V. C. Bennett

The Itsaq Gneiss Complex of southern West Greenland; the world's most extensive record of early crustal evolution (3900-3600 Ma)

The Itsaq Gneiss Complex of southern West Greenland contains the best-preserved occurrences of ⩾ 3600 Ma crust. Its known area is ∼ 3000 km2 with almost continuous exposure in some places. SHRIMP UPb zircon geochronology shows that the gneiss complex had a complicated early history, having been added to, and modified in, several events starting at ∼ 3900 Ma. Supracrustal, mafic and ultramafic rocks comprise approximately 10% of the complex and range in age from ⩾ 3870 to ∼ 3600 Ma. A large portion of the Isua supracrustal belt and some other bodies contain sequences of LREE-enriched, mafic (locally pillow-structured) to felsic volcanic and volcaniclastic rocks (some deposited from turbidity currents) and abundant diverse chemical sediments. These sequences might have formed in an environment analogous to present-day volcanic arcs. Other (mostly ⩾ 3800 Ma?) units dominated by LREE-depleted, high Ti/Zr mafic rocks (as found in komatiites and komatiitic basalts free of crustal contamination) with layers of banded iron formation might have been derived from volcanic edifices formed as a result of plume activity. Only in the youngest supracrustal sequences (∼ 3600 Ma) are detrital sediments derived from mixed-provenance clastic sources an important component. Layered anorthosite, gabbro and peridotite units (some ⩾ 3800 Ma) derived from large deep crustal intrusions are also widespread. In addition there are massive dunites and harzburgites. In the rare cases where their original mineralogy and texture are preserved, these contain aluminous spinel, indicating equilibration in the lowermost crust or upper mantle. Approximately 90% of the complex consists of grey gneisses, the dominant protoliths having been tonalites and less abundant trondhjemites, quartz-diorites, diorites and granodiorites. The protoliths were intruded in several events starting at ∼ 3870 Ma. Like other suites of Archaean grey gneisses, they were formed by partial melting, probably in an arc environment, of buried (subducted?) amphibolite, leaving residual hornblende ± garnet. Granites form approximately 10% of the complex. The oldest are 3650 Ma leucogranites, which probably formed by deep crustal anatexis of predominantly tonalitic gneisses. There are also ∼ 3625 Ma augen granites and ferrogabbros, whose chemistry resembles that of some A-type, within-plate granite suites. The evolution of the Itsaq gneiss complex is marked by increasing compositional diversity with time. Pre-3650 Ma lithologies are predominantly mafic and ultramafic rocks, tonalites and diorites. The first recorded regional metamorphic event occurred at 3650 Ma, and was marked by localised partial melting and intrusion of leucogranites. This might record crustal thickening, brought about by collision of different blocks of tonalite-dominated crust. A further thermal event occurred just after 3600 Ma, as shown by PbPb titanite and feldspar ages and local intrusion of granites. Deposition of sediments at ∼ 3600 Ma which were derived from mixed-age sources suggests their derivation from an extensive block of “continental” crust.

Progressive growth of the Earth's continental crust and depleted mantle: Geochemical constraints

