The extraterrestrial dust accretion rate on Earth at Dome C, Antarctica: a fresh look with 3He

The extraterrestrial dust accretion rate on Earth at Dome C, Antarctica: a fresh look with 3 He G. F´ enisse a,* , D.V. Bekaert a , P.-H. Blard a,b , J. Duprat c , I. Mattia d , M. Genge d , M.D. Suttle d , O. Barres e , C. Engrand f , Y. Marrocchi a a Centre de Recherches P´ etrographiques et G´ eochimiques, CNRS, Universit´ e de Lorraine, UMR 7358, 15 Rue Notre Dame des Pauvres, 54 500 Vandoeuvre-l` es Nancy, France b Laboratoire de Glaciologie, D´ epartement de G´ eosciences, Environnement et Soci´ et´ e, ULB, Brussels, Belgium c Institut de Min´ eralogie, Physique des Mat´ eriaux et de Cosmochimie, UMR 7590, Mus´ eum National d’Histoire Naturelle, CNRS, Sorbonne Universit´ e, Paris 75 231, France d Department of Earth Science and Engineering, Imperial College London, South Kensington, London SW7 2AZ, UK e Universit´ e de Lorraine, CNRS, GeoRessources Laboratory, France f Centre de Sciences Nucl´ eaires et de Sciences de la mati` ere (CSNSM), Universit´ e Paris-Sud, UMR 8609-CNRS/IN2P3, 91 405, Orsay, France ARTICLE INFO Editor: Dr O Mousis Dataset link: https://doi. org/10.24396/ORDAR-152 Keywords: Cosmic dust Concordia station Dome C Comic dust extraction Granulometric-sorting Noble gas database 3 He ET and cosmic dust mass fluxes ABSTRACT Interplanetary dust particles (IDPs) and micrometeorites (MMs), from 1 µm to 5 mm, are the primary source of extraterrestrial (ET) material currently accreted on Earth. The flux of ET particles smaller than ~50 µm is typically determined through optical counting, but it remains uncertain and may deviate from predictions made by numerical simulations. The volatile element content carried by this flux is still not well-constrained and is influenced by the potential effects of atmospheric heating. We developed a clean, pressurized system to extract cosmic dust from ~38 kg of clean snow collected near the Concordia station (Dome C, Antarctica). We measured helium isotope concentrations in various granulometric fractions (>62 µm, 25–62 µm, 5–25 µm and <5 µm). The inferred global 3 He ET annual flux is (1.25±0.03) × 10⁻¹² ccSTP⋅cm⁻²⋅ka⁻¹ (weighted mean±1SD), consistent with previous 3 He ET flux estimates from marine sediments and polar samples. Our data shows that the majority of the 3 He ET flux (70 %) is carried by particles in the 5–25 µm size range, with 20 % attributed to the 25–62 µm fraction. Using an empirical relationship between 3 He ET concentrations and cosmic particle mass, we convert these fluxes into a global ET mass flux for particle diameters <100 µm of (3.5±0.5) kilotons⋅a⁻¹ (weighted mean±1SD). This result is about 3 times higher than collection estimates from (Rojas et al., 2021) and aligns with CABMOD-ZoDy modeling, after atmospheric entry (Carrillo-S´ anchez et al., 2020). This 3 He ET method is suited for detecting particles smaller than 100 µm, while collection results are more relevant for larger fractions. 1. Introduction Interplanetary Dust Particles (IDPs; ranging in size from <1 µm up to a few tens µm) are collected in outer space and the stratosphere, while micrometeorites (MMs; few µm to 1 mm) are extraterrestrial (ET) particles that reached the Earth’s surface. Counting of MMs in deep-sea sediments, deserts, polar ice caps, snow samples, urban environments, as well as observations of IDPs from space and modeling approaches, show that this cosmic dust flux dominates the total mass flux of ET matter arriving on our planet (e.g., Takayanagi and Ozima, 1987; Taylor et al., 1998; Yada et al., 2004; Brook et al., 2009; Genge et al., 2017; Gardner et al., 2014; Rojas et al., 2021). It is by several orders of magnitude larger than the flux carried by larger objects in the meteorite size range (e.g., Wetherill, 1976). The total ET mass flux reaching the top of the Earth atmosphere is 40±20 kilotons.a -1 (Love and Brownlee, 1993), but this number is reduced to a few thousand tons per year by vaporization of the meteoroids during their atmospheric entry (e.g., Duprat et al., 2001; Peucker-Ehrenbrink, 1996). This reduction primarily affects particles with diameters >200 µm (e.g., Farley et al., 1996; Suttle and Folco, 2020). However, most dust flux estimates appear to be method-dependent and are limited to a given size fraction (e.g., Mathews et al., 2001; Plane, 2012; Rojas et al., 2021), making them * Corresponding author. E-mail address: [email protected] (G. F´ enisse). Contents lists available at ScienceDirect Earth and Planetary Science Letters journal homepage: www.elsevier.com/locate/epsl https://doi.org/10.1016/j.epsl.2025.119396 Received 22 November 2024; Received in revised form 4 April 2025; Accepted 28 April 2025 Earth