Nitrogen Isotopic Composition and Density of the Archean Atmosphere

See allHide authors and affiliations

Science  04 Oct 2013:
Vol. 342, Issue 6154, pp. 101-104
DOI: 10.1126/science.1240971

Same As It Ever Was

Nitrogen constitutes approximately 78% by volume of Earth's atmosphere and is a key component in its chemical and physical characteristics. It is not clear whether N2 has been so abundant throughout Earth's geological history. Marty et al. (p. 101, published online 19 September) analyzed the isotopic compositions of nitrogen and argon from fluid inclusions trapped in hydrothermal quartz formed 3 to 3.5 billion years ago. The partial pressure and isotopic composition of atmospheric N2 were similar to today's. Thus, other factors are needed to explain why liquid water existed on Earth's surface despite the Sun being 30% less luminous.


Understanding the atmosphere’s composition during the Archean eon is fundamental to unraveling ancient environmental conditions. We show from the analysis of nitrogen and argon isotopes in fluid inclusions trapped in 3.0- to 3.5-billion-year-old hydrothermal quartz that the partial pressure of N2 of the Archean atmosphere was lower than 1.1 bar, possibly as low as 0.5 bar, and had a nitrogen isotopic composition comparable to the present-day one. These results imply that dinitrogen did not play a significant role in the thermal budget of the ancient Earth and that the Archean partial pressure of CO2 was probably lower than 0.7 bar.

Nitrogen is a key element of planetary atmospheres that is a sensitive tracer of exchanges of volatile elements between planetary interiors and outer space (13). On Earth, the amount of N at Earth’s surface might have varied significantly with time, as a result of exchange with the deep Earth (1, 4) or of loss to space (3, 5). In contrast, the amount of nonradiogenic (not produced by nuclear reactions) noble gas isotopes in the terrestrial atmosphere is unlikely to have varied significantly since Earth's formation (6). Calibrating N against a nonradiogenic noble isotope such as 36Ar in ancient samples provides a way to estimate the variations of the partial pressure of atmospheric N2 (PN2) in the past.

We estimate here the PN2 in the ancient terrestrial atmosphere from the analysis of the N2/36Ar ratios in well-characterized fluid inclusions trapped in hydrothermal quartz from the Archean 3.49-billion-year-old (Gy-old) Dresser (710) and 3.46-Gy-old Apex Basalt (11, 12) formations in Pilbara Craton, Western Australia. Previous geochemical studies of fluid inclusions from this area (9, 1116) and from other geological settings (17) have shown the presence of a paleoatmospheric component, probably incorporated as atmospheric gases dissolved in surface waters. The abundances of atmospheric gases in water are a direct function of their respective partial pressures and of the salinity and temperature of water (18). For the present-day atmospheric partial pressures of N2 (0.7906 bar) and 36Ar (3.20 × 10−5 bar), the N2/36Ar ratio of air-saturated water (ASW) ranges from 1.02 × 104 to 1.31 × 104, for water temperatures between 2°C (the present-day average deep-sea temperature) and 70°C [a proposed model temperature for the Archean oceans (19)] and salinities between 0 and 16 weight % equivalent of NaCl [encompassing the range of salinities observed in Archean Dresser and Apex fluid inclusions (9, 11, 12)]. Nishizawa et al. (13) have shown that, in fluid inclusions trapped in silica dykes and quartz veins from the Dresser Formation, N2 is the main N species, far more abundant than the other identified species (NH4+). These authors proposed an upper limit of 3.3 times the modern value for the Archean N2/36Ar ratio.