A three reservoir model, consisting of the continental crust, depleted mantle and a more primitive mantle reservoir is used as a basis to account for both the present-day as well as the evolving isotopic compositions of the Earths crust and mantle. The rate of growth of the continental crust is used as an input parameter to constrain the concomitant growth and evolution of the depleted mantle reservoir. Recycling of large volumes of bulk continental crust into the mantle is not considered to be an important process, nor is the existence of an additional major enriched reservoir in the early Archean mantle. This relatively simple model of progressively growing continental crust extracted from an increasing volume of depleted mantle can account for the positive ϵNd values which characterise the Archean depleted mantle, the evolution of the strontium, neodymium, hafnium and lead isotopic systems as well as the budgets of a wide range of trace elements in the continental crust and depleted mantle; e.g., the nonprimitive Rb/Cs, Nb/U and Th/U observed in MORBs and OIBs as well as the Sm/Nd in the crust and mantle can be reproduced. The Re-Os isotopic system is most sensitive to the formation of basaltic crust in the early Archean and can potentially provide definitive limits on the volumes of stored mafic or ultramafic crust in the mantle. To account for the relatively radiogenic 206Pb204Pb ratios of modern MORB it is necessary to assume that the overall efficiency of transfer of uranium from the mantle to the crust has decreased markedly since the Archean, a proposed consequence of slab dehydration rather than slab melting. In the post-Archean period, recycling of hydrothermally altered oceanic crust is thus likely to have had a significant influence on the lead isotopic systematics of the mantle. In the model described here it is assumed that the volume of depleted mantle increases in a stepwise manner which is arbitrarily linked to major episodes of rapid crustal formation. From observed crustal age distribution patterns, episodes of rapid crustal formation with high ϵNd values occur at ~3600 Ma, ~2700 Ma and 1800 Ma. Thus, in our first order calculations, the crust is modelled as being extracted from ~10% of the mass of the mantle (upper 220 km) from 4500 Ma to 3600 Ma, ~20% (upper 410 km) from 3600 Ma to 2700 Ma, ~30% (upper 660 km) from 2700 Ma to 1800 Ma and 40% to 50% (upper 800–1000 km) of the mantle from 1800 Ma to the present-day. This type of model with growing volumes of both continental crust and depleted mantle has the general effect of buffering the isotopic and trace element composition of the upper mantle through time to an approximately constant, but incompatible element depleted chemical composition.

Nd isotopic evidence for transient, highly depleted mantle reservoirs in the early history of the Earth

Ma to 3760 Ma metadiorites and tonalites, and older mafic inclusions contained within these rocks from southern West Greenland, have a range of initial end values from + 4.5 to --4.5. The extremely positive initial values determined for nine of the fourteen early Archean samples demonstrate that they were derived from a LREE depleted mantle reservoir which had an end value of ~ +4 prior to 3800 Ma. These rocks provide the best constrained evidence for the existence of highly LREE fraetionated mantle reservoirs in the early Earth. In contrast, previous studies of younger 3700 and 3730 Ma tonalites from the same region show that the latter have lower initial ENd values ranging from -4.6 to + 2.0. The highest initial ENj values determined for 3450 Ma komatiites from western Australia and southern Africa are +2 to +4, and for worldwide ca. 2700 Ma greenstone belts are + 2 to + 5. These values are well below the values for eNd(3450 Ma) (> + 8) and eyd(2700 Ma) (> + 15) that would be expected for the continued evolution of a highly depleted reservoir with Sm/Nd similar to that which produced the ca. 3.8 Ga Greenland metadiorites and tonalites. Thus, if the source region for the oldest Greenland gneisses was the prevalent upper mantle composition, it must have been an ephemeral feature later modified either by mixing with less-depleted mantle or with recycled LREE enriched, negative ENd crust. The generation of highly positive end values by 3.8 Ga requires differentiation of an extremely LREE fractionated reservoir very early in Earths history. The Archean Nd isotope data may record the isolation, depletion by crustal extraction and subsequent partial rehomogenization of limited portions of the upper mantle, or alternatively may reflect transient large-scale differentiation processes unrelated to crustal extraction such as might occur in a terrestrial magma ocean.

The persistence of off-cratonic lithospheric mantle : Os isotopic systematics of variably metasomatised southeast Australian xenoliths

