Secular variations of N-isotopes in terrestrial reservoirs and ore deposits
Published:January 01, 2006
- PDF LinkChapter PDF
R. Kerrich, Y. Jia, C. Manikyamba, S.M. Naqvi, 2006. "Secular variations of N-isotopes in terrestrial reservoirs and ore deposits", Evolution of Early Earth's Atmosphere, Hydrosphere, and Biosphere - Constraints from Ore Deposits, Stephen E. Kesler, Hiroshi Ohmoto
Download citation file:
New δ15N analyses combined with a literature compilation reveal that shale kerogen, VMS-micas, and late-metamorphic vein micas show a secular trend from enriched values in the Archean, through intermediate values in Proterozoic terranes, to the Phanerozoic mode of 3‰–4‰. Kerogen in metashales from the 2.7 Ga Sandur Greenstone Belt, eastern Dharwar Craton, India, is characterized by δ15N 13.1‰ ± 1.3‰, and C/N 303 ± 93. A second population has δ15N 3.5‰ ± 0.9‰, and C/N 8 ± 0.4, close to the Redfield ratio of modern microorganisms, and is interpreted as precipitates of Proterozoic or Phanerozoic oilfield brines that penetrated the Archean basement. Kerogen from 1.7 Ga carbonaceous shales of the Cuddapah Basin average 5.0‰ ± 1.2‰, close to the mode at 3‰–4‰ for kerogen and bulk rock of Phanerozoic sediments. Biotites from late-metamorphic quartz-vein systems of the 2.6 Ga Kolar gold province, E. Dharwar Craton, that proxy for average crust, are also enriched at 14‰–21‰ for three samples, confirming that the N–budget of the hydrothermal fluids is dominated by sedimentary rocks. Muscovites from altered volcanic rocks in 2.7 Ga Abitibi belt VMS deposits have δ15N 12‰–20‰, in keeping with published data for shale kerogen from the same terrane, whereas equivalents in the 1.8 Ga Jerome VMS span 11.7‰–14.1‰.
15N-enriched values in Precambrian rocks cannot be caused by N-isotopic shifts due to metamorphism or Rayleigh fractionation because (1) pre-, and post-metamorphic samples from the same terrane are both enriched in 15N; (2) there is no covariation of δ15N with N, C/N ratios, or metamorphic grade; and (3) the magnitude of fractionations of 1‰ (greenschist) to 3‰ (amphibolite facies) during progressive metamorphism of sedimentary rocks, as constrained from empirical observations and experimental studies, is very small. Nor can 15N-enriched values stem from long-term preferential diffusional loss of 14N, as samples were selected from terranes where 40Ar/39Ar ages are within a few million years of concordant U-Pb ages; nitrogen is structurally bound in micas, whereas Ar is not.
It is possible that the 15N-enriched values stem from a different N-cycle in the Archean, with large biologically mediated fractionations, yet the magnitude of the fractionations between atmospheric N2 and organic nitrogen observed exceeds any presently known, and chemoautotrophic communities tend to depleted values. Earlier results on Archean cherts show a range of δ15N from −6‰ to 30‰. Given the temporal association of chert–banded iron formation (BIF) with mantle plumes, the range is consistent with mixing between mantle N2 of −5‰ and the 15N-enriched marine reservoir identified in this study. The 15N-enriched Archean atmosphere-hydrosphere reservoir does not robustly constrain Archean redox-state. We attribute the 15N-enriched reservoir to a secondary atmosphere derived from CI-chondrite-like material and comets with δ15N of +30‰ to +42‰. Shifts of δ15N to its present atmospheric value of 0‰ can be accounted for by a combination of early growth of the continents with sequestration of atmospheric N2 into crustal rocks, and degassing of mantle N ∼−5‰. If Earth's surface environment became oxygenated ca. 2 Ga, then there were no associated large N-isotope excursions.
INTRODUCTION AND SCOPE
The present nitrogen cycle is quite well documented, including N-isotope fractionations accompanying organic and inorganic transfers of N between terrestrial reservoirs (Delwiche and Steyn, 1970; Macko et al., 1987; Rau et al., 1987; Williams et al., 1995; Kao and Liu, 2000). There are abundant data on modern organic compounds in the biosphere and sediments (Peters et al., 1978; Sweeney et al., 1978; Mazuka et al., 1991; Sadofsky and Bebout, 2004, and references therein), and Phanerozoic rocks (Haendel et al., 1986; Bebout and Fogel, 1992; Williams et al., 1995; Mingram and Bräuer, 2001, and references therein). Holloway and Dahlgren (2002) give a recent review.
The isotopic compositions of N in the mantle (−5‰), organic compounds in sediments (0‰ to −6‰ with a mean of −4‰), atmosphere (0‰), Phanerozoic black shales (kerogen, 3‰–4‰), and granites (range 0 to10‰) are well constrained (Fig. 1). Isotopic differences in terrestrial reservoirs stem from (1) nitrification, denitrification, N-limitation, and other metabolic or inorganic reactions of the near-surface N-cycle (Delwiche and Steyn, 1970; Wada et al., 1975; Sweeney et al., 1978; Saino and Hattori, 1980; Macko et al., 1987; Rau et al., 1987; Hoch et al., 1994; Pinti and Hashizume, 2001; Lehmann et al., 2002); (2) equilibrium fractionation in the geosphere, for example fluid-rock interaction (Hanschmann, 1981); and (3) kinetic effects associated with mantle degassing (Marty and Zimmermann, 1999; Cartigny and Ader, 2003, and references therein), or Rayleigh fractionation during metamorphic devolatilization (Bebout and Fogel, 1992; Mingram and Bräuer, 2001) (Fig. 1).
However, little is known about the early evolution of the N-cycle because records in Archean and Proterozoic rocks are sparse and the oxidation state of the Archean atmosphere-hydrosphere system is uncertain (Ohmoto, 1997, 2004; Holland, 1999; Phillips et al., 2001). Only a few data for N contents and N-isotopes have been published on Archean shales (Zhang, 1988; Jia and Kerrich, 2000, 2004a, 2004b), cherts and stromatolite-bearing sediments (Gibson et al., 1985, 1986), and chert–iron formation (Hayes et al., 1983; Sano and Pillinger, 1990; Beaumont and Robert, 1999; Pinti et al., 2001a). N-isotope data on hydrothermal ore deposits, Precambrian or Phanerozoic, are limited (Jia and Kerrich, 1999, 2000; Jia et al., 2001, 2003a).
Nitrogen-isotopes have been used to address a variety of questions: (1) the isotopic composition of the Archean atmosphere (Sano and Pillinger, 1990; Beaumont and Robert, 1999); (2) secular variation of redox state of the atmosphere and oceans (Beaumont and Robert, 1999); (3) bacterial metabolic pathways (Pinti and Hashizume, 2001; Pinti et al., 2001a); (4) origin of Earth's atmosphere-hydrosphere (Javoy, 1998; Tolstikhin and Marty, 1998; Jia and Kerrich, 2004a, 2004b); (5) the N-isotope characteristics of Archean sedimentary rocks (Hayes et al., 1983; Zhang, 1988; Jia and Kerrich, 2000, 2004a, 2004b); (6) sources of hydrothermal fluids involved in orogenic, or shear zone-hosted mesothermal, gold deposits (Jia and Kerrich, 1999, 2000; Jia et al., 2001, 2003a); (7) nitrogen budgets in convergent margins (Bebout and Fogel, 1992; Sadofsky and Bebout, 2004); and (8) recycling of sedimentary rocks into the mantle (Marty and Dauphas, 2003).
Several observations emerge from the limited database: (1) kerogen in Mesoarchean (3.4–2.9 Ga) cherts are 15N-depleted relative to Neoarchean (2.9–2.5 Ga) counterparts; (2) kerogen in Archean black shales is 15N-enriched compared to Mesoarchean cherts and Phanerozoic equivalents; (3) the δ15N of hydrothermal K-micas in orogenic gold deposits of Phanerozoic accretionary terranes is comparable to that of contemporaneous shales; and (4) K-micas in Archean gold deposits are as enriched as contemporaneous shale kerogen in 15N. Consequently, Archean cherts and shales may sample different N-reservoirs, and shales record a secular variation of δ15N (Fig. 1).