Fluid inclusions trapped in three different Archean hydrothermal quartz samples were analyzed (912, 15, 16). The choice of the samples was driven by the goal of targeting pristine fluids of marine and/or meteoritic origin. Ideally, the best samples to be investigated would be sedimentary rocks. However, precipitated minerals, such as carbonates, sulfates, and halides, are relatively weak mineral phases as compared to quartz, seldom preserving primary fluid inclusions. Silicified sediments were another option, but the grain size is generally too small to preserve fluid inclusions large enough to analyze (several micrometers). For this reason, we chose to investigate quartz-bearing amygdules preserved in komatiitic basalt flow (sample PI-06) at the base of the Dresser Formation and recovered through drilling, and quartz-bearing pods filling retreat voids in exposed pillow basalts (PI-02-39) at the top of the Dresser Formation and within the Apex Formation in the Warranoowa syncline (PB-02-122). Considering the high silica content of Archean seawater and hydrothermal fluids, such voids were probably filled with quartz immediately upon cooling. The fact that the intrapillow pods form well-defined ovoid shapes isolated in the core of the pillow indicates that fluid circulation processes driving mineral precipitation should have occurred through a porous medium shortly after basalt deposition. In order to maximize sampling of primary fluids of surficial origin (seawater and meteoritic) and to minimize the imprint of hydrothermal fluids, all samples were collected in undeformed rocks located in close proximity to the overlying marine/lacustrine sediments. This, together with the shallow-water character of both the Dresser and Apex settings, indicates that the samples investigated contain fluids of marine and/or lacustrine origin, in addition to hydrothermal fluids.

The PI-06 quartz sample is from a drill core (Pilbara Drilling Project 2) in the Dresser formation and was selected in the depth interval from 102 to 110 m. The quartz fills 2- to 10-mm vesicles in komatiitic basalt at the base of the Dresser Formation and contains 2- to 10-μm-sized fluid inclusions. The formation of quartz in amygdules in komatiitic basalt must have occurred early in the lava post-emplacement history, because neither deformation nor pressure tracks were observed (15). Fluids trapped in the inclusions have been interpreted as being a mixture of Archean surface water and hydrothermal fluid (15, 20). Ar-Ar analysis of trapped fluids indicates that they are ≥3.0 ± 0.2 Gy old (15) and that they contain inherited 40Ar, presumably from a hydrothermal end member.

Samples PI-02-39 and PB-02-122 are from isolated quartz-carbonate aggregates forming pods hosted in pillow basalts now exposed at the surface. These pods resemble typical mineralization structures associated with passive hydrothermal circulation of water through shallow crust. Intra-pillow quartz crystals, which were selected for analysis, contain abundant, 1- to 25-μm, two-phase (liquid and <5% CO2 vapor) aqueous inclusions (9). Fluid inclusions are randomly distributed throughout the host quartz, which argues for a primary origin. The absence of crosscutting veins, metamorphic overprint, and deformation features affecting pillow basalts and associated pods indicates negligible fluid remobilization and circulation after deposition and crystallization. Foriel et al. (9) demonstrated that several fluids were trapped in PI-02-39, such as an evolved Archean water component and several (Ba- and Fe-rich) fluid components. A previous noble gas (Ar and Xe) study of sample PI-02-39 (16) has shown that the quartz has preserved, since the meso-Archean era, a paleoatmospheric component having a 40Ar/36Ar ratio of 143 ± 24, mixed with an hydrothermal fluid component rich in Cl, K, and inherited 40Ar . The Ar-Ar analysis of PI-02-39 indicates that fluids are >2.7 Gy old, probably contemporaneous to the deposition of the Dresser formation (16). Although not investigated in detail for its fluid inclusion composition, sample PB-02-122 shows the same type of fluid inclusion distribution and texture as sample PI-02-39. Further sample characteristics, field locations, and photos are given in the supplementary materials (20).

Several aliquots of each sample were crushed under vacuum using different numbers of crushing steps, and the extracted gases were sequentially analyzed for N and Ar isotopic abundances by static mass spectrometry (table S1). For all runs, the N and Ar abundances correlate, showing that it is unlikely that much N in the fluid inclusions has been consumed by postentrapment chemical reactions. Before analysis, two aliquots of sample PI-02-39 were neutron-irradiated, in order that the extended Ar-Ar technique (21) could be used to determine K and Cl in the same extractions as N and Ar. The combined analyses of Cl, K, Ar, and N isotopes confirm previous studies (9, 1316) showing that trapped fluids are mixtures of a low-salinity, low-N, and low–radiogenic 40Ar end member, with several hydrothermal components rich in N, Cl, K, and radiogenic 40Ar.

