The history of the Arctic Ocean remained poorly known until the 2004 IODP coring of Lomonosov Ridge sediments. Early studies of the recovered sequence demonstrated the existence of an Eocene ‘lake-stage’ prior to the transition to marine conditions. The marine stage onset was inferred to be ∼17.5 million years -Ma- ago, thus implying a nearly 26 Ma gap between the lacustrine and marine episodes, and an unusual tectonic history for Lomonosov Ridge, in order to explain this gap. More recently, Rhenium-Osmium (Re-Os) isotope measurements of the transition from the lacustrine to marine sediments suggested a much earlier inception of marine conditions and the absence of any significant gap between both episodes. Here, an improved Osmium isotope stratigraphy and Re-Os data concur to assign a Late Eocene age (∼36 Ma) to the marine invasion, consistent with a relative change in sea level on top of Lomonosov ridge, either from tectonic origin or from another cause.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The Cenozoic history of the Arctic Ocean was very poorly known until the 2004 Arctic Coring Expedition (ACEX), when a sedimentary sequence beginning at about 55 million years (Ma) was obtained from Lomonosov Ridge (Figure S1 of the auxiliary material). Core description and biostratigraphic information used to develop the first age model for this sequence can be found elsewhere [Moran et al., 2006; Jakobsson et al., 2007; Backman et al., 2008]. Briefly, the sequence revealed a unit represented by black shale-like sediment spanning the late Paleocene and most of the Eocene. This unit includes fresh surface water conditions, as revealed by the presence of Azolla fresh-water fern in the sedimentary record [Brinkhuis et al., 2006]. Henceforth, we will refer to this period as the Arctic ‘lake-stage’, keeping in mind that it represents an euxinic setting, with stratified low-salinity surface water and deeper more saline water [Stein et al., 2006; Stickley et al., 2008]. The 44.4 to ∼18.2 Ma succeeding interval has been interpreted as a sedimentary hiatus due to erosion/non-deposition linked to stalled subsidence of the Lomonosov Ridge, and possibly uplift [O'Regan et al., 2008a]. Above a short transitional environment unit, the upper section has been assigned to a Miocene to Holocene marine interval. In that scenario, the opening of Fram Strait, allowing oxygenated Atlantic water to penetrate into the Arctic Basin is dated at ca.17.5 Ma, whereas the marine deposition over the ridge accompanied the resumed subsidence of the Lomonosov Ridge. More recently [Poirier and Hillaire-Marcel, 2009], we presented preliminary Re-Os data from the same core, leading to question both this timing of the marine transition and the existence of the 26 Ma sedimentary hiatus.
 Here, we report revised Re-Os isochron ages and complementary Os-isotope measurements, along with new carbon and nitrogen data, allowing for a better understanding of the redox state of the sediment when deposited, as well as for a better assessment of the timing of events involved. Based on a now complete Re-Os data set, we propose an Os-isotope stratigraphy of the Arctic basin spanning the last 45 Ma that includes a ca. 36 Ma-long marine stage. Our alternative chronostratigraphy of the ACEX core seems to be in agreement with other information about the geology of the Arctic Ocean, such as the timing of the separation of Yermak Plateau and Morris Jessup (see below).
2. Materials and Methods
 Sub-samples from the ACEX core were provided by the Bremen IODP core management facility and treated in GEOTOP laboratories following common procedures for Rhenium (Re) and Osmium (Os). Because both elements are redox-sensitive, concentrations of organic carbon (Corg.) and total nitrogen contents (Ntotal) were measured in bulk sediment in order to assess Re and Os-linkage with Corg fluxes. Re and Os were recovered by standard Carius Tube digestions of the samples, using inverse aqua regia (2 parts HNO3+ 1 part HCl) at 240°C, followed by liquid extraction of Os in Br2 and purification by microdistillation, while Re was separated by anionic exchange resin chromatography [Shirey and Walker, 1995; Birck et al., 1997; Meisel et al., 2003]. Osmium isotopic composition was measured on a VG Sector-54™ in negative ion mode [Creaser et al., 1991]. Re isotope dilutions measurements were obtained on a Micromass IsoProbe™. High-temperature catalytic combustion of samples with a Carlo-Erba NC2500™ furnace was used to determine Ntotal and Ctotal contents, whereas coulometric titration (UIC, Inc.) yielded inorganic carbon (Cinorg.) contents. Corg concentrations were calculated from Ctotal and Cinorg values. Recalibration of our Re-Os spike against geochronological standard material (Oligocene molybdenite [Markey et al., 2007]) leads to an update of the preliminary data set presented by Poirier and Hillaire-Marcel  and the calculation of slightly younger ages than previously estimated (∼2 Ma younger). We now correctly report all depth according to the revised composite depth scale for the ACEX core [O'Regan et al., 2008b]. The corrected and complemented data set is given in Table S1 and illustrated in Figures 1 and 2 (and in Figures S1 and S2).
