobsen and Dymek (1987); 5 = Moorbath et al. (1997). Where appropriate, U-Pb zircon ages are given in brackets.

284

A.P. Nutman et al. / Chemical Geology 141 (1997) 271-287 6

Isua

~

5 4 3

Itsaq Gneiss Complex .~ units 3760-3810 Ma l-rG gneisses .. A6

.... ";~" , ~ " ~

t err°r

EN d 2 1 0

• A3 eB1 o B2

-1 -2 3500

3600

3700 3800 Time (Ma)

3900

4000

Fig. 5. ~'Nd--t plot for Isua supracrustal belt rocks, with end values recalculated on the basis of new age information.

tent of the Isua supracrustal belt, published analyses for which localities can be identified are mostly from the younger ( ~ 3710 Ma) sequence. Supracrustal rocks from the ~ 3710 Ma sequences have end values from ~ --1.0 to + 3.0 at 3710 Ma (Fig. 5), rather than ~ 0 to + 4 at 3800 Ma. Given that these values have typical analytical uncertainties of _ 0.5 end unit, these samples were not drawn directly from only an ultra-depleted reservoir, like that proposed for many > 3750 Ma Greenland rocks by Bennett et al. (1993). Furthermore, the small spread overall in eNd , most notable in well-preserved samples from unit 'B 1' of intermediate to felsic volcanic rocks (Fig. 4), suggests that the Nd isotopic systematics of some early Archaean rocks have not been grossly disturbed in younger events. The average end value of + 1.6 for B 1 samples is almost identical to estimated values of 3710 Ma depleted mantle based on linear evolution from CHUR at 4560 Ma to + 10 eNd units at present (DM, Fig. 5). The simplest explanation of these ~ 3710 Ma samples is that they were extracted at, or just before, 3710 Ma from such a depleted mantle reservoir. However, given that the ~ 3710 Ma package ranges in composition from basaltic to rhyolitic, and the likelihood of incorporating older material during possible formation in an arc-like environment (Nutman et al., 1996; Rosing et al., 1996), such an interpretation might be simplistic, because in many modem arcs there is clear evidence of mixing of sources distinct in their Nd isotopic composition (e.g., DePaolo, 1980). Thus, an equally plausible model is that the ~ 3 7 1 0 Ma Isua supracrustal rocks contain contributions from older (3750-3870 Ma) felsic rocks found in the Itsaq

Gneiss Complex and a ~ 3710 Ma depleted mantle source (Fig. 5). The difficulty in distinguishing between these models is that by 3710 Ma, the likely older crustal components (with end(3800) mostly + 3 to + 4 ; Bennett et al., 1993) had evolved to lower eNd values indistinguishable, within error, from a model linear depleted mantle evolution (Fig. 5). This is in contrast to younger epochs, where average crustal and mantle reservoirs bad sufficient time to evolve distinctive compositions, so that different source components had initial Nd isotopic values differing by tens, rather than just two or three, eNd units (e.g., DePaolo, 1980). The only representatives of unequivocal > 3710 Ma rocks from the Isua supracrustal belt clearly identified in the literature are the felsic gneisses from the 3 8 0 6 + 2 Ma 'A6' unit (Fig. 1 and Table 2; Hamilton et al., 1978, 1983; Moorbath et al., 1986; Jacobsen and Dymek, 1987). However, if these rocks are altered gneiss sheets as suggested by Rosing et al. (1996), then there are no published Nd data for definite >_ 3790 Ma Isua supracmstal rocks! The A6 felsic unit samples are from sites in the central segment of the Isua supracrustal belt where there has been substantial episodic metasomatism involving both hydration and carbonatisation from the early Archaean to the Proterozoic (e.g., Nutman et al., 1984; Rosing et al., 1996). Despite being some of the most heterogeneously altered samples, they show a spread in 1478m//144Nd ratios of only 4% and a range in eNd(3800) of only + 1.0 to + 2.9. Given the coherency of these data, it is possible that the metasomatism did not markedly fractionate Sm from Nd in these rocks, so that their end(3800) values of + 1.0 to + 2.9 could indicate that they were extracted from a strongly depleted reservoir at 3806 Ma. The pattern whereby many _> 3750 Ma rocks have similar or more positive eNd values than < 3750 Ma rocks is observed in the Itsaq Gneiss Complex (Bennett et al., 1993), the Acasta gneisses (Bowring and Housh, 1995) and Labrador samples (Collerson et al., 1991). The meaning of the inferred radiogenic Nd isotopic compositions from the most ancient samples is currently controversial, with arguments centred on whether these compositions reflect derivation from an ultra-depleted mantle reservoir or have resulted from secondary parent-daughter element fractionation. We note here that the Isua supracrustal belt is a

