Insight into the interactions between crust and hydrosphere, through the protracted evolution of the Greenland Shield, can be provided by oxygen isotopes in the mineral remnants of its denuded crust. Detrital zircons with ages of 3900 Ma to 900 Ma found within an arkosic sandstone dike of the Neoproterozoic (?Marinoan) Mørænesø Formation, North Greenland, provide a time-integrated record of the evolution of part of the Greenland Shield. These zircon grains are derived from a wide variety of sources in northeastern Laurentia, including Paleoproterozoic and older detritus from the Committee-Melville orogen, the Ellesmere-Inglefield mobile belt, and the subice continuation of the Victoria Fjord complex. Archean zircon crystals have a more restricted range of δ18OSMOW values (between 7.2‰ and 9.0‰ relative to standard mean ocean water [SMOW]) in comparison to Paleoproterozoic 1800–2100 Ma grains, which display significant variation in δ18OSMOW (6.8‰–10.4‰). These data reflect differences in crustal evolution between the Archean and Proterozoic Earth. Through time, remelting or reworking of high δ18O materials has become more important, consistent with the progressive emergence of buoyant, cratonized continental lithosphere. A secular increase in the rate of crustal recycling is implied across the Archean-Proterozoic boundary. This rate change may have been a response to differences in the composition of sediments and/or the stabilization of continental crust.
One Eoarchean oscillatory-zoned zircon grain, free of cracks and with concordant U-Pb systematics, has an elevated δ18OSMOW value of 7.8‰. This is interpreted to reflect a primary magmatic signature, supporting the presence of heavy oxygen that may be compatible with a hydrosphere on early Earth, as previously determined only from Jack Hills zircons.
The Greenland Shield can be divided into three basement provinces, namely: (A) Archean rocks (3100–2600 Ma, with local older units up to ca. 3900 Ma) that have been essentially unaffected by Proterozoic or later orogenic activity; (B) Archean terranes that were reworked in the early Proterozoic ca. 1850 Ma; and (C) terranes composed of juvenile early Proterozoic crust (2000–1750 Ma old; Henriksen et al., 2000). Direct study is limited to exposures around the edge of the Greenland Ice Sheet. However, a complementary record of the crustal evolution of these basement blocks is available from detrital (sedimentary rock–hosted) zircon crystals that are derived from them. Zircon is a refractory mineral and is a common accessory phase in intermediate to acid igneous rocks, high-grade metamorphic rocks, and clastic sedimentary rocks. Zircon grains can survive multiple episodes of magmatic and metamorphic reworking as well as transport via the sedimentary rock cycle. Detrital zircon grains carry with them important information on crustal evolution, and a multigrain population may elucidate the evolution of vast tracts of crust (e.g., Fedo et al., 2003). Thus, they have potential to preserve a more complete record of igneous episodes than the fragmentary exposed basement of a region (e.g., Knudsen et al., 1997).
Analysis of oxygen isotopes in detrital igneous zircon crystals of known age can be used to trace both the evolution of crustal recycling and crust-mantle interaction (e.g., Valley et al., 2005). Zircon crystals diffuse oxygen slowly, even under high-temperature conditions, and hence their measured δ18O value can approximate the crystallization value, provided that no late alteration has occurred (Peck et al., 2001). Incorporation of high δ18O material, for example, rocks and/or minerals altered by low-temperature near-surface processes, will increase the δ18O value of a melt. Hence, the zircon crystallized from such melts will also have elevated δ18O values. For example, granites with a dominant metasedimentary component have bulk-rock δ18OSMOW values of 9%–15‰ relative to standard mean ocean water (SMOW) (O'Neil and Chappell, 1977) and would crystallize zircon in the range ∼7‰–13‰ (e.g., Lackey et al., 2005), whereas zircons in equilibrium with mantle-derived melts have δ18OSMOW values of ∼5.3‰ ± 0.6‰ (two standard deviations; Valley, 2003). Changes in whole-rock δ18O values do occur during temperature decreases accompanying magmatic differentiation (Valley, 2003), but these effects are small over the typical range of magmatic zircon crystallization temperature. Therefore, significant deviations of δ18O in zircon from mantle values are primarily the consequence of magma interaction with materials altered by low-temperature near-surface processes. A δ18O value of ∼6.3‰–6.5‰ is commonly taken as the value above which incorporation of an elevated δ18O component is implied (Valley, 2003; Cavosie et al., 2005; Kemp et al., 2006). It is noteworthy that zircon grains crystallized in both young ocean crust (e.g., Cavosie et al., 2009) and lunar melts (e.g., Nemchin et al., 2006b) have δ18OSMOW values less than this limit.