The ReOs systematics of a suite of spinel peridotite xenoliths from southeast Australia provide insight into the effects of melt extraction and metasomatism on Re and Os and place strong constraints on the evolution and long-term stability of post-Archean lithospheric mantle in a tectonically complex region. Data from variably melt-depleted and non-modally metasomatised xenoliths demonstrate that Re abundances are largely controlled by melt extraction, with Re similarly distributed to Os. Ratios of ReOs correlate strongly with indices of melt extraction (e.g. Al2O3, Ni and Yb), and with the calculated bulk partition coefficient of Re, comparable to that of Yb over a large range of melt extraction (∼ 4–20%). Hence, if Re is controlled by sulfide phases, sulfur:clinopyroxene ratios should remain essentially constant over large degrees of melt extraction. Eight of the 24 samples analysed were wehrlites or apatite-bearing peridoties, representing residual peridotite which has interacted with a carbonatitic melt. In comparison with the non-modally metasomatised xenoliths, these samples show no evidence for disturbance of Os isotopic composition, or addition of Re or Os during metasomatism. The entire suite provides a 220 km long, WNW-ESE lithospheric mantle transect, east of, and perpendicular to, the presumed Australian Precambrian shield margin. The Os model ages indicate at least three episodes of mantle depletion: ca. 1960 Ma, 800–1000 Ma and < 500 Ma. The older age is found only in the two westernmost localities where a subset of four samples define a ReOs age of 1959 ± 100 Ma, with an initial γOs = +0.2. Although the oldest exposed rocks in the region are of Cambrian age, and the presence of early Proterozoic basement is highly contentious, the Os isotopic data require that early Proterozoic basement extends some 400 km further east than the easternmost exposed early Proterozoic crust. Model ages of 800–1000 Ma are common to all but one locality, indicating at least two melt extraction events in the western localities. Paleozoic ages are only identified in the eastern localities, suggesting the lithospheric mantle becomes younger to the east. Importantly, this and other ReOs isotopic studies provide increasing evidence for the long-term stability and persistence of lithospheric mantle of Proterozoic as well as of Archean age.

Behaviour of Platinum-group elements in the subcontinental mantle of eastern Australia during variable metasomatism and melt depletion

Increasing recognition of complexities in the Platinum-group element (PGE) and Re concentration patterns in mantle samples are challenging the view of chondritic relative abundances in the upper mantle. To investigate the possible causes of PGE abundance variations, a suite of east Australian, mantle-derived, spinel peridotite xenoliths, ranging from fertile lherzolites to depleted harzburgites, and including apatite ± phlogopite ± amphibole bearing samples, have been analysed for their whole rock PGE and Re abundances. Whole rock abundances for 21 samples, combined with mineral separate analyses of 2 xenoliths, are presented to constrain the distribution of the PGEs and Re, their inherent heterogeneity at difference scales, and their behaviour during both melt extraction and metasomatism. Fertile (>2.9 wt% Al2O3) xenoliths have broadly chondritic relative PGE abundances, with the significant exception of positive Rh anomalies and variable negative Os anomalies. The high Rh abundances cannot be attributed to melt extraction or metasomatism. Bulk mineral separate PGE-Re analyses of 2 fertile xenoliths indicate less than 6% of the whole rock PGE budget resides in either silicate or oxide (spinel) phases. The remainder of the PGEs, and at least 80% of the whole rock Re budget, are sited in acid-leachable sulfides and less soluble trace phases such as PGE-sulfides or alloys. Individual PGEs partition into different trace phases resulting in small scale heterogeneity of both PGE ratios and concentrations on the order of 8%–20%. Although these trace phases may be present within the mantle, it is more likely at least some exsolved from monosulfide solid solutions at low temperatures. Ir and Rh abundances are consistent with compatible behaviour during melt extraction, whereas Ru, Pt and Pd abundances are consistent with slightly incompatible behaviour and can be modeled by assuming all reside in sulfides within the mantle, with DsulfRu ∼ DsulfPt > DsulfPd. Comparison of PGE abundances between ‘dry’ xenoliths and modally metasomatised xenoliths, suggests the PGEs are not significantly mobilised during interaction with carbonate melts or during metasomatism leading to hydrous mineral growth. Given the problems of various types of secondary alteration processes, including melt extraction and surficial alteration that commonly affect xenoliths, and as within-locality heterogeneity is on a comparable order to any proposed regional heterogeneity, it may be premature to define significant regional differences, or ‘primary’ non-chondritic PGE patterns in lithospheric peridotites.