A.P. Nutman et al. / Chemical Geology 141 (1997) 271-287

problematic area to answer this question, because of locally extreme alteration particularly of mafic and ultramafic rocks, the range of lithologies of different age, and based on the zircon record, evidence of at least local incorporation of older crustal components. In contrast, other p~u'ts of the Itsaq Gneiss Complex locally contain much better preserved ultramafic rocks, gabbros, diorites and tonalites with an age range of ~ 300 m.y. (e.g., Nutman et al., 1996). We suggest that some ,of these are much better targets than the Isua supracrustal belt rocks in trying to understand early Archaean mantle evolution via radiogenic isotope tracers (e.g., Bennett et al., 1993). 4.2.2.

146Sm-142Nds y s t e m

Isua supracrustal belt volcanic and sedimentary rocks (Harper and Jacobsen, 1992) as well as other rocks from the Itsaq Gneiss Complex (e.g., McCulloch and Bennett, 1993) have been analysed in the search for 142Nd anomalies. 142Nd forms from now extinct 1465m (half life of only 103 × 106 years). Any preserved ten,estrial 142Nd anomalies would record early planetary fractionation events during the first 300-400 Ma of Earth history, when 1 4 6 5 m w a s still 'alive'. 142Nd anomalies have been measured in lunar, martian and meteorite samples and are evidence of early planetary fractionation in those bodies (e.g., Prinzhofer et al., 1992; Nyquist et al., 1995). Only one terrestrial rock, an Isua supracrustal belt felsic gneiss, has yielded a possible 142Nd (positive) anomaly (Harper and Jacobsen, 1992). The 0.3 e unit size of the possible anomaly is at the limit of that resolvable from 142Nd//144Nd measured on 'normal' terrestrial materials, using the best analytical techniques and equipment available; therefore there has been some doubt expressed whether it exists, although it has been generally accepted (Sharma et al., 1996). The anomalous sample is a felsic gneiss from the B1 felsic formation and is likely to be part of the ~ 3710 Ma volcanosedimentary sequence. The significance of a 142Nd anomaly in this rock is not clear; generation and preservation of a positive ~42Nd anomaly requires that a LREE depleted (high S m / N d ) mantle reservoir formed and became isolated during the first 300 Ma of Earth history and was then preserved until 3710 Ma. At this time melt extraction generated a LREE-enriched felsic volcanic rock with only a moderate eNd value

285

(eNd(3710) = +2.9), rather than an expected extreme composition. In contrast, 142Nd anomalies have not been identified in the older ( > 3800 Ma) gneisses with more radiogenic initial 143Nd/144Nd compositions. We suggest that the possible 142Nd anomaly represents a rogue ancient component (possibly some very ancient detrital zircons?) in this felsic rock rather than the anomaly being indicative of the general state of the ~ 3800-3700 Ma Earth.

Acknowledgements Supported by the Australian National University, Oxford Brookes University, the Royal Society of London, the Carlsberg Foundation, the European Community Science Programme, the Danish Natural Sciences Research Council, the Danish National Research Foundation, the Geological Survey of Greenland, and NERC grant GR3/8879 to C.R.L. Friend. Ole Christiansen and Nunaoil A / S are thanked for logistical assistance. Stephen Moorbath is thanked for discussions on SHRIMP geochronology of the Itsaq Gneiss Complex and Richard Armstrong is thanked for comments on an earlier draft of this paper. The reviews of M. Whitehouse and K. Ludwig are appreciated. Prof. W. Compston is thanked for allowing the SHRIMP to be used for this study.