A compilation of δ18O values in igneous zircons of known age has been used to trace the evolution of intracrustal recycling and crust-mantle interaction through Earth history (Valley et al., 2005). This data set raises two significant questions: (1) Does it truly reflect global processes, given that early Earth tectonic styles may have varied in space and time, as they do on modern Earth, with crust formed through plumes and through subduction and/or accretion at the same time in different places (e.g., Foulger et al., 2005)? (2) Was the Archean homogeneous in terms of crustal recycling from 4400 to 2500 Ma (as indicated by Valley et al., 2005, their Fig. 4), which is in apparent contrast to many models for Earth evolution and crustal growth over this period (e.g., Collerson and Kamber, 1999)?
In order to help address these questions, we analyzed oxygen isotopes in detrital zircons from a sedimentary sample of the Neoproterozoic (?Marinoan) Mørænesø Formation (CKG38) from North Greenland (Fig. 1). Results from this study contribute data from previously unsampled orogens to the terrestrial oxygen isotope data set. This permits a more thorough evaluation of crustal maturation on a global scale, and it allows us to test the global applicability of the δ18O evolution model (Valley et al., 2005).
Significantly, the data illustrate a more gradual secular change in magmatic δ18O than has been hitherto recognized. This is used to suggest an ongoing increase in the remelting or reworking of high δ18O supracrustal materials through Earth history, consistent with a secular increase in the volume of continental crust (e.g., McCulloch and Bennett, 1994; Collerson and Kamber, 1999). Additionally, a single analysis from the data set confirms heavy oxygen in Eoarchean igneous rocks, previously only reported from the Yilgarn craton of Western Australia (Peck et al., 2001).
The Proterozoic Morænesø Formation is preserved in a series of paleovalleys and records a range of glacial and postglacial sedimentary processes (Collinson et al., 1989). The formation lies unconformably on the Mesoproterozoic Inuiteq Sø Formation and is overlain by the Portfjeld Formation, which, at least in part, may extend back into the late Neoproterozoic (Dewing et al., 2004). The Morænesø Formation is thus broadly constrained in age between ca. 1380 Ma, the age of dolerites that cut the Inuiteq Sø Formation, but not the Morænesø Formation (Upton et al., 2005), and the late Neoproterozoic (ca. 590–580 Ma). The stratigraphy of the deposit, which consists of carbonates directly overlying diamictites, has prompted correlations with Neoproterozoic Marinoan-Varanger glaciations (Surlyk, 1991).
The sample from the Mørænesø Formation selected for this study (CKG38) represents a sandstone dike that cuts down into diamictites and is interpreted to represent a Neoproterozoic frost polygon crack filled by sand. Its zircon age distribution is typical of the Mørænesø Formation as a whole, consisting of a characteristic Mesoproterozoic to Neoproterozoic detrital zircon population, as well as prominent Paleoproterozoic (ca. 1.9–2.1 Ga) and Neoarchean populations, together with relatively minor Eoarchean-Mesoarchean components (Kirkland et al., 2009). The Mørænesø Formation is an ideal target for a regional oxygen isotope study because its detrital zircon provenance (Kirkland et al., 2009) records a wide variety of sources within northeast Laurentia, including Paleoproterozoic and older detritus from the Committee-Melville orogen, the Ellesmere-Inglefield mobile belt, and the subice continuation of the Victoria Fjord complex (e.g., Dawes, 2006; Nutman et al., 2008).