In this paper, new N-isotope data are reported for 2.7 Ga and 1.8 Ga carbonaceous shales from India to further test for secular variations. The first data for 2.7 and 1.8 Ga volcanic hosted massive base metal (Cu-Zn-Pb) sulfide (VMS) deposits are also presented, together with new data for hydrothermal K-micas from the ca. 2.6 Ga Kolar gold province, India. We compile existing data together with the new results to evaluate the δ15N of various lithologies through time. From this database we consider the origin of N in orogenic gold, rare metal pegmatite, and VMS deposits, examine the implications of secular variations of δ15N, and address the question of whether N-isotopes record “the great oxygenation event” ca. 2.3 Ga.
Carbonaceous metashales were sampled from the 2.7 Ga Sandur Greenstone Belt (SGB), and the Paleoproterozoic Cuddapah Basin, India (Fig. 2). The Sandur belt is one of a series of composite tectonostratigraphic supracrustal terranes, separated by granitoids, in the eastern Dharwar Craton (Manikyamba et al., 1997). The eastern SGB is dominated by arc-related tholeiitic to calc-alkaline flows. Minor sedimentary units include polymictic conglomerate, graywacke, and carbonaceous shales. Rhyolites have zircon U-Pb ages of 2658 ± 14 Ma using the SHRIMP technique (Nutman et al., 1996). The western SGB is prevalently 2.7 Ga high-Mg basalts and komatiites. Shales were sampled from Vibhutigudda and Bhimangundi (Fig. 2). Fossil cyanobacteria have been described from carbonaceous cherts at the former locality (Naqvi et al., 1987; Venkatachala et al., 1990). The eastern arc and western plateau terranes accreted post 2.6 Ga (Manikyamba et al., 1997; Naqvi et al., 2002). Metamorphic grade varies from greenschist facies in the west to mid-amphibolite facies in the east (Manikyamba et al., 1997). Deformation is of low intensity except proximal to faults or shear zones.
The intracratonic Cuddapah Basin developed over 1.9–1.7 Ga (Fig. 2). Siliciclastic rocks are prevalent in this Proterozoic sequence, including conglomerates, current-bedded and rippled arenites, and diverse shales (Nagaraja-Rao et al., 1987). Well-preserved stromatolitic units and microfossils in cherts associated with stromatolites are present (Schopf and Prasad, 1978; Nagaraja-Rao et al., 1987). Samples of carbonaceous shales were obtained from near Mangampeta and Marcapur. Metashales are at prehnite-pumpellyite facies in the former locality, and at greenschist facies in the latter. The basin's western margin lies unconformably on Archean craton, whereas the eastern margin tectonically underlies the Eastern Ghat Mobile Belt (EGMB), which was thrust over the basin. From dating of anorthosite, alkali plutons, and granitoids emplaced along the accretionary zone, accretion of the EGMB, which was part of Antarctica before the breakup of Gondwanaland, to the Dharwar Craton is estimated to have occurred ca. 1600 Ma (Dasgupta and Sengupta, 2003). Carbonaceous shales were sampled from fresh rock in road or rail cuttings for both the Sandur belt and Cuddapah Group.
Kolar Gold Province
The Kolar terrane is one of several supracrustal greenstone sequences in the eastern Dharwar Craton. Tholeiitic basalts and komatiites dated at 2.7 Ga are prevalent, at amphibolite facies. Gold mineralization was coeval with brittle-ductile deformation (Hamilton and Hodgson, 1986) associated with accretionary tectonics (Balakrishnan et al., 1999). Biotite-rich alteration selvedges bounding gold-bearing quartz-calcite veins were obtained from the Oriental reef (see Siddaiah and Rajamani, 1989, and references therein). Nitrogen, as NH4 +, may substitute for K given similar valence and ionic radius (Honma and Itihara, 1981; Bos et al., 1988). Accordingly, K-silicates are preferred minerals for N-isotope studies of hydrothermal ore systems.
VMS deposits reflect the conjunction of magmatic, hydrothermal, and sedimentary processes proximal to the seafloor. The Abitibi greenstone, composite, supracrustal terrane, dated to ca. 2.7 Ga, is the largest and best-preserved greenstone belt, with several VMS districts (for a review see Jackson and Fyon, 1991). Metamorphic grade is prehnite-pumpellyite to greenschist facies, and deformation intensity is low except proximal to regional shear zones. The VMS deposits formed on the seafloor during a hiatus of bimodal volcanism in several areas, notably Kidd Creek, Ontario, and the Noranda and Matagami districts of Quebec. Alteration of footwall volcanic rocks, by modified seawater-derived hydrothermal fluids, generated domains of muscovite and chlorite bearing alteration (Franklin et al., 1981). Muscovite-rich samples were obtained from mafic volcanic flows in the footwall of the 2.7 Ga Kidd Creek VMS deposit, Ontario, and from felsic pyroclastic units subjacent to the Amulet deposit, Noranda, and Mattagami Lake deposit, Matagami, Quebec. Geological relationships of Abitibi VMS deposits have been reviewed by Franklin et al. (1981), Bleeker et al. (1999), and Hannington et al. (1999). Seawater-rock ratios in the footwall are known to be large compared to possible mantle or magmatic contributions (Beaty and Taylor, 1982; Costa et al., 1983).
The Jerome VMS deposit is located in the upper of two cycles of bimodal magmatism of the 1.8 Ga Ash Creek Group, Arizona. Seafloor massive sulfides and mineralized breccias are present, and “black smoker” chimneys are preserved. Muscovite-rich, hydrothermally altered felsic volcanic rocks were sampled from the footwall. Metamorphic grade is greenschist facies (Anderson et al., 1971; Sangster and Scott, 1976; Lindberg and Gustin, 1987).
Studies of progressively metamorphosed sedimentary rocks show shifts of ∼1‰ in δ15N from protoliths to greenschist facies counterparts, and ≤3‰ to amphibolite facies (Haendel et al., 1986; Bebout and Fogel, 1992; Mingram and Bräuer, 2001; Busigny et al., 2003; Jia, 2004). Alternatively, some authors attributed variations of 27‰–36‰ in their data sets to shifts due to Rayleigh fractionation accompanying metamorphism (Beaumont and Robert, 1999; Pinti et al., 2001a). The former studies are all of siliciclastic sedimentary rocks whereas the latter are of kerogen from chert–banded iron formation (BIF). Consequently, differences between the studies could be sample or environment dependent, rather than due to metamorphism, or alternatively they could reflect assumptions in Rayleigh modeling.
Given the time-series and spatial association of chert–iron formation with volcanic sequences erupted from mantle plumes (Isley and Abbott, 1999; Condie et al., 2001), mantle N may be incorporated into kerogen in chert–iron formation, but probably not in distal carbonaceous shales. Accordingly, we selected carbonaceous shales distal from chert-BIF to obtain a marine biogenic signature. The magnitude of metamorphic shifts of δ15N values was estimated by analysis of pre-metamorphic and late-metamorphic materials from the same terrane. Analyses of fine-grained hydrothermal K-micas from VMS deposits allow comparison with published data for kerogen in carbonaceous shales from the same 2.7 Ga terrane. In turn, data from these two types of pre-metamorphic samples are compared with late-metamorphic hydrothermal micas from the Kolar gold deposit, and with published data for micas from 2.7 Ga late-metamorphic gold deposits. Alteration associated with gold deposits overprints regional metamorphic fabrics and the deposits retain primary fluid inclusions, equilibrium quartz-muscovite oxygen isotope fractionations, and upper plateau 40Ar/39Ar ages. Accordingly the deposits have not been overprinted by a later metamorphic event (Kerrich and Cassidy, 1994; McCuaig and Kerrich, 1998).
Separation and Analysis
Previous studies of Precambrian carbonaceous sedimentary rocks revealed low bulk rock N contents. Accordingly, kerogen was separated for analysis of N and δ15N, revealing that kerogen N dominated the whole rock N-budget. Minor N is likely to be in K-silicates (Hayes et al., 1983; Beaumont and Robert, 1999; Jia and Kerrich, 2004b). Kerogen was separated using the following procedure. All selected sedimentary rock samples were washed in dichloromethane-ethanol to remove modern organic contamination, ground in a steel mortar to <50 µm, and then treated with HCl and HF to remove carbonates and silicates using the technique of Durand and Nicaise (1980). All kerogens in this study have been screened by X-ray diffraction (XRD); graphite peaks were at or below detection, in keeping with greenschist facies shale and the results of Landis (1971). Pure muscovite separates, where N as NH4 + substitutes for K, from gold and VMS deposits were obtained by standard mineral separation procedures.