References Appel, P.W.U., 1983. Tourmaline in the early Archaean Isua supracrustal belt, West Greenland. J. Geol. 92, 599-605. Baadsgaard, H., Nutman, A.P., Bridgwater, D., McGregor, V.R., Rosing, M.T., Allaart, J.H., 1984. The zircon geochronology of the Akilia association and the Isua supracrustal belt, West Greenland. Earth Planet. Sci. Lett. 68, 221-228. Bennett, V.C., Nutman,A.P., McCulloch, M.T., 1993. Nd isotopic evidence for transient, highly depleted mantle reservoirsin the early history of the Earth. Earth Planet. Sci. Lett. 119, 299317. Black, L.P., Williams,I.S.. Compston,W., 1986. Four zircon ages from one rock: The history of a 3930 Ma-old granulite from Mount Sones, Enderby Land, Antarctica. Contrib. Mineral. Petrol. 94, 427-437. Bowring, S.A., Housh, T., 1995. The Earth's early evolution. Science 269, 1535-1540. Bowring, S.A., Williams, I.S., Compston, W., 1989. 3.96 Ga gneisses from the Slave province, Northwest Territories, Canada. Geology 17, 971-975. Bridgwater, D., McGregor, V.R., 1974. Field work on the very

286

A.P. Nutman et a L / Chemical Geology 141 (1997) 271-287

early Precamhrian rocks of the Isua area, southern West Greenland. Rapp. Grcnl. Geol. Unders. 65, 49-54. Bridgwater, D., Rosing, M.T., 1995. Anomalous er~dt values from early Archaean rocks - evidence for major heterogeneities in the early Archaean mantle or the effects of secondary processes. Early History of the Earth, European Union of Geochemistry Meeting in Cambridge, U.K., Progr. Abstr, p. 19. Chauvel, C., Dupr6, B., Jenner, G.A., 1985. The Sm-Nd age of Kambalda volcanics is 500 Ma too old!. Earth Planet. Sci. Lett. 74, 315-324. Claou6-Long, J.C., Compston, W., Cowden, A., 1988. The age of the Kambalda greenstones resolved by ion-microprobe: Implications for Archaean dating methods. Earth Planet. Sci. Lett. 89, 239-259. Claou6-Long, J.C., Compston, W., Roberts, J., Fanning, C.M., 1995. Two Carboniferous ages: A comparison of SHRIMP zircon dating with conventional zircon ages and 4°Ar/39Ar analysis. Soc. Sediment. Geol. Spec. Publ. 54, 3-21. Collerson, K.D., McCulloch, M.T., Nutman, A.P., 1989. Sr and Nd isotope systematics of polymetamorphic Archean gneisses from southern West Greenland and northern Labrador. Can. J. Earth Sci. 26, 446-466. Collerson, K.D., Campbell, L.M., Weaver, B.L., Palacz, Z.A., 1991. Evidence for extreme fractionation in early Archaean ultramafic rocks from northern Labrador. Nature 349, 209-214. Compston, W., Williams, I.S., Myer, C., 1984. U-Pb geochronology of zircons from lunar breccia 73217 using a sensitive high mass-resolution ion microprobe. J. Geophys. Res. 89B, 525534. Compston, W., Kinny, P.D., Williams, I.S., Foster, J.J., 1986. The age and Pb loss behaviour of zircons from the Isua supracrustal belt as determined by ion microprobe. Earth Planet. Sci. Lett. 80, 71-81. Corfu, F., Davis, D.W., 1992. A U-Pb geochronological framework for the western Superior Province. In: Thurston, P.C., Williams, H.R., Sutcliffe, R.H., Stott, G.M. (Eds.), Geology of Ontario. Ont. Geol. Surv., Spec. Vol. 4, 1335-1346. DePaolo, D.J., 1980. Sources of continental crust: Neodymium isotope evidence from the Sierra and Peninsular Ranges, California. Science 209, 684-687. De Wit, M.J., Roering, C., Hart, R.J., Armstrong, R.A., De Ronde, C.E., Green, R.W.E., Tredoux, M., Pederby, E., Hart, R.A., 1992. Formation of an Archaean continent. Nature 357, 553-562. Dymek, R.F., Klein, C., 1988. Chemistry, petrology and origin of banded iron-formation lithologies from the 3800 Ma Isua supracrustal belt, West Greenland. Precambrian Res. 39, 247302. Gill, R.C.O., Bridgwater, D., Allaart, J.H., 1981. The geochemistry of the earliest known basic metavolcanic rocks at Isua, West Greenland: A preliminary investigation. Spec. Publ. Geol. Soc. Aust. 7, 313-325. Goldstein, S.L., O'Nions, R.K., Hamilton, P.J., 1984. A Sm-Nd study of atmospheric dusts and particulates from major river systems. Earth Planet. Sci. Lett. 70, 221-236. Gruau, G., Rosing, M., Bridgwater, D., Gill, R.C.O., 1996. Reset-