Throughout the Mørænesø Formation, the Mesoproterozoic 1000–1100 Ma Grenville detrital population appears together with ca. 1600 Ma Labradorian detritus, and thus it has been inferred that one provenance component was derived from the Grenville orogen far to the south (Kirkland et al., 2009). Rainbird et al. (1992) also reported similar detrital populations in the Minto Inlier within the Canadian high Arctic and also inferred detrital transport northward, away from the denuding Grenville orogen. The expansive source region represented by the Mørænesø Formation reflects the ability of the Marinoan fluvioglacial environment to transport material over considerable distances (Allen and Leather, 2006). In addition, it is possible that the Mørænesø Formation contains some aeolian material.
Oxygen isotope data were obtained during three separate analytical sessions; during each session, analyses were performed in an automated sequence using a Cameca IMS1270 ion microprobe. After U-Th-Pb data acquisition and prior to oxygen isotope analysis, the zircon grains were lightly polished. As an additional precaution against bias from implanted primary beam oxygen ions, analytical locations clearly within the same crystal domain (as identified in cathodoluminescence [CL] images) but not overlapping the geochronology pit were used. The basic instrument setup and analytical procedure closely followed those described by Nemchin et al. (2006b). A 20 keV Cs+ primary beam (+10 kV primary, −10 kV secondary) of ∼7 nA was used in aperture illumination mode to sputter an ∼15 μm sample area. Charge compensation was provided by a normal incidence electron gun. Analytical data were acquired under fully automated runs that consisted of an ∼2 min presputter with a raster of 25 μm, followed by field aperture, entrance slit and mass centering, using the 16O signal. After centering, 4 min of signal integration in two Faraday detectors, operating at a common mass resolution of ∼2500, measured the 16O and 18O isotopes. A minor linear drift correction (<1 ppm/run) was applied to the data, where applicable, in order to minimize the external error on the standards. Data were normalized to the 91500 zircon standard assuming a δ18OSMOW value of +9.86‰ (Wiedenbeck et al., 2004). Results obtained in three separate analytical sessions are presented in Figure 2 and Table 1 in sequential order of analysis. Overall uncertainties on δ18O values are reported at the 1σ level and include a propagated external reproducibility from the standard measurements in each session that ranged from 0.21‰ to 0.24‰ (std. dev.).
All analytical spots were located within 6 mm of the standard and were positioned in a standard east-to-west orientation to avoid a small systematic bias observed previously in samples oriented north-south (Whitehouse and Nemchin, 2009). In all analytical sessions, there was a <50 digital units difference in DT1 centering between the sample and standard in the y direction and <100 units difference in the x direction, indicating relative flatness of the sample mount over the analyzed area.
Imaging of grains using cathodoluminescence, backscattered (BSE) and secondary electrons (SE), and reflected light optical microscopy both before and after analysis was conducted to assess data quality and significance. In particular, Cavosie et al. (2005) demonstrated that abnormal pit shape was a proxy for analyses that sampled either various amounts of fractured zircon, metamict zircon, or mounting epoxy. Using this criterion, analytical locations that could be from mixed domains, nonmagmatic zircon, inclusions, or on fractures, accounting for approximately one third of analyses, were excluded. The remaining oxygen isotope analyses from CKG38 zircons represent crystal domains (15–20 μm) that exhibit well-preserved magmatic zonation (Fig. 3) and have yielded concordant U-Pb ages (Kirkland et al., 2009); they are therefore inferred to represent magmatic values.
The oxygen isotope results obtained from sample CKG38 can be considered in three age sets corresponding to periods of supercontinent assembly (Condie, 1998, 2004): ca. 1000–1400 Ma Rodinia; ca. 1800–2100 Ma Paleoproterozoic; and ca. 2500–3000 Ma Archean (Table 2; Fig. 4). The Archean zircons display a relatively restricted range of δ18O values of between 7.2‰ and 9.0‰ (average 8.1‰ ± 1.0‰, 2 std. dev.), in comparison to Paleoproterozoic grains that display variation in δ18O from 6.8‰ to 10.4‰, with an average of ∼8.0‰ ± 2.2‰ (2 std. dev.) for crystals interpreted to retain magmatic values (Fig. 4). Rodinian-aged grains (1000–1400 Ma) span a considerable range of δ18O values, from ∼6.5‰ to 11.9‰ (Fig. 2), within the range reported by Peck et al. (2000) from the Grenville Province. Two late Grenvillian (ca. 1050 Ma) crystals have δ18O values greater than 10‰, and one ca. 1180 Ma grain has a value of 11.5‰.