SHRIMP zircon ages constraining the depositional chronology of the Hamersley Group, Western Australia

The Mt Bruce Supergroup of Western Australia was laid down between ca 2.8 Ga and ca 2.2 Ga in the Hamersley Basin, unconformably over a basement of the older, granite‐greenstone, component of the Pilbara Craton. The Mt Bruce Supergroup consists of three groups: the Fortescue Group, Hamersley Group and Turee Creek Group in upward sequence. The Hamersley Group, which is divided into eight formations, has a general thickness of ∼2.5 km, and is characterised by major banded iron‐formation (BIF) units. Reported here are SHRIMP U–Pb zircon results (406 grain analyses) from 13 samples taken from the Hamersley Group and near the top of the underlying Fortescue Group. In combination with SHRIMP results previously published from 12 Hamersley Group samples, the present results provide significant new constraints on the depositional chronology of the group, and suggest that the average (compacted) depositional rates of each of the main depositional lithologies (BIF, carbonate, shale) were ∼180 m per million years, 12 m per million years and 5 m per million years, respectively. Some recently published SHRIMP ages from the Joffre Member differ slightly from those that are interpreted from the present data, and it is suggested that the two datasets may be reconciled if non‐zero‐age Pb loss is taken into account. The total body of zircon U–Pb age data from the Fortescue and Hamersley Groups is consistent with a model involving continuous accumulation of basin‐fill for at least 330 million years, from ca 2780 Ma to the top of the Hamersley Group at ca 2449 Ma. The word ‘continuous’ in this context means that there may have been no breaks in deposition longer than 1 million years. However, this model is not proven, and a major challenge for future work is to measure the length of any proposed non‐depositional intervals.

Rhenium and platinum group element abundances correlated with mantle source components in Hawaiian picrites: sulphides in the plume

Core addition and crustal recycling models that seek to explain the radiogenic Os isotopic compositions of primitive Hawaii tholeiites predict distinctive geochemical consequences for chalcophile and siderophile element abundances in the mantle plume. To test these models and to improve our understanding of compositional variability in the Hawaiian plume, the platinum group element (PGE) and Re contents of primitive shield picrites from several Hawaiian volcanoes were measured. PGE abundances span a large range, from similar to MORB for a picrite from Koolau, to compositions similar to those of basaltic komatiites for picrites from Kilauea and Loihi. Re concentrations range from 0.25 to 0.95 ng/g and with a mean of 0.73 ng/g, higher than previously compiled global averages for ocean island basalts (OIB) (0.38 ng/g) and closer to average MORB (0.98 ng/g) than previously recognised. Some subaerial tholeiites, notably from Kilauea and Mauna Kea, have anomalously low Re abundances and high Cu/Re ratios, possibly reflecting Re loss upon eruption or during degassing of shallow magma chambers. These data show that the PGE and primary Re contents of primitive Hawaiian picrites are well correlated with isotopic compositions of these lavas, linking the PGE and Re characteristics directly with source features of the mantle plume. However, mixing models that describe the isotopic effects of core addition and crustal recycling do not account for the PGE and Re abundances. The range of PGE and Re contents in these lavas does not appear to reflect abundance variations in the plume components, but some aspect of the melting process that is linked to source characteristics of the plume. One possibility is that the PGE and Re characteristics of Hawaiian tholeiites may reflect variable amounts or compositions of residual sulphide during melting. In this scenario, the high PGE and Re contents of Kilauea and Loihi picrites may be indicating a relatively small amount of residual sulphide during melting, whereas the low PGE and Re contents of Koolau primitive magmas may be indicating greater amounts of residual sulphide in the plume. The systematic compositional variations of PGE and Re in primitive tholeiites must be accounted for by any model for the origin of the Hawaiian plume.

Sm-Nd and U-Pb Zircon Isotopic Constraints On the Provenance of Sediments From the Amadeus Basin, Central Australia - Evidence for Ree Fractionation