ring of Sm-Nd systematics during metamorphism of > 3.7-Ga rocks: implications for isotopic models of early Earth differentiation. Chem. Geol. 133, 225-240. Hamilton, P.J., O'Nions, R.K., Evensen, N.H., Bridgwater, D., Allaart, J.H., 1978. Sm-Nd isotope investigations of Isua supracrustals and implications for mantle evolution. Nature 272, 41-43. Hamilton, P.J., O'Nions, R.K., Bridgwater, D., Nutman, A.P., 1983. Sm-Nd studies of Archaean metasediments and metavolcanics from West Greenland and their implications for the Earth's early history. Earth Planet. Sci. Lett. 62, 263-272. Hamilton, W.B., 1993. Evolution of Archaean mantle and crust. In: Reed, Jr., J.C., Bickford, M.E., Houston, R.S., Link, P.K., Rankin, D.W., Sims, P.K., Van Schmus, W.R. (Eds.), The Geology of North America, C-2, Precarnbrian of the Conterminous United States. Geological Society of America, Boulder, CO, pp. 597-614; 630-636. Harper, C.L., Jacobsen, S.B., 1992. Evidence from coupled 147Sm-143Nd and 1465m-142Nd systematics for very early (4.5 Gyr) differentiation of the Earth's mantle. Nature 360, 728-732. Jacobsen, S.B., Dymek, R.F., 1987. Nd and Sr isotope systematics of clastic metasediments from Isua, West Greenland: Identification of pre-3.8 Ga differentiated crustal components. J. Geophys. Res. 93, 338-354. KrOner, A., Hegner, E., Wendt, J.I., Byerly, G.R., 1996. The oldest part of the Barberton granitoid-greenstoneterrain, South Africa: evidence for crust formation between 3.5 and 3.7 Ga. Precambrian Res. 78, 105-124. Maruyama, S., Masuda, T., Nohda, S., Appel, P., Otofuji, Y., Miki, M., Shibata, T., Hagiya, H., 1992. The 3.9-3.8 Ga plate tectonics on the Earth; evidence from Isua, Greenland. Evolving Earth Symposium, Okazaki, Progr. Abstr., p. 113. McCulloch, M.T., Bennett, V.C., 1993. Evolution of the early Earth: constraints from 142Nd-143Nd isotopic systematics. Lithos 30, 237-255. McCulloch, M.T., Bennett, V.C., 1994. Progressive growth of the Earth's continental crust and depleted mantle: geochemical and geodynamical constraints. Geochim. Cosmochim. Acta 58, 4717-4738. McDonough, W.F., Ireland, T.R., 1993. Intraplate origin of komatiites inferred from trace elements in glass inclusions. Nature 365, 432-434. Michard-Vitrac, A., Lancelot, J., All~gre, C.J., Moorbath, S., 1977. U-Pb ages on single zircons from the early Precambrian rocks of West Greenland and the Minnesota River Valley. Earth Planet. Sci. Lett. 35, 449-453. Moorbath, S., O'Nions, R.K., Pankhurst, R.J., 1973. Early Archaean age for the Isua iron formation, West Greenland. Nature 245, 138-139. Moorbath, S., Allaart, J.H., Bridgwater, D., McGregor, V.R., 1977. Rb-Sr ages of early Archaean supracrustal rocks and Am~tsoq gneisses at Isua. Nature 270, 43-45. Moorbath, S., Taylor, P.N., Jones, N.W., 1986. Dating the oldest terrestrial rocks - fact and fiction. Chem. Geol. 57, 63-86. Moorbath, S., Whitehouse, M.J., Kamber, B.S., 1997. Extreme Nd-isotope heterogeneity in the early Archaean - fact or