The Paleoproterozoic and Archean detrital zircon populations studied herein most probably originated from the North-West and western North Greenland regions (Kirkland et al., 2009), which include the Committee-Melville orogen (Archean), the Ellesmere-Inglefield mobile belt (Paleoproterozoic), and the subice continuation of the Victoria Fjord complex (Archean) (Dawes, 2006; Nutman et al., 2008). The crystalline rocks in this region of Greenland record an apparently continuous increase in oxygen isotopic values from a relatively restricted range in the Archean, with values less than ∼9‰, to more variable δ18O values in the Proterozoic, including much higher values of ∼12‰. As shown in Figure 4, there appears to be a change in the rate at which oxygen isotopic values increased with time across the Archean-Proterozoic boundary, broadly consistent with—but in detail different from—the findings of Valley et al. (2005).
Significant numbers of Archean zircons in CKG38 have δ18O values greater than 7.5‰, which has previously been proposed as the upper limit for Archean zircon grains (Cavosie et al., 2005; Valley et al., 2002, 2005). Fifteen Archean (>2500 Ma) grains analyzed have concordant U-Pb systematics with igneous-like Th-U ratios and low common Pb contents. These grains have either sharp oscillatory zoning or domains of oscillatory zoning overgrown by rims with homogeneous, low-cathodoluminescence response. Oscillatory-growth zoning is interpreted as a primary magmatic feature. The oldest identified, oscillatory-zoned, magmatic grain has a concordia age of 3953 ± 18 Ma (2σ), and is one of a few reported detrital Eoarchean magmatic zircons from Greenland (Fig. 3). The oxygen isotope analytical spot was sited away from any visible fractures and has a calculated accumulated alpha-radiation dosage of 0.254 × 1016 events/mg, based on age, U and Th concentrations, a value that correlates with a “well-crystallized” structure (Murakami et al., 1991). Its oxygen isotopic composition of 7.76‰ ± 0.50‰ (2σ) is greater than that observed in crystals in high-temperature equilibrium with mantle rocks (i.e., 5.3‰ ± 0.6‰; Valley, 2003). All other Archean grains analyzed herein also show elevated oxygen signatures in excess of mantle-like values. Excluding the possibility of unrecognized low-temperature disturbance, which we have attempted to identify within these concordant analyses, the elevated values thus indicate the presence of heavy oxygen by at least the Eoarchean, consistent with evidence for recycling of sedimentary rocks in early Earth history (Chauvel et al., 2008; Shirey et al., 2008).
Similar results from other Eoarchean to Hadean zircon crystals in the Jack Hills of Western Australia have led to marked differences in interpretation (e.g., Coogan and Hinton, 2006; Harrison et al., 2005; Nemchin et al., 2006a). Oxygen isotope and trace-element zonation from Jack Hills zircon has been suggested as evidence for Hadean igneous processes involving supracrustal material (Wilde et al., 2001). Unradiogenic Hf isotope compositions in other ancient Jack Hills zircons imply the existence of enriched (crustal) reservoirs by at least 4300 Ma (Amelin et al., 1999; Harrison et al., 2005). However, both the size of this crust and the mechanism whereby it was generated are debated (e.g., Shirey et al., 2008). Thermobarometry of mineral inclusions within Jack Hills zircons indicates that they formed in a region with low heat flow, consistent with, but not exclusive to, a subduction-zone setting (Hopkins et al., 2008; Rollinson, 2008; Nutman et al., 2008). Detailed discussion of the significance of Hadean Jack Hills zircon has been presented elsewhere (e.g., Cavosie et al., 2005; Nemchin et al., 2006a) and is not reiterated here.
While the oxygen isotopic data from Greenland cannot, on its own, resolve the question as to when subduction initiated on early Earth, it can help to elucidate the conditions for melting in the Eoarchean Greenland Shield. Our oxygen isotope results suggest that the majority of detrital zircons in the Mørænesø Formation were derived from melts that incorporated material modified by near-surface processes. The host melts to these detrital zircon crystals appear to have been generated during crustal reworking events accompanying supercontinent assembly (Condie, 1998, 2004). Additionally, the heavy oxygen signature in a single Eoarchean zircon is consistent with the presence of liquid water at ca. 4 Ga, permitting wet melting conditions. This documents that the heavy oxygen isotopic signal from the Jack Hills zircons is not unique to the Yilgarn of Western Australia, but it is also recorded in Eoarchean grains from other cratons.