The Amadeus Basin of central Australia is a Late Proterozoic to Late Palaeozoic ensialic depositional basin located between two Proterozoic basement blocks of different Nd crustal formation ages, the older Arunta Block (TDMNd = 2.0 to 2.2 Ga, U-Pb zircon ages = 1.5–1.9 Ga) to the north and the younger Musgrave Block (TDMNd = 1.7 to 1.9 Ga, U-Pb zircon ages = 1.0–1.7 Ga) to the south. Initial Nd isotopic compositions of the Amadeus Basin sediments generally fall into the region defined by the evolutionary trajectories of these two basement blocks, indicating that the sediments are dominated by essentially two source components: the older Arunta Block [147Sm/144Nd ≈ 0.114 and ϵNd(0) ≈ −20.4] and younger Musgrave Block [147Sm/144Nd ≈ 0.118 and ϵNd(0) ≈ −14.7] and/or their equivalents. Stratigraphically higher sediments plot more closely to the evolutionary trajectory of the Musgrave Block. This result, together with the average provenance ages, indicates the proportion of material from the Musgrave Block increases as the sediments become younger. The Sm-Nd isotopic data preclude a substantial involvement of Archaean sources. U-Pb zircon ion probe analyses of the Late Proterozoic Heavitree Quartzite and the late Cambrian Goyder Formation are consistent with the Sm-Nd isotopic constraints. Most of the zircons from the basal Heavitree Quartzite give concordant U-Pb zircon ages of 1500 to 1900 Ma, consistent with derivation from an Arunta-type source, whilst zircons from the younger Goyder sandstone give six discrete U-Pb zircon age groups of 511 ±20, 615 ± 15, 960 ± 109, 1190 ± 54, 1633 ±24 and 1878 ±48 Ma, indicative of a dominantly Musgrave-type source with a minor contribution from the Arunta Block or reworked preexisting Amadeus Basin sediments. The two youngest U-Pb zircon ages probably indicate some contribution from Late Proterozoic-Cambrian volcanics in central Australia or a previously unrecognised younger local source in the basement blocks. The 511 ±20 Ma constrains the maximum depositional age of the sediment. The Amadeus Basin sediments have a surprisingly wide range of 147Sm/144Nd ratios (0.077–0.136), irrespective of grain size and rock types. In addition, the calculated TDMNd values for samples from the same stratigraphic unit are well correlated with 147Sm/144Nd ratios, with samples of different stratigraphic units forming sub-parallel arrays. These phenomena are interpreted in terms of REE fractionation and preferential sorting of preexisting REE-rich phases during sedimentary recycling and deposition. This process can lead to either an increase or a decrease in 147Sm/144Nd ratio of a sediment relative to its source and therefore an erroneous estimate of its TDMNd provenance age. A theoretical model is developed which accounts for the observed correlation between TDMNd and 147Sm/144Nd ratios for suites of Stratigraphically related sediments and allows a reliable estimate to be made of the provenance age of individual stratigraphic units, assuming the average 147Sm/144Nd ratio of the source is known or can be estimated. The provenance age for each stratigraphic unit of the Amadeus Basin has been calculated using this method. The Sm-Nd isotopic systematics in the Amadeus Basin sediments suggest that 1. (1) REE fractionation during sedimentary recycling can in some cases be an important factor and needs to be considered; 2. (2) the provenance of the Amadeus Basin sediments was controlled by local source regions and cannot be used to infer large-scale continental averages. Hence, some caution must be evoked in using Sm-Nd isotopic constraints from restricted sampling for more generalised models of crustal evolution.

Rhenium systematics in submarine MORB and back-arc basin glasses: laser ablation ICP-MS results

Abstract Rhenium and other trace elements, including the moderately chalcophile elements Mo and Cu, were determined for 37 submarine basaltic glasses from the Lau and Coriolis Troughs (CT) back-arc basins and Woodlark marginal basin, as well as 30 mid-ocean ridge basalt (MORB) glasses from the Pacific and Atlantic Oceans, using laser ablation ICP-MS. Rhenium is strongly positively correlated with Yb for all these submarine basaltic glasses. Enriched (E-) and normal (N-) MORB as well as Kings Triple Junction samples show similar correlations with constant Yb/Re ratios, indicating that Re and Yb exhibit similar compatibility during melt evolution [Chem. Geol. 139 (1997) 185]. In contrast, samples from the East and Central Lau Spreading Centers have much higher ratios compared with MORB samples and form steeper arrays on Re–Yb variation diagrams, similar to komatiites. More incompatible element-depleted samples including those from the Lau and Woodlark Basin spreading centers and the more depleted (D-) MORB samples are also distinguished from E- and N-MORB and samples from Kings Triple Junction and Coriolis Troughs Basin on the basis of their higher Cu/Re ratios. These observed elemental systematics are interpreted to reflect progressive melting of depleted mantle, where previous melting events result in the elimination of sulfides in the source regions of the depleted samples. Using the determined Yb/Re and Ce/Mo ratios and assuming that the abundances of Yb and Ce are 10% and 40% reduced in the DMM compared to the primitive mantle (PM), average concentrations of 0.12 ppb for Re and 34 ppb for Mo are estimated for the DMM. The partition coefficients of the analysed moderately incompatible elements are in the order of Mo