The analytical techniques used in this study involved a high-precision continuous flow-isotope ratio mass spectrometer (CF-IRMS) at the Soil Science Laboratory, University of Saskatchewan, and followed the techniques used by Jia and Kerrich (2000). Analytical precision (reproducibility, 1σ; n ≥ 3) is typically ∼0.3‰ for kerogen δ15N, ∼0.5‰ for muscovite δ15N, and ∼0.3‰ for kerogen δ13C. The long-term reproducibilities for international nitrogen isotope standard materials are as follows: IAEA-N1 = 0.54 ± 0.07‰ (n = 15, accepted value 0.53‰); IAEA-N2 = 20.35 ± 0.08‰ (n = 10, accepted value 20.41‰); and for the internal laboratory standard material BLN.SOIL, 5.20 ± 0.21‰ (n = 20, accepted value 5.15‰). Ten replicate analyses of “in house” muscovite separates of samples KAII-1 and CD11-21-1 yielded mean δ15N values of 19.4‰ ± 0.09‰ and 3.4‰ ± 0.06‰, respectively (Jia et al., 2003b). Nitrogen and carbon concentrations were obtained from each sample based on system calibration using known standards. Blanks for this technique are <0.075 µg N2 for routine runs of the types of samples analyzed in this study. Isotope data are reported in standard δ-notation relative to atmospheric N2 for nitrogen and to the Peedee Belemnite limestone (PDB) standard for carbon.
Isobaric Interferences and Modern Organic Contamination
Studies show that isobaric interference of carbon monoxide (CO) may produce erroneous nitrogen isotope ratios. One percent CO in an analyte would cause ∼7‰ errors in δ15N (Beaumont et al., 1994). The analytical approaches used in this study have carefully eliminated any such isobaric interferences.
Contamination by organic nitrogen compounds during the preparation of the kerogen residue could change the initial isotopic composition. However, samples were treated with dichloromethane-ethanol, as in other studies of Precambrian rocks (Beaumont and Robert, 1999). Consequently, the very positive δ15N values of Archean samples cannot be attributed to contamination because modern organic nitrogen compounds have δ15N of close to 0‰, or negative down to −6‰ (Nadelhoffer and Fry, 1988; Sachs and Repeta, 1999; Kao and Liu, 2000), and modern kerogen ranges from 1‰ to 6‰ (Williams et al., 1995; Ader et al., 1998; Kao and Liu, 2000; Sephton et al., 2002). Also, several samples have the extremely 13C-depleted values of −33‰ to −48‰, characteristic of some Archean kerogen (Wellmer et al., 1999); accordingly, significant contamination by modern hydrocarbons can be ruled out.
Total Inorganic Carbon (TIC)/Total Organic Carbon (TOC)
Total inorganic carbon, and total organic carbon were determined on carbonaceous shales using a Leco CR-12 carbon analyzer, following the procedure of Wang and Anderson (1994).
Data are reported in Table 1 and Figure 3. Also shown in Tables 2 and 3, for comparison, are compilations of recent data on nitrogen contents and δ15N values of Archean to Phanerozoic siliciclastic sediments, chert-BIF, granitoids, and ore deposits.
Carbonaceous shales from the Archean Dharwar Craton have two compositional-isotopic populations. The six kerogens from Vibutigudda (P1) are characterized by relatively uniform N (73 ± 11 ppm) and C (18,345 ± 3710 ppm) contents, with C/N ratios of 179–435. The δ15N values (12‰ to 15‰) are much higher than those of most Phanerozoic sedimentary rocks (1‰–6‰; Table 3). Four kerogens from Bhimangundi define a second population (P2); they feature greater N (841 ± 24 ppm) but lower C (5500 ± 365 ppm) contents, giving uniform C/N ratios of 7–8. Both the δ15N values (3.5‰ ± 0.9‰) and δ13C (−32.0‰ ± 0.1‰) are lower than counterparts from Vibutigudda (Table 1). Proterozoic carbonaceous shales constitute a third population (P3). Their average N (395 ppm) and C (21,309 ppm) contents are greater than in P1, but average C/N ratios (73) and δ15N are lower (Table 1, Fig. 3). Shale kerogens at the Mangampeta locality are enriched by ∼2‰ relative to greenschist facies equivalents at Macapur. Given shifts of ∼1‰ from sedimentary protoliths to greenschist facies in several studies (see Discussion section), the 2‰ difference may reflect primary variations of δ15N.
P2 “Archean” samples are characterized by C/N values close to the Redfield ratio, averaging 6.6 (C/N = 106:16) of modern organic compounds (Chen et al., 1996; Fraga et al., 1998). In contrast, P1 and most Archean kerogens are characterized by C/N spanning 35–600 (Hayes et al., 1983; Sano and Pillinger, 1990; Beaumont and Robert, 1999; Jia and Kerrich, 2004b). P2 samples also feature the conjunction of less-dispersed compositional and isotopic values than the other two Precambrian populations (P1 and P3), with total organic carbon (TOC) to total inorganic (TIC) ratios greater than P1 (Table 1). The data plot to the lower end member of the colinear array, in δ15N vs. δ13C coordinates, of organic compounds in modern marine sediments of Peters et al. (1978). Considering these lines of evidence together, we interpret P2 data in terms of infiltration of Proterozoic or Phanerozoic hydrocarbon-bearing formation brines locally into Archean basement, forming a secondary hydrocarbon signature. Penetration of Proterozoic and Phanerozoic formation brines, and Quaternary groundwaters, into Archean crust has been documented for several cratons (for a review see Kerrich and Ludden, 2000).
Hydrothermal K-micas from Volcanogenic Massive Sulfide (VMS) and Gold Deposits
The δ15N values of hydrothermal biotite from 2.6 Ga quartz vein systems at Kolar are 14‰–21‰ and overlap the range for P1 kerogen in the eastern Dharwar Craton; both sets of samples are enriched in 15N compared to Proterozoic or Phanerozoic N-reservoirs (Tables 1,3; Figs. 1, 3). K-micas generated by modified seawater-derived hydrothermal fluids in 2.7 Ga Abitibi VMS deposits are enriched in 15N at Kidd Creek (16‰ to19‰), Amulet (12‰–16‰), and Matagami (13‰–20‰), overlapping the range of δ15N values of kerogen in metashales from this terrane. All micas from the Archean and Proterozoic VMS deposits have low N contents of 10–25 ppm (Tables 1,3; Fig. 3). In the 1.8 Ga Jerome VMS deposit, four δ15N values of muscovite are 12‰–14‰, intermediate between Phanerozoic kerogen or bulk sedimentary rocks, and Archean VMS counterparts or kerogen (Tables 2,3).
COMPARISON WITH THE LITERATURE
Mantle and Diamonds
From worldwide sampling of diamonds, Boyd et al. (1987, 1992) reported negative δ15N values (−8.7‰ to −1.7‰) with a mode of −6‰ to −5‰. Independently, Cartigny et al. (1997, 1998) showed that mantle δ15N values of diamonds were between −8‰ and −5‰ (see also Marty and Humbert, 1997; Marty and Zimmermann, 1999; Javoy and Pineau, 1991). The diamonds are mostly Archean, as old as 3.2 Ga (Richardson et al., 1984, 2001). Accordingly, upper mantle δ15N may have been uniform at about −6 ± 1‰ since ∼3.2 Ga (Fig. 1). However, lower δ15N values of down to −20‰ were found in some rare diamonds, signifying another N reservoir (Cartigny et al., 1997).
Marty (1995) estimated the N content of undegassed mid-oceanic ridge basalt (MORB) to be 1–2 ppm; these values are inferred from (1) observed covariations of N2/36Ar, 40Ar/39Ar, and 3He/4He in MORB; (2) an assumption that nitrogen, like He and CO2, behaves as an incompatible element during partial melting of rocks; and (3) the mantle carbon content, which is ∼400 ppm. The upper mantle would then have an N content of ∼0.16 ppm (Porcelli and Turekian, 2004). According to Cartigny and Ader (2003), uncertainties in these estimates for MORB may arise from isotopic fractionation during partial degassing of basaltic rocks. Alternatively, the mantle N content is estimated at ∼40 ppm from δ13C-N systematics for diamond, assuming that N is not highly incompatible (Cartigny et al., 2001).
Phanerozoic and Modern Sediments
Sedimentary rocks have similar bulk nitrogen isotopic compositions as indigenous kerogen, consistent with the bulk rock N budget being dominated by kerogen N. The total range of δ15N in Phanerozoic bulk sedimentary rocks is 1.0‰–10.8‰ (Fig. 1, Table 2). However, the averages of most rock data sets cluster between 3‰ and 5‰, within one standard deviation, the global average being 3.65‰ ± 0.55‰ (Table 2). Similarly, the total range of kerogen is −2.0‰ to 6.0‰, but averages of each data set cluster between 1.5‰ and 4.6‰, and the global average is 3.7‰ (Table 2).