A.P. Nuhnan et al./Chemical fiction? Case histories from northern Canada and West Greenland. Chem. Geol. 135, 213-231. Myers, J.S., 1988. Early Archaean Narryer Gneiss Complex, Yilgarn Craton, Western Australia. Precambrian Res. 38, 297307. Nutman, A.P., Collerson, K.D., 1991. Very early Archean crustal-accretion complexes preserved in the North Atlantic Craton. Geology 19, 791-794. Nutman, A.P., Allaart, .J.H., Bridgwater, D., Dimroth, E., Rosing, M., 1984. Stratigraphic and geochemical evidence for the depositional environment of the early Archaean Isua supracrustal belt, southern West Greenland. Precambrian Res. 25, 365-396. Nutman, A.P., Kinny, P.D., Compston, W., Williams, I.S., 1991. SHRIMP U-Pb zircon geochronology of the Narryer Gneiss Complex, Western .$ustralia. Precambrian Res. 52, 275-300. Nutman, A.P., Friend, C.R.L., Kinny, P.D., McGregor, V.R., 1993. Anatomy of an early Archean gneiss complex: 3900 to 3600 Ma crustal evolution in southern West Greenland. Geology 21, 415-418. Nutman, A.P., Hagiya, H., Maruyama, S., 1995. SHRIMP U-Pb single zircon geochronology of a Proterozoic mafic dyke, Isukasia, southern West Greenland. Bull. Geol. Sot. Denm. 42, 16-20. Nutman, A.P., McGregor, V.R., Friend, C.R.L., Bennett, V.C., Kinny, P.D., 1996. The Itsaq Gneiss Complex of southern West Greenland; the world’s most extensive record of early crustal evolution (3900-3600 Ma). Precambrian Res. 78, l-40. Nyquist, H.E., Wiesmann, H., Bansal, B., Shih, C.-Y., Kieth, J.E., formation interval for the Harper, C.L., 1995. ‘46Sm-‘42Nd lunar mantle. Geochim. Cosmochim. Acta 59, 2817-2837. Prinzhofer, A., Papana.stassiou, D.A., Wasserburg, G.J., 1992. Samarium-Neodymium evolution of meteorites. Geochim. Cosmochim. Acta 56, 797-815.

Geology 141 (1997) 271-287

287

Roddick, J.C., van Breemen, O., 1994. U-Pb zircon dating: a comparison of ion microprobe and single grain conventional analyses. In: Radiogenic Age and Isotopic Studies, Geol. Surv. Can. Current Res. 1994-F Rep. 8, pp. l-9. Rose, N.M., Rosing, M.T., Bridgwater, D., 1996. The origin of metacarbonate rocks in the Archaean Isua supracrustal belt, West Greenland. Am. J. Sci. 296, 1004-1044. Rosing, M.T., Rose, N.M., 1993a. The role of ultramafic rocks in regulating the concentrations of volatile and non-volatile components during deep crustal metamorphism. Chem. Geol. 108, 187-200. Rosing, M.T., Rose, N.M., 1993b. Reappraisal of the earliest part of Earth’s stratigraphic record. A sedimentary origin for the Garbenschiefer amphibolite. EOS 74, 656. Rosing, M.T., Rose, N.M., Bridgwater, D., Thomsen, H.S., 1996. Earliest part of the Earth’s stratigraphic record: A reappraisal of the > 3.7 Ga Isua (Greenland) supracrustal sequence. Geology 24, 43-46. Salters, V.J.M., 1996. The generation of mid-ocean ridge basalts from the Hf and Nd isotope perspective. Earth Planet Sci. Lett. 141, 109-123. Sambridge, M.S., Compston, W., 1994. Mixture modeling of multi-component data sets with application to ion-probe zircon ages. Earth Planet. Sci. Lett. 128, 373-390. Sharma, M., Papanastassiou, D.A., Wasserburg, G.J., Dymek, R.F., 1996. The issue of the terrestrial record of ‘46Sm. Geochim. Cosmochim. Acta 60, 2037-2047. Song, B., Nutman, A.P., Liu, D., Wu, J., 1996. 3800 to 2500 Ma crustal evolution in the Anshan area of the Liaoning Province, northeastern China. Precambrian Res. 78, 79-94. Vervoort, J.D., Patchett, P.J., Gehrels, G.E., Nutman, A.P., 1996. Constraints on early Earth differentiation from hafnium and neodymium isotopes. Nature 379, 624-627.