Several high δ18O values were obtained from Paleoproterozoic grains. However, these grains have a homogeneous low-cathodoluminescence response and show no textural evidence of retaining a primary magmatic signature, such as oscillatory-growth zoning (e.g., Nemchin et al., 2006a). The oxygen isotope compositions for these grains therefore could reflect a variety of nonmagmatic alteration processes, including interaction with low-temperature surface or subsurface water, which could have resulted in elevated δ18O values for these grains. Although such alteration processes cannot be clearly identified by strong discordance, increase in common Pb, or change in Th-U ratio (e.g., Peck et al., 2003), the textural features of these grains caution that they may have experienced some minor degree of postcrystallization alteration that did not demonstrably affect their U-Pb systematics (e.g., Nemchin et al., 2006a) (Fig. 4). Therefore, we have opted for a conservative interpretation and consider these results to represent potentially modified δ18O values.
The high δ18O values (12‰) for some ca. 1000 Ma zircons obtained in this study are consistent with those of Valley et al. (2005). These grains have similar oxygen isotopic compositions to crystals in anorthosite–mangerite–charnockite–granite plutons in the Fronetenac arc of the Grenville Province, which are known to have elevated δ18O values of up to 14‰ (King et al., 1998). These elevated oxygen isotopic values support the conclusion of Kirkland et al. (2009) that the detrital population of the Morænesø Formation contains a contribution from a distal Grenville source.
Secular Change in δ18O
There are numerous models for both the growth of continental crust and the rate of crustal recycling through the mantle (Fyfe, 1978; Hawkesworth and Kemp, 2006; Hurley and Rand, 1969; Moorbath, 1985; Nägler and Kramers, 1998; Taylor and McLennan, 1981). The relatively uniform upper values of δ18O in magmas across the whole of the Archean have been used to suggest that rates of crustal recycling balanced rates of crustal growth on early Earth (Valley et al., 2005), as also suggested for modern Earth (Scholl and von Huene, 2009). Our Greenland data, however, suggest an increase in the heaviest δ18O values (albeit strongly constrained by the single Eoarchean analysis). This apparent increase in δ18O may signify the increasing incorporation of high δ18O materials by magmas over time. One mechanism to achieve this could be through increased weathering, transport, and subduction recycling of sedimentary material. However, there is no unique requirement for sediment subduction, since reworking in the crust leading to melting is also feasible. In either case, the change in maximum δ18O value through time is consistent with a secular increase in the volume of continental crust, as suggested by a wide range of isotope and geochemical evidence (e.g., Allègre, 1982; Allègre and Rousseau, 1984; Collerson and Kamber, 1999; McLennan and Taylor, 1983; Veizer and Compston, 1976), and the decline in the rate of mantle magmatism (e.g., Dickinson and Kröner, 1981).
Changes across the Archean-Proterozoic Boundary
The Greenland data set presented shows an apparent continuous rise in δ18O values throughout the Proterozoic, distinct from the Archean, and changing around the Archean-Proterozoic boundary (Fig. 3; see also Valley et al., 2005). Assuming an average sediment value of ∼+15‰ and a highest Archean magmatic value of +9‰, this would imply a maximum input of ∼30% high δ18O material in the Archean magmas of the Greenland Shield relative to mantle δ18O values. For Proterozoic zircon grains, up to ∼50% of high δ18O material could have been incorporated into the magmas.