Recent marine sediments have a δ15N range of −3‰–9‰, with an average of 3.9‰ ± 1.1‰ (Table 2; Peters et al., 1978; Sweeney et al., 1978; Rau et al., 1987; Williams et al., 1995; Schubert and Calvert, 2001; Sadofsky and Bebout, 2004). Variations in δ15N have been attributed to varying mixtures of terrestrially derived (13C and 15N depleted) and enriched marine-derived organic compounds, given a linear correlation between δ15N and δ13C (Peters et al., 1978; Sweeney et al., 1978; Minoura et al., 1997). Alternatively, according to Sadofsky and Bebout (2004) variations may stem from some combination of changes in bioproductivity, complex diagenetic processes, and differing proportions of marine and terrestrial organic matter in continental margin versus intraoceanic arc settings.
A larger range of N-isotope compositions has been recorded from specific niches. δ15N values of −12‰ to 4‰ are documented from chemoautotrophic bacteria in warm seeps on the seafloor (Conway et al., 1994). Reduced nitrate availability may generate sedimentary organic compounds (SOC) with δ15N as low as −2.7‰. In contrast, intense denitrification, characterized by a kinetic isotope effect, may generate δ15N values as high as 19‰ (Cline and Kaplan, 1975; Sweeney et al., 1978; Rau et al., 1987). For example, Mazuka et al. (1991) report δ15N values of 0.9‰–18.9‰, with a mode of 8‰, for Pliocene and younger marine sediments at ODP site 724.
The principal process that controls changes in isotope composition of organic nitrogen is thermal decomposition of SOC during diagenesis. Preferential release of 14N, given lower energy to break 14N-12C than 15N-12C bonds, shifts the residual kerogen to ∼3‰ (Delwiche and Steyn, 1970; Wada et al., 1975; Macko et al., 1987), implying a flux of 15N depleted nitrogen back to the surface. Wada et al. (1975) and Sweeney et al. (1978) also considered that denitrification may be responsible for nitrogen isotope fractionation between atmospheric and organic nitrogen in kero-gens, because denitrification, a bacterial process, involves preferential loss of 14N to the atmosphere, leaving nitrate enriched in 15N; nitrate is the source of marine organic matter, which retains its isotopic signature in marine sedimentary rocks. Williams et al. (1995) report two stages of N production during progressive burial: from microbiological activity during diagenesis, and then release of N in the “oil window” when NH4 + becomes incorporated into illite during the smectite-illite transition. The remaining organic matter as kerogen in sedimentary rocks increases in δ15N relative to precursors. Freudenthal et al. (2001) report a decrease from 1000 to 250 ppm N during diagenesis of recent sediments accompanied by a shift of 1.9‰ by Rayleigh fractionation.
Fractionations between kerogen N (3.2‰ ± 0.3‰) and fixed N in mudstones (3.0‰ ± 1.4‰) are small (Williams et al., 1995). Given that Phanerozoic upper crust is dominated by recycled sedimentary rocks (Taylor and McLennan, 1985), and the kerogen-sediment fractionation is insignificant, bulk upper crust has a nitrogen isotope composition of 3‰–4‰, indistinguishable from kerogen (see Fig. 1 for references).
In summary, notwithstanding extreme δ15N values in special niches, sedimentation, maturation, and diagenesis clearly generate kerogen with a restricted range of values (Figs. 1, 4E, 4G; Table 2). The most robust comparison for this study is of Precambrian carbonaceous shales with modern and Phanerozoic counterparts (Tables 1 and 2).
Carbonaceous shales sampled in the 2.2 Ga Paleoproterozoic lower greenschist facies Birimian volcanic-sedimentary terrane, Ghana, have δ15N values ranging from 9.3 to12.6‰ (Table 2, Fig. 4C) (Jia and Kerrich, 2004b). Boyd and Philippot (1998) reported bulk rock values of 8.4–16.6 ‰ for the mid-amphibolite facies Moine succession, Scotland, deposited between 1.5 and 1.0 Ga (Table 2). Micas separated from these samples yield similar isotopic compositions of 7.4‰–16.2‰. Assuming an oxidized atmosphere-hydrosphere system by this time in the Mesoproterozoic, and therefore contemporaneous organic matter similar to δ15N in modern marine sediments of −3‰ to 9‰ (Peters et al., 1978; Sweeney et al., 1978; Rau et al., 1987; Williams et al., 1995; Schubert and Calvert, 2001; Sadofsky and Bebout, 2004), they concluded that metamorphism to amphibolite facies had induced a shift of ≥ 8‰. However, if shifts to amphibolite facies are ≤3‰ (see Discussion, below), primary values may have been 3‰–12‰. Haendel et al. (1986) documented 15N-enriched values for the Proterozoic Saschsisches Erzebirge, Germany (11.3‰; Table 2). The 1.6 Ga Tetsa samples, and Cuddapah kerogen data of this study have δ15N values that are indistinguishable from the Phanerozoic data set. In summary, three of the five Proterozoic data sets are enriched in 15N compared to Phanerozoic counterparts (Table 2; Figs. 1, 4C).
Hayes et al. (1983) reported an extensive database on Precambrian sedimentary rocks spanning 3.8–0.8 Ga, including X-ray characteristics, H and C contents, and δD and δ13C values. They reported N contents and δ15N on a subset of ten samples 3.4 to 1.6 Ga in age, of which five are cherts mostly associated with BIF, two are shales, and three are stromatolitic dolomites. δ15N values range from 0.8‰–5.7‰. Chert kerogen has a similar range of δ15N values as in the larger database for Precambrian cherts of Beaumont and Robert (1999). Kerogen N in chert could be mixtures of mantle N having δ15N of ∼−5‰ with enriched oceanic N.
Zhang (1988) documented δ15N ranging from 2‰ to 39‰ in Archean carbonaceous metashales, the most enriched values being in greenschist facies shales from the 2.6 Ga Ventersdorp Group, South Africa (Hayes et al., 1983). Greenschist facies carbonaceous shales in the 2.7 Ga Archean Abitibi Greenstone Belt are characterized by variably enriched δ15N values from 12‰, with an average of 15.3‰ ± 1.8‰ (Table 2, Fig. 4A) (Jia and Kerrich, 2000). Two other ca. 2.7 Ga Archean data sets on unweathered rocks in drill core from the greenschist facies Penhalonga Formation, Botswana, and western Abitibi Greenstone Belt also have overlapping enriched averages δ15N of 17.3‰ ± 1.9‰, and 16.0‰ ± 1.7‰, respectively (Table 2; Figs. 1, and 2A) (Jia and Kerrich, 2004b, 2004c). Contemporaneous samples of this study from the Dharwar Craton endorse systematically 15N-enriched sedimentary kerogen in Neoarchean carbonaceous shales (Table 1, Fig. 1).
Precambrian Chert–Iron Formation
Beaumont and Robert (1999) analyzed nitrogen concentrations and isotopic compositions of kerogen from Precambrian cherts, ranging in age from 3.5 to 0.7 Ga. They found a very large range of δ15N values, from −4.7‰ to 30.0‰, with generally low bulk rock N contents of 2–106 ppm. There is a general increase of δ15N values from −4.7‰ to 5.9‰ in the Mesoarchean (3.5 to 2.9 Ga), through −2.5‰ to 30.0‰ in the Neoarchean (2.8 to 2.5 Ga), to 2.1‰ to 7.7‰ in the Proterozoic (<2.5 to 0.6 Ga). They concluded that Archean atmospheric N2 had a similar isotopic composition to the present, and that the secular change of δ15N records the “great oxygenation event” (Fig. 4A).
Using a step-heating procedure, Pinti et al. (2001a) documented δ15N ranging from −7‰ to 20‰ in Archean chemical sedimentary rocks (3.8–2.8 Ga). Nitrogen released at low-temperature steps, in all samples, had negative or near present-day atmospheric δ15N, which was interpreted to reflect modern atmospheric or/and organic compound contamination; on the other hand, for all but one sample the nitrogen extracted from high-temperature steps showed positive δ15N, with most between 2‰ and 20‰; enriched values were modeled as secondary shifts by Rayleigh volatilization during metamorphism (Pinti et al., 2001a).