 Elemental abundances, shown in Figure 1, suggest that the budget of the redox-sensitive metals Re and Os is tightly controlled by the organic carbon content of the sediment and thus by redox-conditions in bottom waters and sediments during deposition. The Corg. content of the lake-stage sediments is approximately one order of magnitude higher than in the overlying marine sequence (previously shown by Stein et al. ). Accordingly, Re and Os contents are approximately two orders of magnitude and twofold higher, respectively, in the lake-stage sediments. The lowest Corg content was found in the first oxidised levels (160–195 mcd), where Re and Os contents are minimum as well. In addition, whereas Corg./Norg. ratios generally exceed 20 in these sediments, they are significantly lower (as low as ∼5) in the marine unit. Maximum values of Re and Os concentrations, as well as of the Corg. content and Corg./Ntotal ratio, highlight a final phase of organic-rich sediment deposition.
 The high Re- and Os-concentrations (and org-C) observed in the lacustrine unit indicate that reducing conditions prevailed in the sediment and possibly in the deep-water column during this episode, whereas Os-isotopes in the overlying marine unit yielded an initial signature consistent with oceanic Os [Pegram and Turekian, 1999; Peucker-Ehrenbrink and Ravizza, 2000a; Burton, 2006]. As already demonstrated by Poirier and Hillaire-Marcel , the layers defining respectively the top of the lacustrine unit and the bottom of the marine unit yielded linear isochrons with very similar ages, now better constrained (Figure 2). Data at 200.38 mcd, on top of the lacustrine unit, just below the inferred hiatus, can be fitted through a regression line yielding an age of 36.64 ± 0.74 Ma (3-point isochron; MSWD = 0.26). At 196.5 mcd, just above the inferred hiatus, data yield an age of 36.2 ± 2.2 Ma (MSWD = 14). The first age is ∼8 Ma younger than the stratigraphic age estimate of 44.4 Ma at a core depth of 198.7 mcd. On the other hand, our 36.6 Ma isochron at 200.38 mcd seems to be in excellent agreement with biostratigraphic information from diatoms, suggesting a late middle to early late Eocene age in the 202.5–203.5 mcd interval (at 36.7 Ma [Stickley et al., 2008; Backman et al., 2008]. Above the inferred hiatus, the Re-Os isochron yields an age much older than the Burdigalian age (16–20 Ma) assigned to the early marine sediment based on biostratigraphic inferences from dinocysts [Sangiorgi et al., 2009]. We cannot rule out, in an absolute manner, the possibility that the 196.5 mcd regression represents a mixing line. However, we emphasize some facts regarding this: 1) both Re-Os regression ages are within error of each other; 2) the intercepts of these two regressions correspond to a) a continental source below the 198.7 mcd hiatus, during lake-stage, and b) a marine source (at 36 Ma) for the marine-stage above hiatus; and finally 3) the initial Os isotopic compositions for post-ventilation sediments closely follow the pattern of global marine Os evolution for the 36 Ma to present interval (Figure 2). These facts all provide additional compelling evidences in support of a much older lacustrine to marine stage transition. The high to very high level of Re-Os authigenic enrichment above fractions linked to detrital sources in the alternating black and gray interval above the 198.7 mcd level (the so-called zebra zone, Table S1 and Figure S3), leads to conclude that the organic-rich sediments, above the inferred hiatus, were deposited under newly inflowing marine waters, resulting in an overprint of the inherited, detrital sediment Os-isotope composition, by low 187Os/188Os-ratio marine osmium. Because there is such a drastic change in the 187Os/188Os ratio (Figure 2) that accompanies the Re-Os enrichment (Figure S3) just above 198.7 mcd, any sedimentary mixing scenario would nonetheless imply a major “external” marine input, otherwise the 187Os/188Os ratio would have remained radiogenic. In our scenario, the transition from the Arctic Lake to the Arctic Ocean, linked to the opening of Fram Strait, had to be marked by a drainage/flooding event, through a crustal stretching-related corridor within proto-Fram strait. It has been previously shown that the width of the strait (and wind-forcing) is more important than depth to ventilate the Arctic basin [Jakobsson et al., 2007; Thompson et al., 2010]. Whereas a drainage event might have been quite short (within error bars of the Re-Os isochrons), the subsequent inception of full marine conditions in bottom waters might have required more time (Figure 3). The setting of these marine conditions has been recorded by a ‘transitional estuarine sea phase’ (the zebra-zone of Jakobsson et al. ), followed by full marine oxic conditions from sub-unit 1/4 upwards. The duration of this marine inception phase, recorded between 198.7 and 193 mcd (excluding potential sediment erosion at 198.7 mcd), could have required up to 2 Ma, if and only if the sedimentation rate remained constant from the Re/Os-isochron age to the oldest 10Be age, above. For comparison, Jakobsson et al.  estimated a duration of 0.7 Ma for this inception phase.