For all samples, step-crushing data define well-resolved mixing correlations in a 40Ar/36Ar versus N2/36Ar diagram that are consistent with mixing between several hydrothermal fluids and a low–40Ar/36Ar, N2/36Ar end member (Fig. 1). The slopes of the correlations differ widely between samples, indicating variable enrichment of 40Ar in the different trapped hydrothermal end members. The lowest measured N2/36Ar ratios in step-crushing runs of PI-02-39 aliquots range from 0.72 ± 0.03 × 104 (the last crushing step of PI-02-39-3, table S1 and fig. S3) to 1.43 ± 0.02 × 104 (the first crushing step of PI-02-39-4). A higher value of 3.13 ± 0.02 × 104 was obtained for aliquot PI-02-39-2; however, we suspect the contribution of hydrothermal N in this case (20). We consider a N2/36Ar range of 0.7 × 104 to 1.4 × 104 as representative of the low-salinity end member of sample PI-02-39. Sample PI-06 presents much lower 40Ar/36Ar ratios, in the range from 386 to 1014, than those of sample PI-02-39, in agreement with the less evolved character of its trapped fluids (15, 20). Despite this difference, its lowest N2/36Ar ratios (0.85 × 104 to 1.50 × 104, table S3) are in the same range as those of sample PI-02.

Fig. 1 N-Ar isotope variations for inclusion fluids trapped in Archean quartz.

Open, black, and gray symbols are duplicate step-crushing runs for samples PI-02-39-39, PI-06, and PB-02-122, respectively. (B) is an enlargement of the zone outlined with dashes in (A). The lines in (B) represent upper and lower limits for sample PI-02-39 data points, which converge toward ASW. This sample contains a mixture of several hydrothermal fluids having different N/36Ar compositions, with a Cl-poor water component. ASW: air-saturated water with N2/36Ar values corresponding to modern atmospheric partial pressures of N and Ar (table S3) and with 40Ar/36Ar ratios <298 (the modern value). The corresponding modern partial pressure of atmospheric N2 is also indicated. A two to three times higher N2 pressure, as suggested by (1, 2), would correspond to the gray bar on the right-hand side of ASW. Such a higher value is not supported by the convergence of data point correlations.

The convergence of data points toward a common end member (Fig. 1) constrains the possible range of N2/36Ar ratios for the presumed Archean ASW value, which is consistent with the modern ASW value. Based on the results summarized in table S3, we propose that the Archean ASW N2/Ar ratio was ≤ 0.7 × 104 to 1.4 × 104, comparable to that of modern ASW (1.0 × 104 to 1.3 × 104). Assuming a constant concentration of atmospheric 36Ar since 3.5 billion years ago (Ga) (6) implies that the Archean PN2 was 1≤1 bar (scaled to the modern PN2 of 0.79 bar), and possibly as low as 0.5 bar.

Variations of the N isotopic composition (expressed as δ15N relative to the modern atmospheric value) are also consistent with mixing between a crustal/sedimentary end member (δ15N within 3 to 8 per mil (‰) for metagabbros (4) and 5 to 15‰ for sediments (4, 14, 22, 23), and an Archean atmospheric δ15N value within ~2 to 3‰ of the present-day value (Fig. 2). A modern-like N isotope ratio in the Archean is in agreement with the conclusions of a near-constant atmospheric δ15N through time from the analysis of ancient cherts (14, 23), although others (24) have proposed drastic 15N enrichments of atmospheric N2 during the Archean eon. The near-constancy of the atmospheric δ15N value and of PN2 since 3.0 to 3.5 Ga is consistent with the presence of a significant terrestrial magnetic field in the Archean. In the absence of such magnetic shielding, atmospheric dinitrogen would have interacted with charged particles from the solar wind, resulting in the nonthermal loss to space of this element and N isotope fractionation (3), as in the atmospheres of Mars (25) and Titan (26). In order to efficiently shield the amosphere against N2 loss, a magnetic field at least 50% of the present-day intensity would have been required 3.5 Ga (3), which is in agreement with paleomagnetic data from 3.2 Gy-old single silicate crystals (27).

Fig. 2 N isotopic composition (δ15N is the deviation in parts per mil from the modern atmospheric 15N/14N ratio of 3.6765 × 10−3) versus the 36Ar/N2 ratio for all extraction steps (small gray symbols) and for the total extracted gases of each sample (large black symbols).