In a study of the Cornubian batholith of southwest England, Boyd et al. (1993) reported N concentrations of 8–187 ppm and δ15N values in the range of 5.2‰–10.2‰. According to Bebout et al. (1999), the early Devonian Skiddaw peraluminous granite in the English Lake District has nitrogen contents of 49 ± 27 ppm and δ15N values of 3.5‰ ± 1.1‰ (n = 7). Given that Phanerozoic peraluminous granites have metasedimentary precursors (Hawkesworth and Kemp, 2004), these results are consistent with melting of sedimentary rocks having initial δ15N of 2‰–6‰, which is shifted by ∼3‰ during metamorphism prior to partial melting (Table 2).
Sparse data for 2.7 Ga tonalitic rocks in the Uchi subprovince, Superior Province, range from −5.3 to +5.2‰, averaging −0.9‰ (Jia and Kerrich, 1999, 2000). The Archean tonalite-trondhjemite-granodiorite (TTG) suite is considered to have formed by partial melting of basaltic oceanic crust on a subducting slab (Drummond and Defant, 1990; Martin, 1999; Smithies, 2000). Archean diamond data indicate that Archean oceanic crust may have had a δ15N similar to that of modern MORB (averaging −5‰). Secondary isotopic shifts could arise from some combination of seawater alteration and addition of carbonaceous sediments, which are unconstrained, followed by a shift of ≤3‰ during subduction metamorphism prior to partial melting. Nitrogen contents of the tonalites at 5–27 ppm are comparable with Phanerozoic metaluminous granitoids (Hall, 1999), implying similar concentrations of N in the source.
Orogenic, or mesothermal, lode gold deposits constitute a distinct class of Au- and Ag-rich structurally hosted vein systems. They developed post peak-metamorphism, late in the development of accretionary orogenic belts. This deposit type is common in Neoarchean greenstone belts and Paleoproterozoic terranes, and formed continuously through the Phanerozoic, including large metallogenic provinces in the lower Paleozoic Lachlan orogen and Jurassic–Tertiary of the North American Cordillera. The vein systems extend tens of kilometers laterally, and up to 6 km vertically (McCuaig and Kerrich, 1998; Goldfarb et al., 2001).
Muscovite, or more rarely biotite, is abundant as an alteration phase, and therefore a candidate for determining the N-isotope composition of the hydrothermal fluids and source reservoir. According to H, C, and O-isotope data, and K-Cs-Rb-Ba systematics, the veins precipitated from metamorphic fluids. Given a “deep later” P-T-t path where peak-metamorphism migrated down through the crust, metamorphic fluids were probably generated syn peak-metamorphism at mid-crustal levels, and advected to shallow crustal levels where vein minerals precipitated post peak-metamorphism (McCuaig and Kerrich, 1998). In the 2.7 Ga Abitibi Greenstone Belt, the deposition of sedimentary-volcanic host rock sequences was temporally separated from precipitation of the post-metamorphic hydrothermal micas by only 30 m.y. (Corfu et al., 1989; Kerrich and Cassidy, 1994).
Stable (H, C, O, and S) and radiogenic (Sr, Nd, and Pb) isotope studies of these vein systems of Archean to Phanerozoic age suggest that the fluids acquired a signature of bulk crust, as expected for fluids evolved during regional metamorphism (for reviews see McCuaig and Kerrich, 1998; Hagemann and Cassidy, 2000; Kerrich et al., 2000). The dilute, aqueous dominated, but carbonic- and N-bearing fluids, were generated by dehydration of hydroxyl-silicates for H2O and NH4 +, and by decarbonation of carbonates and/or oxidation or hydrolysis of organic compounds for aqueous carbonic species and N2 (Jia and Kerrich, 2004b).
Hydrothermal micas in Phanerozoic gold quartz veins have δ15N values similar to those in contemporaneous sedimentary rocks (Tables 1–3, Fig. 4). This result is in keeping with the bulk mid to upper crustal N budget being dominated by sedimentary N. Crustal sedimentary rocks contain (0.22 ± 0.10) × 1020 mol N2, mostly in shales where N abundances are hundreds to thousands ppm. The mass of N is comparable in lower crustal igneous rocks (Zhang and Zindler, 1993, and references therein) (Table 5).
Averages of δ15N values for deposits in accretionary orogenic belts are as follows: 3.5‰ ± 0.4‰ for the Paleozoic Lachlan fold belt in southeastern Australia; 3.0‰ ± 1.2‰ for the western North American Cordillera, which hosts many Jurassic–early Tertiary gold-bearing quartz vein systems from the Mother Lode of southern California to Alaska; 4.0‰ ± 2.0‰ for Mesozoic quartz veins in the western Qilian orogen, North China; and 4.9‰ ± 0.6‰ for the middle to late Tertiary Monte Rosa quartz veins in the Alpine orogen (Table 2; Figs. 4F, 4H) (Jia and Kerrich, 2004b).
δ15N values for hydrothermal micas from gold-bearing quartz veins in the 2.1 Ga Ashanti belt of Ghana, and the 1.8 Ga Trans-Hudson orogen, Canada, are 7‰–10‰, intermediate between Archean and Phanerozoic counterparts (Table 3, Fig. 4D) (Jia and Kerrich, 2004b).
For the Archean Superior Province, Jia and Kerrich (1999, 2000) reported data for hydrothermal micas from nine 2.7 Ga deposits (Abitibi belt: Kerr-Addison, Dome, Hollinger, Goldhawk, Beaumont; Wawa belt: Hemlo; Geraldton-Beardmore belt: Geraldton, Pickle Crow; Red Lake belt: Red Lake); the total range of δ15N values is 11.8‰–21.0‰, with a mean of 16.3‰ ± 3.0‰ (Table 3, Fig. 2B). Hydrothermal micas from the Norseman terrane, Western Australia, and Harare Greenstone Belt, Zimbabwe, are also characterized by 15N-enriched values, averaging 17.3‰ ± 4.3‰ and 20.7‰ ± 3.0‰, respectively. Including the 15N-enriched data for the Kolar deposit of this study, high δ15N values of late-metamorphic hydrothermal micas have been recorded on four Neoarchean cratons (Tables 1,3; Fig. 4B).
δ15N values of micas in the 1.8 Ga Jerome deposit are within the range for Proterozoic siliciclastic rocks and kerogen. Hall (1989) documented an increase of N content from 1 ppm to 182 ppm (average 53 ppm), correlated with secondary K addition, in basalts altered by seawater. High N contents of 144–238 ppm have been reported from submarine hydrothermal fluids in the Sea of Cortez (Von Damm et al., 1985). Nitrogen in zones of potassic submarine hydrothermal alteration could be sourced in seawater, organic-rich sediments, hydrolysis of K-silicates in the host volcanic rocks, or some combination. Given bottom seawater N contents of 0.5 ppm (Létolle, 1980), we tentatively interpret this variably enriched signature in the VMS alteration zone as reflecting N-bearing organic compounds in ambient sediments. Micas in the Archean deposits have the same high δ15N values as contemporaneous shale kerogen, consistent with a sedimentary N source.
Rare Metal Pegmatites
Rare metal pegmatites containing Mo and W occur in per-aluminous, S-type domains of the post-metamorphic 2.6 Ga Preissac-Lacorne post-tectonic batholith, Abitibi Greenstone Belt, Canada (Feng and Kerrich, 1992). Muscovites are characterized by δ15N values of 1.6‰–5.3‰, with an average +3.2‰ (Jia and Kerrich, 1999, 2000). Peraluminous granites are considered to result from melting of average mid-crust (Sylvester et al., 1997; Hawkesworth and Kemp, 2004). Neoarchean greenstone belts are dominated by syntectonic TTG batholiths, with supracrustal volcanic-sedimentary sequences. Notwithstanding uncertainties in melt-residue fractionations, these results are consistent with fusion of a mix of relatively 15N-depleted tonalites and 15N-enriched clastic sedimentary rocks (Table 1, Fig. 1).
Nitrate deposits formed in arid climates possess up to 163,000 ppm N, and δ15N ∼0‰ consistent with atmospheric deposition of N (Böhlke et al., 1997). Sparse data on NH4 +–bearing K-feldspars from a variety of epithermal springs and deposits in the western United States range from 2700 to 19,000 ppm N, with δ15N of −0.6‰–12.3‰ interpreted as N mobilized from sedimentary rocks (Krohn et al. 1993).
Nitrogen isotopic compositions of Archean shale kerogen, hydrothermal gold quartz vein systems that proxy for average crust, and seawater-altered micas in the VMS deposits all show 15N-enriched values. The three types of samples show parallel secular trends from 15‰–24‰ at 2.7 Ga, through intermediate values in the Proterozoic, to 3‰–4‰ in the Phanerozoic (Tables 1–3, Fig. 4). We first constrain the magnitude of possible isotope fractionation during metamorphism or Rayleigh volatilization, then address possible long-term (chronic) diffusional loss of N.