These changes most likely reflect differences in the composition and/or abundance of sediment and/or altered crust available for recycling (Albarède, 1998). Other factors may account for the change in slope of the δ18O values across the Archean-Proterozoic boundary, such as the development of cyanobacteria and the rise in atmospheric oxygen. In the Archean, an atmosphere exhibiting higher pCO2 and possibly pCH4 would result in higher degrees of chemical weathering (Kump, 2008), and dissolved weathered rocks would be transported into the oceans as solutes mixing with the bulk ocean. The resulting subduction-related magmas (produced from material derived or influenced by the oceans) would thus reflect isotopic dilution by the water mass and the buffering effects of ocean crust alteration at mid-ocean ridges (e.g., Jean-Baptiste et al., 1997). After the development of an oxygenic atmosphere, however, the style of rock weathering changed to a regime of principally physico-mechanical breakdown of materials (e.g., Condie, 1993; Lowe and Tice, 2004). Hence, oxygen isotope signatures in zircons produced from material affected by near-surface processes, where fractionation is large, remained locked within the mineral grains and were preserved for later reworking.
Another possible factor accounting for the change in slope of the δ18O values across the Archean-Proterozoic boundary is the development of larger, cooler, and stiffer plates by the end of the Archean. This would have given rise to longer, higher orogens (Campbell and Allen, 2008; Squire et al., 2006). In this model, rates of sediment recycling in the Archean were limited by a lack of uplift and erosion of continents. This limit reflects greater radiogenic heat production, as judged from K, Th, and U, in the Archean (e.g., O'Nions et al., 1978; Davies, 1980; Wanke et al., 1984; Franck, 1998), which resulted in softer, weaker continental lithosphere (Cagnard et al., 2006; Choukroune et al., 1995; Rey and Houseman, 2006). This inherent weakness meant that continental crust did not significantly thicken upward during subduction-accretion and continental collision events (e.g., the widespread development of low-grade granite-greenstone terrains), and hence there was limited uplift and erosion. In the Proterozoic, secular decrease in radioactivity and overall mantle heat, combined with melt depletion of lower crust (Rudnick and Fountain, 1995), would have resulted in larger plates and progressive strengthening of continental lithosphere, in turn allowing crustal thickening, uplift, orographic precipitation, and erosion at greater rates and volumes than in the Archean. This would have increased sediment recycling, thereby increasing the δ18O values of Proterozoic magmas.
The δ18O values of detrital zircon crystals in the Mørænesø Formation indicate a maximum input of ∼30% high δ18O material into Archean magmas of the Greenland Shield relative to the mantle. For Proterozoic zircons, up to ∼50% of high δ18O material could have been incorporated into these magmas. Therefore, recycling of high δ18O sources, such as sediment or altered basement, was an important process during crustal growth events within the Greenland Shield from ca. 4000 Ma, and became an even more significant process during the Proterozoic.
Detrital zircons of the Mørænesø Formation show a progressive increase in δ18O from the end of the Archean to the present day, similar to that seen in the global compilation of Valley et al. (2005). In contrast to this earlier study, however, our data set suggests a progressive increase in δ18O values also during the Archean. These secular changes in δ18O imply increasing incorporation of high δ18O material into silicic magmas over time via increased weathering, transport, and reworking of high δ18O sources. This observation supports models for increasing continental crustal volume during the Archean and the progressive stiffening and exposure of continental crust.
The Mørænesø Formation data set further documents a minor change in the rate of increase of heavy oxygen signatures across the Archean-Proterozoic boundary. This rate change may reflect (1) changing sediment composition from material derived from chemically weathered, buffered solutions in the Archean to mechanically broken-down material in the Proterozoic; and/or (2) a greater abundance of sediment available for recycling due to erosion of longer and higher Proterozoic orogens.
One Eoarchean (3953 ± 18 Ma) grain yields a heavy oxygen isotope signature of 7.8 ± 0.5‰, which is interpreted as a primary magmatic value. This is consistent with the existence of liquid water during Eoarchean times and implies wet melting conditions.
Kirkland wishes to thank J. Ineson, P. Frykman, J.S. Peel, and L. Stemmerik for insightful discussion in the field. Financial support from the Swedish Research Council (to Pease) and Ymer-80 (to Kirkland) is gratefully acknowledged. We thank reviewers Tony Kemp, Anna Pietranik, and Jon Patchett for constructive comments that improved this manuscript. Kirkland and Van Kranendonk publish with the permission of the Executive Director of the Geological Survey of Western Australia. The NORDSIM facility is funded by the research councils of Denmark, Norway, Sweden, the Geological Survey of Finland, and the Swedish Museum of Natural History. This is NORDSIM publication 249.