In this format, mixings will yield straight lines. All extractions were done by crushing except for a heating run for PI-02-39 (gray and black squares). ASW is the modern air-saturated water composition (36Ar/N2 computed with data given in table S3, δ15N = 0‰). The range of crustal and sedimentary values is also indicated (4, 14, 22, 23). Observed ratios in Archean fluid inclusions are consistent with mixing between typical crustal/sedimentary values from the hydrothermal end members and an ASW component having a N isotopic composition comparable to the modern one within ~2 to 3‰.

An Archean atmospheric PN2≤ 1.1 bar, together with a N isotopic composition similar to that of the present-day atmosphere, has important implications for the thermal conditions of Earth’s surface. The energy delivered by the ancient Sun might have been as little as 70% of what it is today, requiring other sources of energy or, more probably, a larger greenhouse effect than today (28). The presence of greenhouse gases such as NH3 and CH4 has been proposed [(29) and references therein], but their stability in the ancient atmosphere has been questioned. Others have postulated the occurrence of higher PCO2 in the past, although the geological record of Archean sedimentary rocks suggests that the PCO2 could have been only a few times larger the present-day value (29). Two studies (1, 2) have proposed that a PN2 two to three times the present-day one [with the presence of other gaseous species such as H2 (2)] was sufficient to maintain a clement surface temperature, with partial pressures of greenhouse gases consistent with the geological record. The present results are not consistent with these proposals and can be used to set an upper limit of the maximum Archean PCO2 pressure. A recent study (30) based on fossil imprints of rain droplets proposed an atmospheric pressure <2 bar at 2.7 Ga, probably less than an upper limit in the range from 0.5 to 1.14 bar, which is similar to our estimate of the PN2 in the Archean atmosphere (0.5 to 1.1 bar). Although the ages of our Archean samples may differ, this comparison suggests by the difference between the two estimates that the PCO2 was not more than ~0.7 bar and may have been far less. This is qualitatively consistent with conditions necessary to maintain a temperature of ~15°C at Earth’s surface with a mixture of CO2 and other greenhouse gases (28, 29).

Supplementary Materials

Materials and Methods

Figs. S1 to S4

Tables S1 to S3

References (3441)

References and Notes

  1. The excellent match between the atmospheric 38Ar/36Ar ratio (two Ar isotopes of primordial origin) and that of Ar trapped in primitive meteorites attests that atmospheric Ar was not quantitatively lost to space [noble gas isotopic ratios are sensitive to thermal and nonthermal escape processes (5)]. Independently, noble gas isotope systematics indicate that mantle degassing did not contribute quantitatively to the atmospheric inventory of these elements after Earth’s formation. The large difference in the 40Ar/36Ar values between the mantle and the atmosphere implies that less than 2% of mantle 36Ar has been transferred to the atmosphere since the time of Earth’s formation (31). A similar conclusion is obtained from 36Ar flux estimates. The modern flux of 36Ar from the mantle is ~2000 mol/year [from scaling to the present-day 3He flux from the mantle (32)]. Integrated over 3.5 Gy (the age of the formations studied here) and assuming a constant flux through time, the amount of 36Ar is only about 0.1% of the 36Ar atmospheric inventory (5.55 × 1015 mol). The degassing rate of the mantle was probably higher in the past because of a higher thermal regime. Coltice et al. (33) estimated that the melting rate of the mantle was a factor of ~20 higher 3.5 Gy ago as compared to the present-day rate. In this case, the mantle contribution of degassed mantle 36Ar to the atmosphere represents only 0.8% of the 36Ar inventory, which is still negligible in the context of this study.
  2. Supplementary materials on Science Online.
  3. Acknowledgments: This study was supported by Agence Nationale de la Recherche Grant eLife-2 to P.P. and B.M. and by the European Research Council under the European Community’s Seventh Framework Programme (FP7/2007-2013 grant agreement no. 267255 to B.M.). P.P. acknowledges the Institut de Physique du Globe de Paris, the Institut des Sciences de l’Univers, and the Geological Survey of Western Australia for supporting the Pilbara Drilling Project. We thank six anonymous reviewers and D. Catling for constructive comments. This is Centre de Recherches Pétrographiques et Géochimiques contribution no. 2248 and Institut de Physique du Globe de Paris contribution no. 3424.
View Abstract

Stay Connected to Science

Navigate This Article