Effects of Metamorphism
The sample design of this study involves comparing pre-metamorphic kerogens from carbonaceous shales with late-metamorphic hydrothermal micas from the eastern Dharwar Craton. Also, published data for pre-, and late-metamorphic samples from other Precambrian terranes are compared. These sample types permit testing for large shifts of δ15N during metamorphism, as proposed by Boyd and Philippot (1998), Beaumont and Robert (1999), and Pinti et al. (2001a).
From inspection of Tables 1–3 and Figure 4, it is clear that the δ15N values of sedimentary kerogen are consistent with those of post peak-metamorphic hydrothermal micas through geological time. Jia et al. (2001, 2003a) and Jia and Kerrich (2004c) showed that hydrothermal micas in the Paleozoic Lachlan accretionary orogen and the Mesozoic–Cenozoic western North American Cordillera have δ15N of 3.5‰ ± 0.4‰ (n = 20) and 3.0‰ ± 1.2‰ (n = 100) respectively, comparable to Phanerozoic bulk metasedimentary rocks and average crustal values (Haendel et al., 1986; Bebout and Fogel, 1992; Mingram and Bräuer, 2001; Busigny et al., 2003), and Phanerozoic kerogens (Williams et al., 1995; Ader et al., 1998; Kao and Liu, 2000; Sephton et al., 2002). This result endorses the use of hydrothermal micas to proxy for crust in the Precambrian.
Mean δ15N values of the 2.2–2.1 Ga pre-metamorphic carbonaceous shales and post peak-metamorphic hydrothermal micas hosted in the Birimian sediments overlap at 10.8‰ ± 1.1‰ and 10.2‰ ± 1.5‰, respectively (Tables 2,3; Figs. 4C, 4D). Kerogen and hydrothermal biotites from the eastern Dharwar Craton are both enriched in 15N. Similarly, δ15N values of the 2.7 Ga Three Nations carbonaceous shales (16.0‰ ± 1.7‰) and VMS micas (16.5‰ ± 2.9‰) from the Abitibi belt are both enriched, and comparable to the isotopic composition of hydrothermal micas (16.3‰ ± 2.9‰) in the same terrane (Figs. 4A, 4B) (Jia and Kerrich, 1999, 2000, 2004c). The hydrothermal quartz-mica veins precipitated post peak-metamorphism (Kerrich and Cassidy, 1994), and accordingly their δ15N values cannot have been influenced by peak-metamorphic conditions.
These results demonstrate minimal isotope fractionation of nitrogen between metasedimentary rocks, hydrothermal fluids, and minerals precipitated from the fluids. They also rule out any significant shifts in δ15N during metamorphism to greenschist facies, or mid amphibolite for the Sandur shales, consistent with the results of previous studies as indicated below.
Progressively Metamorphosed Sediments
Bebout and Fogel (1992) reported data for progressively metamorphosed sedimentary rocks of the Catalina Schist complex, California (Tables 2,4; Fig. 5A). They obtained δ15N of 2.2‰ ± 0.6‰ in low-grade rocks (350 °C) and 4.3‰ ± 0.8‰ in amphibolite facies equivalents (600 °C). They calculated fluid rock (N2-NH4 +) N-isotope fractionations of −1.5 ‰ ± 1‰ with the Rayleigh distillation equation at temperatures ranging from 350 to 600 °C.
Mingram and Bräuer (2001) also found shifts of <2‰ in δ15N from low-grade carbonaceous shales (300 °C) at 2.2‰ ± 0.6‰, through greenschist facies equivalent (470 °C) at 3.5‰ ± 0.9‰, to amphibolite facies mica schists (550 °C) at 3.9‰ ± 0.8‰ (Table 4, Fig. 5B). In a more recent report on N-isotope fractionation due to metamorphism, Busigny et al. (2003) found that the δ15N values of metasedimentary rocks from the Schistes Lustrés complex (Western Alps), which were subducted from shallow level to depth of 90 km, are between 3.1‰ and 4.8‰ and show no systematic isotopic shifts with increasing metamorphic grade (Tables 2,4; Fig. 5C). A more closed system behavior may explain why shifts were smaller than in the Catalina.
Haendel et al. (1986) documented shifts of ∼1‰ from sediments to greenschist facies counterparts, and ≤3‰ to amphibolite facies in progressively metamorphosed siliciclastic sequences of the Sachsisches Erzgebirge, Germany (Table 4, Fig. 5D). However, Haendel et al. (1986) emphasized that the total range of δ15N may not be attributable to progressive metamorphism alone because (1) the different sedimentary lithologies may have had intrinsically different primary N contents and δ15N values; (2) there were multiple metamorphic events in some sequences, but not others; (3) there is a range of age of the lithologies from Ordovician to Late Precambrian; and (4) the Precambrian gneisses alone are characterized by the most 15N-enriched values.
Hence, there is systematic 15N-enrichment in the three Proterozoic data sets (see Table 2), the Erzegebirge (δ15N = 11.3‰ ± 4.0‰), Moine (13.7‰ ± 2.6‰), and Ashanti belt (10.8‰ ± 1.1‰), notwithstanding the fact that the former two are at amphibolite facies, but the Ashanti belt is prevalently lower greenschist facies (Tables 2 and 4, Jia and Kerrich, 2004b). Accordingly, the “metamorphic trend” of Haendel et al. (1986) could be reinterpreted in terms of a secular evolution of crustal N and δ15N, in keeping with this study
Experimental and Theoretical Studies
Ader et al. (1998) observed a decrease of nitrogen content but uniformity of δ15N in anthracite at temperatures of up to 600 °C, in keeping with the empirical studies of N in progressively metamorphosed sedimentary rocks as indicated above (Table 4). Hanschmann (1981) calculated nitrogen isotope fractionations between NH4 + in solid and fluid (N2) phases; interpolation of the data yields fractionation of ∼–2.25‰ at temperatures of 350–600 °C, in accord with the results of Bebout and Fogel (1992) and Haendel et al. (1986). Collectively, studies of progressively metamorphosed terranes, theoretical and empirical studies, as well as data for hydrothermal micas, show that the isotope composition of N2 in metamorphic fluids is close to that of the source rock reservoir (Hanschmann, 1981; Bebout and Fogel, 1992; Jia et al., 2001, 2003a).
Rayleigh fractionation generates large isotopic shifts where volatilization nears completion (Valley, 1986). Pinti et al. (2001a) model their range of data as a shift of ≥27‰ from primary values of −7‰, stemming from Rayleigh devolatilization using fractionation factors of Hanschmann (1981) and no constrained f (the fraction of residual nitrogen in the rocks) values. Modeling was on a mix of data from 3.8 Ga amphibolite facies chert-BIF and 3.4 Ga greenschist facies cherts. Two 15N-enriched samples of Pinti et al. (2001a) are amphibolite facies metasedimentary rocks having C/N ratios of ∼2000 and 5000 that have experienced multiple Archean metamorphic events and a Proterozoic disturbance (e.g., Rollinson. 2002). Of the remaining four samples, two metamorphosed to high-grade have δ15N intermediate between two low-grade samples. Accordingly, there is a data cluster, rather than a correlation of δ15N values with either C/N or metamorphic grade, and hence no progressive Rayleigh fractionation with increasing grade (Fig. 6 of Pinti et al., 2001a).
Within the seven Precambrian data sets there are only small trends of increasing δ15N to lower N contents as expected for Rayleigh processes (Table 2; Figs. 3, 4). As a corollary, in the Onverwacht data of Beaumont and Robert (1999) two samples with the most positive δ15N have greater N contents than two of the more negative.
Rayleigh effects can also be addressed via the quartz-mica vein systems, which are precipitated from fluids generated by loss of volatile species during progressive metamorphism. If Rayleigh effects were significant during metamorphic devolatilization, then H, C, O, N, and S isotope values would be depleted in the veins relative to ambient crust, but this is not the case (McCuaig and Kerrich, 1998).
Constraints from C/N and δ 13C
Precambrian kerogens collected in this study do not show correlations between C/N ratios and δ13C or δ15N values (Fig. 6). Both δ15N and δ13C values of samples, for each given age, shift ≤3‰ over the span of C/N ratios, consistent with previous estimates (Haendel et al., 1986; Watanabe et al., 1997).