 Calculating a new age model from our Re-Os isochron and the oldest 10Be datum (12.31 Ma at 151.28 mcd [Frank et al., 2008]) yields a new linear apparent sedimentation rate for the 151 to 198 mcd interval, of ca. 1.8 m/Ma (ca. four times lower than Backman et al. , see Figure 3 for age models comparison). Using this rate, the deposition period of the low metal and low organic section (between ∼193 and 155 mcd), would have required about 16 Ma. We emphasize that the age model for this initial oxic interval is still provisional, since no intermediate age is available. Os-isotope stratigraphy is not truly conclusive for this sub-section of sedimentary deposits, since the oceanic Os signature during this time interval remained quite stable, and can thus be contracted or expanded without any visible effect on the Os isotopic signature (Figure 2). Moreover, we suspect that the low metal section was deposited at a faster pace than suggested by the rate derived by simple linear interpolation. Such a very low sedimentation rate with close to nil-authigenic enrichment would tend to enhance the relative abundance of (micro-) meteoritic particles. Such cosmic dust regularly accumulates on the surface of the planet (30000 ± 15000 tons/year [Peucker-Ehrenbrink and Ravizza, 2000b]), and this would effectively drive the Re-Os contents of the sediments upwards, and the 187Os/188Os composition to low values (chondritic material has a typical 187Os/188Os ∼ 0.13, with 0.6 ppm Os). The interpolated low sedimentation rate yields ∼40 picogram of chondritic Os that should have been added to the detrital fraction per gram of sediments; this should have driven the 187Os/188Os composition to much lower values which we do not find. This might imply a variable sedimentation rate for this section and/or a yet unrecognized hiatus above the Re-Os isochron-ages (see Bruvoll et al.  about this last possibility).
 We note that there are published geodynamic models suggesting a connection between the North Atlantic and the Arctic much earlier than 17.5 Ma [Jokat et al., 2008], while Arctic aerogeophysical data were interpreted to suggest a rifting between Morris Jesup Rise and Yermak Plateau (Northern section of the Fram Strait), which must have occurred before Chron 13 (before 34 Ma) [Brozena et al., 2003]. All of which concurs with our temporal interpretation. The suggested closer temporal proximity to the Eocene thermal maximum might also help to account for the surprisingly high TEX86 surface water temperatures found above the inferred hiatus [Sangiorgi et al., 2008]. Another consequence of the new age model presented here, is to give a slightly younger age for the onset of seasonal sea-ice evidence found in the deeper part of the ACEX record [Stickley et al., 2009], from 47.5 to about 44 Ma.
 An increase in Re-Os-Corg contents (Figure 1) occurred just before the 198.7 mcd level, i.e., nearly at the end of the lacustrine stage, and continues until full oxic conditions seem settled in the Arctic Basin. This increase first corresponds to high 187Os/188Os initial ratios up to a depth of ca. 200 mcd, and is seen as the result of authigenic-Os enrichment resulting from diffusion of radiogenic continental-like Os (originating from the enormous watershed of the Arctic Lake) from the water column to the sediments. Above 200 mcd, which in view of the above revised chronology might now be seen as a transitional “catastrophic” layer linked to the marine inception event, thus presenting some erosional features, a steadier sedimentation resumed, with still high values of Re-Os-Corg contents, but with now very low 187Os/188Os initial ratios (similar to contemporary oceanic water, Figure 2). This implies a strong dominance of seawater derived Os (from the inflowing low 187Os/188Os North Atlantic at ∼36 Ma) that was buried in the organic-rich sediment. A budget calculation of the Re and Os deposited on Lomonosov Ridge (roughly 1800 km × 110 km) within the sub-unit 1/5 (assuming constant thickness of 5.7 m) implies an input of seawater of >0.01 Sv for the period the whole sedimentary sub-unit accumulated, if the Os and Re contents of the 36 Ma seawater were similar to present-day levels; ca. 10 and 8200 pg/kg respectively. If this sub-unit is not confined solely to the Lomonosov Ridge (i.e., Arctic basin wide), then the required inflow will be larger. The low 187Os/188Os ratio of oceanic water at the end of the Eocene Epoch, fully captured here at the very bottom of the marine layer (Figures 2 and S3), is generally attributed to meteoritic impacts [Whitehead et al., 2000; Paquay et al., 2008], which, on top of ‘resetting’ the oceanic Os isotopic composition to a low value, would have significantly raised its content in Re and Os, as a transient perturbation in Re-Os fluxes. Some particulate impact-derived Os could also have been involved here, increasing the content of these metals in sediments, given that the late Eocene Popigai impact site is located on nearby Siberian continental crust.