Constraints on mantle evolution from 187Os/188Os isotopic compositions of Archean ultramafic rocks from southern West Greenland (3.8 Ga) and Western Australia (3.46 Ga)

Initial 187Os/188Os isotopic compositions for geochronologically and geologically well -constrained 3.8-Ga spinel peridotites from the Itsaq Gneiss Complex of southern West Greenland and chromite separates from 3.46-Ga komatiites from the Pilbara region of Western Australia have been determined to investigate the osmium isotopic evolution of the early terrestrial mantle. The measured compositions of 187Os/188Os(0) = 0.10262 ± 2, from an olivine separate, and 0.10329 ± 3, for a spinel separate from ∼3.8-Ga peridotite G93/42, are the lowest yet reported from any terrestrial sample. The corrections for in situ decay over 3.8 Ga for these low Re/Os phases are minimal and change the isotopic compositions by only 0.5 and 2.2% for the spinel and the olivine, respectively, resulting in 187Os/188Os(3.8 Ga) = 0.1021 ± 0.0002 and 0.1009 ± 0.0002, respectively. These data extend direct measurement of Os isotopic compositions to much earlier periods of Earth history than previously documented and provide the best constraints on the Os isotopic composition of the early Archean terrestrial mantle. Analyses of Pilbara chromites yield 3.46-Ga mantle compositions of 0.1042 ± 0.0002 and 0.1051 ± 0.0002. These new data, combined with published initial Os isotopic compositions from late Archean and early Proterozoic samples, are compatible with the mantle, or at least portions of it, evolving from a solar system initially defined by meteorites to a modern composition of 187Os/188Os(0) = 0.1296 ± 0.0008 as previously suggested from peridotite xenolith data ( Meisel et al., 2001); the associated 187Re/188Os(0) = 0.435 ± 0.005. Thus, chondritic 187Os/188Os compositions were a feature of the upper mantle for at least 3.8 billion years, requiring chondritic Re/Os ratios to have been a characteristic of the very early terrestrial mantle. In contrast, nonchondritic initial compositions of some Archean komatiites demonstrate that Os isotopic heterogeneity is an ancient feature of plume materials, reflecting the development of variable Re/Os mantle sources early in Earth history. The lower average 187Os/188Os = 0.1247 for abyssal peridotites (Snow and Reisberg, 1995) indicate that not all regions of the modern mantle have evolved with the same Re/Os ratio. The relative sizes of the various reservoirs are unknown, although mass balance considerations can provide some general constraints. For example, if the unradiogenic 187Os/188Os modern abyssal peridotite compositions reflect the prevalent upper mantle composition, then the complementary high Re/Os basaltic reservoir must represent 20 to 40% by mass of the upper mantle (taken here as 50% of the entire mantle), depending on the mean storage age. The difficulties associated with efficient long-term storage of such large volumes of subducted basalt suggest that the majority of the upper mantle is not significantly Re-depleted. Rather, abyssal peridotites sample anomalous mantle regions. The existence of 3.8-Ga mantle peridotites with chondritic 187Os/188Os compositions and with Os concentrations similar to the mean abundances measured in modern peridotites places an upper limit on the timing of a late accretionary veneer. These observations require that any highly siderophile element -rich component must have been added to the Earth and transported into and grossly homogenized within the mantle by 3.8 Ga. Either large-scale mixing of impact materials occurred on very short (0–100 myr) timescales or (the interpretation preferred here) the late veneer of highly siderophile elements is unrelated to the lunar terminal cataclysm estimated to have occurred at ∼3.8 to 3.9 Ga.