A range of δ13C values (−18‰ to −48‰) has been reported from greenschist facies sedimentary units in 2.7 Ga Superior Province supracrustal terranes (Schoell and Wellmer, 1981; Wellmer et al., 1999). According to Wellmer et al. (1999), no correlations exist between δ13C and H/C or C/N. Rather, they interpreted variations in δ13C values of kerogens as resulting from organic matter synthesized by different metabolic pathways. The two Paleoproterozoic sample sets from the Cuddapah Basin (−27‰ and −31‰) are interpreted in this way (Table 1). Two Archean kerogen sample sets from the Abitibi belt and Botswana, each from greenschist facies carbonaceous shales, possess distinct δ13C populations averaging −20‰ and −26‰, respectively (Table 1 of Jia and Kerrich, 2004b), consistent with different processes during deposition, as suggested by Wellmer et al. (1999). These results complicate the evaluation of metamorphic effects on kerogen samples compiled in Table 2, including new data from India, but endorse preservation of primary signatures (Table 1, Fig. 4).
Long-Term Diffusional Effects
Possible long-term kinetically induced diffusional loss of 14N from kerogen or hydrothermal micas has been evaluated with respect to the K-Ar system. Whereas N, as NH4 +, substitutes for K in K-silicates, radiogenic Ar gas is not structurally sited in silicates. Diffusion-dependent blocking temperatures are known for amphibole (∼500 °C), biotite (300 °C), and muscovite (350 °C) (McDougall and Harrison, 1988). The sample design for kerogen and N-isotope analysis in four terranes 2.7 Ga in age and three Proterozoic terranes was predicated on U-Pb zircon and 40Ar/39Ar ages on granitoids within a few tens of million years. Given undetectable Ar loss in these terranes, N-isotope values should also be primary (Jia and Kerrich, 2004b).
Given that Ar-loss is grain-size and temperature dependent (McDougall and Harrison, 1988), the compliance of 15N-enriched data between fine-grained pre-metamorphic micas in VMS deposits and kerogen, with coarse-grained late-metamorphic micas from the same terrane provides further constraints against preferential diffusional loss of 14N (Table 3).
The effects on preferential loss of 14N of (1) structural sitting of N between kerogen and silicates (Jia and Kerrich, 2004b), (2) metamorphism, and (3) age can also be evaluated from data sets for Phanerozoic rocks. There are no obvious differences in δ15N between these three classes of samples from the Tertiary to Cambrian (Table 5, Fig. 7).
Kerogen for this study, as well as other Precambrian kerogens analyzed by Jia and coworkers, was selected for minimal secondary disturbance on the following criteria: (1) preservation of sedimentary structures; (2) large distance from VMS or gold deposits; (3) large distance from faults; (4) coherent REE patterns and Eu anomalies; and (5) coherent LILE/REE/HFSE systematics (see Jia and Kerrich, 2004b) (Table 2).
IMPLICATIONS AND CONCLUSIONS
Precambrian Oxidation State
Two conflicting models based on various lines of evidence have been developed for the evolution of oxygen in the atmosphere in the Precambrian: low pO2 in the Archean, with a rapid increase in oxygen ca. 2.3 Ga (Holland, 1999, 2002 and references therein), or alternatively, an oxygenated atmosphere from the early Archean (Ohmoto, 1997, 2004; and references therein). Beaumont and Robert (1999) interpreted their secular isotopic “trend” as a record of changes in the redox potential of Earth. Negative δ15N values reflect an unspecified metabolic isotopic fractionation under anoxic conditions, with microorganisms using reduced forms of nitrogen, whereas positive δ15N values reflect an increase in pO2 after the Paleoproterozoic, which promoted the biologic production of nitrate species (Beaumont and Robert, 1999).
Beaumont and Robert reported data for 12 Mesoarchean cherts (33 analyses), 9 Neoarchean cherts, of which one is split into bedded and homogeneous domains (19 analyses), and 12 Proterozoic cherts (21 analyses). Some samples were analyzed once, others had duplicate or triplicate analytes prepared, and measurements of some analyses were made more than once. Statistically, for a given population multiple analyses of a sample reveal “within sample” variance, whereas differences between samples represent “between sample” variance (Searle, 1971). Consequently, samples with multiple analyses are over-represented as plotted by Beaumont and Robert (1999), Holland (2002), and Marty and Dauphas (2003).
When the data are replotted to take into account “between sample” variance alone, the major shift is from δ15N averages of −0.5‰ in Mesoarchean chert kerogen, through 7.2‰ in Neoarchean kerogen to Paleoproterozoic and Neoproterozoic averages of 3.9 ‰ and 2.7‰ respectively (Figs. 8, 9). If the model of an anoxic environment until the “great oxidation event” ca. 2.3 Ga is correct, then a rapid rise of pO2 is not reflected in the Archean to Proterozoic N-isotope record, with the implication that the N cycle was not profoundly changed ca. 2.3 Ga. The redox state of the early atmosphere, and the timing and mechanism(s) of redox transitions, remain controversial (Kasting, 1993; Ohmoto, 1997; Holland, 1999; Phillips et al., 2001; Ohmoto, 2004) and beyond the scope of this paper. New approaches are required as suggested by Anbar and Knoll (2002).
Pinti et al. (2001a) reported N and C contents and δ15N from a variety of Archean chert-BIF, where δ15N spans from −7‰–20‰. They identify the most depleted values of −7‰, obtained in high-temperature heating steps, as primary and related to metabolic isotopic fractionation of NH4 + by chemoautotrophic bacteria at Archean hydrothermal vents. In a related discussion paper, Pinti and Hashizume (2001) also suggested that 15N-depleted compositions in Archean cherts could stem from input of mantle N, but their preferred explanation for values of −6‰ obtained in low-temperature steps by Sano and Pillinger (1990), and 15N-depleted values in the data set of Beaumont and Robert (1999), is for chemoautotrophic processes.
All data sets from Precambrian chert-BIF, metamorphosed from greenschist to amphibolite facies, are characterized by (1) some depleted δ15N values (−5‰); (2) a large range of δ15N values, up to 30‰, in greenschist facies cherts; and (3) an absence of correlation of δ15N with either N content or C/N ratios. The third observation rules out Rayleigh fractionation as a cause of enriched values.
Precambrian chert and BIF are often interbedded; their spatial and temporal association with volcanic sequences erupted from mantle plumes has been used as evidence for a hydrothermal origin of these sediments (Barley et al., 1998, and references therein). Isley and Abbott (1999) and Condie et al. (2001) established a statistical correlation of the time-series for BIF and volcanic sequences erupted from mantle plumes, corroborating the empirical association (Fig. 9).
Consequently, the large spread of values reported by Beaumont and Robert (1999) and Pinti et al. (2001a) may be primary. Pinti et al. (2001a) measured δ15 N of −7.4‰ in sample Pano D-136 from 3.5 Ga cherts of the Pilbara Craton, Western Australia, for which 40Ar/36Ar at 58,500 and N2/36Ar ratios are close to those estimated for the upper mantle of δ15N = −5 ± 2‰, 40Ar/36Ar = 42,000, and N2/36Ar = (5 ± 2) × 106 (Marty and Zimmermann, 1999). Independently, Pinti et al. (2001b) reported Xe-isotope evidence for input of mantle volatiles to the Pilbara cherts.
In light of these observations and constraints on the magnitude of metamorphic δ15N shifts to ≤3‰, we reinterpret the range of δ15N values from depleted to enriched in all of the data sets for Precambrian chert-BIF as mixing between a 15N-depleted mantle source emitted from plumes and a 15N-enriched marine sedimentary kerogen component, identified by Jia and Kerrich (2004a, 2004b) and this study (Fig. 3A). Given a mantle plume association, BIFs are not a record of near-surface redox conditions.
Nitrogen contents of micas from orogenic gold deposits mirror the secular variation in N content of carbonaceous shales (Table 2, Fig. 4B). This result is consistent with the breakdown of sedimentary kerogen as the primary source of N in the hydrothermal fluids from which the micas were deposited, with dehydration of K-micas as a secondary source of N. This interpretation is also in keeping with the second-order observation that Archean micas are characterized by a larger range of δ15N than Phanerozoic equivalents (Tables 2,3,5). Jia et al. (2001) accounted for this distribution in that Archean terranes are volcanic dominated, with subordinate sedimentary units, whereas Phanerozoic accretionary belts possess a higher proportion of turbidites. Consequently, metamorphism of Archean terranes would generate metamorphic fluids having a variable N content and a δ15N depending on the proportions of metaigneous (low N content, depleted δ15N) and metasedimentary (high N content, enriched δ15N) rocks. Metamorphism of Phanerozoic terranes would generate fluids in which the N-budget was dominated by metasedimentary rocks. This interpretation is also consistent with fluid inclusions, characterized by high N2 content (0.6–99.0 mol) from quartz-gold veins in organic-rich slate belts (Bottrell et al., 1988; Ortega et al., 1991; de Ronde et al., 1992).