 The fact that some of the highly radiogenic Os-rich material from the late lacustrine stage and the marine inception event has been eroded from the Lomonosov Ridge, and perhaps from other areas of the former Arctic Lake, due to strong currents during this event, might suggest that some of this Os of continental origin might have been exported from the basin into the open ocean within the same time-interval. This Os-pulse might have contributed to the global oceanic ‘recovery’ to more radiogenic compositions, as observed following the Eocene-Oligocene Os-isotopic minimum. On the other hand, the highly anoxic nature of the sediments on both sides of the inferred hiatus suggests that water mass settings at that time would have tend to render osmium (and Re) very particle reactive and thus, be sequestered in nearby sediments. Future long cores from other parts of the Lomonosov Ridge and from the surrounding deep Arctic basins might help to better estimates this.
 The deepest oxic layers (above ∼193 mcd; Figure 2), deposited during marine conditions, depict a sharp decrease in organic carbon, Re and Os contents. The very low organic carbon content of these sediments would have precluded strong authigenic enrichment in Re and Os, which are both present as (relatively) stable oxy-anions in oxic seawater. In this respect, we concur with previous published views that the inflow of North-Atlantic water resulted in oxic conditions beginning above subunit 1/5 [e.g., Jacobson et al., 2007; Backman et al., 2008].
 The Re-Os data set shed some light on the ventilation and oxidation stage recorded in subunit 1/5, either related to a basin-wide event or to local events (i.e., ridge crest subsiding with penetration of a distinct water mass). In Figure S3, one can see, from the shift in the initial 187Os/188Os ratios to low values in the organic-rich sediment, the first inflow of marine water coincident with the inferred hiatus at 198.7 mcd. Moreover, the level of enrichment in Re-Os during this very early marine stage implies a rather large amount of marine water input, suggesting basin-wide ventilation (otherwise the radiogenic inputs would dominate the signal, although the basin might have kept a somewhat stratified waters configuration). Here again, geochemical data from other deep Arctic sites, including tracers that are not so sensitive to redox conditions are needed to fully address this question.
 Finally, we note that remains of agglutinated benthic foraminifers (including species known to have appeared during the Oligocene in the Atlantic [Kaminski, 2007]) were found within the zebra zone [Sangiorgi et al., 2008], indicative of the decline of anoxic bottom water condition. Our data do not invalidate the possibility that a relatively nearshore/shallow environment characterized the Lomonosov Ridge during deposition of the ∼198.7 mcd-deep sediment [i.e., Sangiorgi et al., 2008], as long as the sediment was bathed by enough marine water to overprint the Os isotopic composition of the detrital fraction.
 The revised Re/Os ages from this study essentially close the gap in the ACEX core and thus significantly modify the acknowledged chronology and tectonic evolution of Lomonosov Ridge. In this new scenario, the sedimentary unit between the lake-stage and marine sequence (Figure 2) recorded a marine inflow event resulting from an early opening of Fram Strait, 36 Ma ago; thus, an inception of the Arctic Ocean about 18 Ma earlier than previously estimated. We conclude at a first inflow of North Atlantic waters, perhaps through corridors, linked to early seafloor spreading stages of a proto-Fram strait [e.g., Jakobsson et al., 2007]. This scenario involves a smooth relative change in sea level for the Lomonosov Ridge (from tectonic origin or else, since our data cannot discriminate). Based only on the available ACEX core, it seems difficult to shed more light into the processes leading to the onset of marine conditions in the Arctic basin. Similar Re-Os investigations on long cores from other Arctic ridges and deep basins, yet to be obtained, are now needed to further constrain the timing and mechanism leading to the ventilation of the Arctic Basin.
 We are grateful to A. Zimmerman (AIRIE, Colorado) for having provided molybdenite Re-Os reference material. Constructive reviews from Matthew O'Regan and an anonymous reviewer helped to improve significantly our manuscript. The Canadian Foundation for Climate and Atmospheric Sciences (award to the Polar Climate stability network) and NSERC-Canada (discovery grant of CHM) are acknowledged for funding.
 The Editor thanks Matthew O'Regan and an anonymous reviewer for their assistance in evaluating this paper.