Hayes et al. (1983) documented a general trend of increasing δ13C with decreasing organic C content for Precambrian sedimentary rocks. They modeled the trend as a Rayleigh fractionation where δ13C of residual organic C shifted by ∼6‰ for each factor-of-ten loss of C. However, both δ13C and TOC overlap for their Archean and Proterozoic data sets (Figure 8 of Hayes et al., 1983). In addition, differences of δ13C and δ15N between populations of samples within given Archean and Proterozoic terranes are preserved (Table 1) (Wellmer et al., 1999; Jia and Kerrich, 2004b). We interpret increasing bulk N contents of sedimentary rocks as follows: Most Archean siliciclastic sequences are first-cycle volcanogenic turbidites shed off of bimodal arcs into tectonically active basins. Stable passive margins developed in the Proterozoic, receiving cratonic detritus at relatively slow sedimentation rates (Taylor and McLennan, 1985). Accordingly, organic compounds were diluted by detritus in the former setting relative to the latter (Jia et al., 2001).
The implication from the shale and mica record is that Nfixing microorganisms have progressively drawn down atmospheric N2, to sequester it as NH4 + in crustal K-silicates, increasing the crustal N-inventory from an original basaltic value of 1–2 ppm (Table 6). These processes concurrently shift atmospheric and crustal δ15N values down. An analogous process is sequestration of atmospheric CO2, with transfer to the crustal carbonate and kerogen budgets. According to Delsemme (1998; 2001), the crustal carbonate budget translates into a Hadean atmosphere with 20 times the present CO2.
Origin of the Archean Atmosphere-Hydrosphere
Archean sedimentary kerogens and crustal hydrothermal systems as recorded in this study, together with compilations from the literature, signify systematically 15N-enriched values of 15‰–24‰ (Tables 2,3; Fig. 4). If the isotopic fractionation between kerogens and atmosphere in the Precambrian was approximately the same as at present, then it implies a 15N-enriched atmosphere in the Archean (13‰–21‰ ca. 2.7 Ga). It is possible that the 15N-enriched values stem from a different N-cycle in the Archean, with large biologically mediated fractionations, yet the magnitude of the fractionations exceeds any presently known.
Kramers (2003) conducted an evaluation of Earth's volatile budget. He compared volatile element abundances in the outer Earth reservoirs (OER: atmosphere-hydrosphere, continental crust, MORB-source mantle) with average carbonaceous chondrites (CC) on an Al-normalized basis. Normalized abundances of I, Br, N, and C decrease relative to CC in a manner consistent with mass-dependent hydrodynamic loss during energetic impacts and extreme UV (EUV) solar radiation. Kramers accounted for overabundance of H and Cl by their residence dominantly in an ocean, and for Ne-isotope characteristics of mantle plumes by incorporation of ∼10− 7 of Earth's volatile budget from solar atmosphere. The results of this study are consistent with the model of Kramers (2003). An implication is that if isotopic fractionation of N accompanied its hydrodynamic loss, then Earth's secondary atmosphere could have acquired an initial δ15N >CI chondrite.
Several authors have proposed an E-chondrite model for Earth based on rare 15N depleted diamonds (Javoy, 1998; Tolstikhin and Marty, 1998). Those diamonds at ∼–25‰ (Cartigny et al., 1997, 1998) are close to the range of δ15N in E-chondrites of −15‰ to −43‰ (Kung and Clayton, 1978; Grady et al., 1986). However, a largely pure E-chondritic mantle can be ruled out from chemical and isotopic compositional data of Earth (e.g., Allègre et al., 2001; Drake and Righter, 2002), Cr-isotope data of the mantle and various classes of meteorites (Shukolyukov and Lugmair, 1998), and 15N-enriched Archean carbonaceous shales (this study).
The best material to fit the isotope characteristics for 15N-enriched atmosphere and carbonaceous shales is CI-chondrite-like material having δ15N of 30‰–42‰ (Kerridge, 1985) at the end of Earth accretion ca. 4.5 Ga, and/or comets formed in the vicinity of Jupiter, which likely had the same N-isotopic composition as CI-chondrites, because they originated in the same zone of the early solar system. Volatiles in CI-chondrites and comets may be similar (Delsemme, 2001).
After considering the δ18O and other isotopes, and the major element abundances (e.g., Mg/Si vs. Al/Si) in Earth and chondritic meteorites, Drake and Righter (2002) ruled out either E- or CI-chondrites as source materials for Earth. Rather, they considered Earth to have formed by accretion in a narrow feeding annulus at ∼1 astronomical unit (AU), close to its present orbital radius. They argued that Earth's budget of water cannot have been acquired primarily from a late veneer of CI chondrites in the asteroid belt, and that comets originating in the Oort cloud at 2.6 AU can deliver no more than 50% of Earth's minimum water budget; consequently, Earth's water is indigenous. However, Robert (2001) argued that the source of Earth's water is consistent with current understanding of the water content of the asteroid belt, as inferred from the chondritic meteorite record and from the mean of D/H ratios of clay minerals in carbonaceous chondrites, which is close to standard mean ocean water. According to Morbidelli et al. (2000), from early stages of accretion to late stage gas-free sweep-up of planetesimals, water was delivered to Earth from a mix of objects in the asteroid belt, in the vicinity of the giant planets, and in the Kuiper Belt.
If the 15N-enriched values of Archean samples reflect a commensurately enriched atmosphere, then it is possible that N is not dominantly indigenous, but was acquired from 15N- enriched solar system materials including CI-chondrite type material and comets. The secular trend of 15N in shale kerogen documented in this study reflects a corresponding atmospheric trend from >21‰ ca. 2.7 Ga to 0‰ now: the shift can be accounted for by a combination of (1) mantle degassing (continuous addition of N2 with δ15N −5‰); (2) progressive sequestration of atmospheric N2 into crustal rocks by nitrogen fixing organisms, with a return flux of 15N-depleted N2 stemming from diagenetic fractionation; and (3) recycling of 15N-enriched Archean sediments into the mantle. Such nitrogen recycling has been proposed by Zhang and Zindler (1993). Given the limited database for Archean rocks this interpretation is of necessity speculative, and other explanations may emerge. Isotopic mass balance for N is difficult given uncertainties in mantle N-content, mass of subducted material, and the conservative or volatile behavior of N at convergent margins. Kerrich and Jia (2004) attempted a simplified model, which indicated that subduction has caused a shift of <0.1‰ in the mantle δ15N using the “chert”-like kerogen recycling assumptions of Marty and Dauphas (2003).
We thank Steve Kesler and Hirishi Ohmoto for the invitation to submit this manuscript. We are grateful to M. Stocki for assistance with the nitrogen analysis in the Department of Soil Sciences, University of Saskatchewan; P. Lindgren for orchestrating a field trip to Jerome; and K.M. Ansdell, T. Oberthür, B. Pratt, A. Still, and S. Vearncombe for providing some of the samples. Y. Jia acknowledges receipt of a CSIRO postdoctoral fellowship, an honorary position at Monash University, and a research grant from SEG Foundation, Inc., USA. R. Kerrich acknowledges a Natural Sciences and Engineering Research Council (NSERC) Discovery Grant, an NSERC MFA grant, and the George McLeod endowment to the Department of Geological Sciences, University of Saskatchewan. The critiques of H. Ohmoto, S. Kesler, D. L. Pinti and an anonymous journal reviewer significantly improved an earlier version of the manuscript.
Figures & Tables
Evolution of Early Earth's Atmosphere, Hydrosphere, and Biosphere - Constraints from Ore Deposits
- amphibolite facies
- banded iron formations
- chemically precipitated rocks
- clastic rocks
- Dharwar Craton
- gold ores
- greenschist facies
- Indian Peninsula
- iron formations
- isotope fractionation
- isotope ratios
- Kolar schist belt
- massive deposits
- massive sulfide deposits
- metal ores
- metamorphic rocks
- mica group
- organic compounds
- organic nitrogen
- P-T conditions
- quartz veins
- secular variations
- sedimentary rocks
- sheet silicates
- stable isotopes
- total organic carbon
- Sandur greenstone belt
- Jerome Deposit