Planet. Sci. Lett. 663 (2025) 119396 Available online 12 May 2025 0012-821X/© 2025 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ). uncertain especially for the smallest cosmic particles (<50 µm, Rojas et al., 2021). Cosmic dust is essentially composed of anhydrous silicate, carbonaceous and hydrous phase aggregates (some of which are rich in volatiles; e.g., Kurat et al., 1994; Engrand et al., 2018), Fe-Ni sulfides and magnetite minerals (Mukhopadhyay and Farley, 2006). Two typical types of particles - unmelted (uMMs) and cosmic spherules (CSs) - are defined based on their shape and petrography, that directly result from the temperature they experienced during their atmospheric entry, determined by their mass, entry angle and velocity in Earth’s atmosphere (Flynn, 1989). The transformation of hydrous minerals into olivine and pyroxene takes place at temperatures ≥800 ◦C (Greshake et al., 1998). However, the deposition and preservation of cosmic dust, as well as the collection procedures, may affect these compositions and lead to a sampling bias in favor of the largest particles melted upon atmospheric entry (especially CSs) (Maurette et al., 1987; Taylor et al., 2000; Toppani et al., 2001). While IDPs have also been collected in the stratosphere by high altitude aircrafts from the NASA JSC (Johnson Space Center) (e.g., Brownlee et al., 1994), they remain much less well documented than larger particles. The central regions of Antarctica are an ideal location to constrain the flux of ET dust on Earth since it is far from terrestrial dust sources and anthropogenic contamination (Duprat et al., 2007, 2010; Rojas et al., 2021). At the vicinity of the Concordia station (75 ◦06 ′ S, 123 ◦20 ′ E and 3200 m elevation, Dome C), the snow accumulation rate low and well constrained (2.7±0.1 g.cm -² .a -1 , water equivalent, Frezzotti et al., 2005; Le Meur et al., 2018). This low accumulation rate allows for the accumulation of a significant amount of ET dust in a reasonable volume of snow (Duprat et al., 2007; Rojas et al., 2021) Antarctica remains the only place on Earth where fragile, unmelted particles such as UCAMMs (so-called Ultra-CArbonaceous MicroMeteorites) have been recovered (Duprat et al., 2010). These UCAMMs contain abundant (up to 50 wt.% C) labile organic matter components (e.g., Engrand et al., 2018), pointing to a high preservation potential of cosmic dust in such Antarctic sites. Rojas et al. (2021) used some ultra-clean snow samples of Concordia to collect large numbers of MMs and determine the size distribution of both melted and unmelted MMs in the 20–200 µm size range. The global ET flux deduced from this study is 5.2±1.5 1.2 kilotons.a -1 . For the smallest fractions (<50 µm), the ET particle mass-distribution at Concordia exhibits strong discrepancies with simulations, especially CABMOD-ZoDy, which is the combination of the University of Leeds Chemical Ablation MODel and the Dynamical model of the Zodiacal Cloud from Carrillo-S´ anchez et al. (2020). This apparent contradiction may be due to the limits of the identification protocol, as the optical counting of ET dust can be complicated by the presence of anthropogenic and terrestrial dust contaminations (Rojas et al., 2021). Models simulating ET mass fluxes are limited by assumptions and empirical constraints related to the dust sources (e.g., Nesvorny et al., 2006). New approaches to identity and quantify the ET dust particles in snow samples are thus necessary to improve our understanding of the global ET mass flux on Earth. The use of 3 He is a well-known approach to quantify the abundance of ET particles in terrestrial material (Farley and Patterson, 1995). While 4 He is a radiogenic isotope mainly produced by the radioactive decay of 235 U, 232 Th and 147 Sm at the Earth’s surface, terrestrial 3 He is mostly of primordial origin, inherited from Earth’s accretion (Tolstikhin, 1975). The high 3 He concentrations observed in ET particles are due to both the surface-implantation of solar wind and 3 He production by cosmic rays (e.g., Pepin et al., 2000). Due to their high specific surface areas, the smallest IDPs are thus expected to contain the largest 3 He concentrations from solar wind implantation (e.g., Eberhardt et al., 1965; Stuart et al., 1999). ET and terrigenous reservoirs have distinct 3 He/ 4 He isotopic ratios (~200 Ra vs. 0.02 Ra, respectively, where Ra =1.384 ×10 –6 (Ozima and Podosek, 2002 is the atmospheric ratio). The bulk 3 He/ 4 He measured in terrestrial matrices (e.g., sediments, ice core and firn snow) reflects a mixture of these two reservoirs and is highly sensitive to trace ET contributions, especially in the finest fractions (<50 µm) that are difficult to quantify with optical counting (McGee and Mukhopadhyay, 2013). Since 3 He ET incorporated in sediments has been proven to be preserved for multi-million-year timescales (Patterson et al., 1998; Mukhopadhyay and Farley, 2006), 3 He concentrations in sediment records the ET particle fluxes through geological times and can also be used in some cases to determine sedimentation rates (e.g., Farley and Patterson, 1995; Winckler and Fischer, 2006; McGee and Mukhopadhyay, 2013; Blard et al., 2023). For instance, Takayanagi and Ozima (1987) used marine sediments to estimate an average 3 He ET flux of (1.5±1.0) ×10 -12 cc STP.cm -² .ka -1 (pcc STP.cm -2 .ka -1 , hereafter pcc STP.cm -2 .ka -1 ) over the last 40 Myr. In 2013, McGee and Mukhopadhyay (2013) compiled 3 He ET concentrations from deep-sea sediments over various timescales, offering the possibility to constrain variations in past sedimentation rates and/or 3 He ET flux. Brook et al. (2000) measured 3 He in the GISP2 ice core (Greenland) and 3.8 ka-old ice core from Vostok (Antarctica) to derive global 3 He fluxes of 0.62±0.27 pcc STP.cm -² .ka -1 and 0.77±0.25 pcc STP.cm -² .ka -1 , respectively. Brook et al. (2009) filtered 9 replicates of 3.8 ka-old Vostok ice using 0.45 µm silver filters. These authors derived an average 3 He concentration of (525±174) ×10 –6 pcc STP per gram of ice, and a global 3 He flux of 0.77±0.25 pcc STP.cm -² .ka -1 . They also showed that the 5–10 µm and 0.45–5 µm fractions represent 74 % and 10 % of the total flux, respectively. Additionally, Farley et al. (2021) measured 3 He ET flux of (1.4±1.2) pcc STP.cm -² .ka -1 and (1.2±0.3) pcc STP.cm -² . ka -1 at high temporal resolution in modern air and in a shallow (201 m long) ice core (Antarctica, South Pole), respectively. By measuring the amount of 3 He ET , Os and Ir accumulated in deepsea sediments and assuming average concentrations, Takayanagi and Ozima (1987), Peucker-Ehrenbrink and Ravizza (2000) derived variable estimates for the flux of ET material, with values ranging from 0.4 up to 30 kilotons.a -1 . However, mass estimates of dust based on element abundance measurements are significantly dependent on concentrations used for the abundance-to-concentration ratios, forming a methodological bottleneck. Using MMs recovered from ice cores drilled collection at Dome C, Yiou et al. (1989) deduced an ET flux of ∼1.5 kilotons.a -1 . Using the CS collection (50–700 µm) from the South pole of the water well (bottom part) at the Scott-Amudsen station (South Pole Water, SPWW), the flux and mass distribution of melted cosmic dust was estimated to be 2.7±1.4 kilotons.a -1 (Taylor et al., 1998). The accretion rate before atmospheric entry inferred from craters observed on the LDEF NASA satellite (Long Duration Exposure Facility) is estimated at 20 to 60 kilotons.a -1 in the 20–400 µm size range (Love and Brownlee, 1993). The comparison of this flux with the value inferred by Taylor et al. (1998) suggests that ∼90 % of submillimeter cosmic dust is vaporized during atmospheric entry. In Greenland, using optical counting, Maurette et al. (1987) deduced an ET flux of about 2.2 kilotons.a -1 . However, poor age constraints and the post-depositional weathering of particles from blue ice might bias these results. Here, we aim to document the 3 He ET and the cosmic dust flux recorded in the modern snow of Concordia by analyzing the helium isotopic inventory of variable size fractions of the dust, using a new, pressurized filtering system to separate cosmic dust into distinct granulometric fractions (from <5 µm to >62 µm). Our dataset is used to derive a mean 3 He ET flux, that we compare to previous estimates from various archives (snow, ice and marine sediments). Using a new exhaustive review of He abundances measured in cosmic dust particles of various masses, we use the obtained 3 He ET flux to derive a total ET mass flux for several granulometric fractions from 5 µm to 100 µm. Finally, we discuss the ability of the 3 He ET approach to trace the largest ET particles (>100 µm). G. F´ enisse et al. Earth and Planetary Science Letters 663 (2025) 119396 2 2. Materials & methods 2.1. Cosmic dust in the clean snow from the Concordia station We extracted cosmic dust from ∼38 kg of clean snow from the French-Italian Concordia station, about 1100 km from the closest coast (Fig. 1). Because of the low snow accumulation rate (2.7±0.1 g.cm -2 .a -1 water equivalent; Frezzotti et al., 2005; Le Meur et al., 2018) and the high elevation and remote location, the concentration of cosmic dust relative to terrestrial particles is exceptionally high in the firn, allowing reliable reconstruction of the ET mass flux with low uncertainty (Duprat et al., 2007; Rojas et al., 2021). We processed and melted 4 different snow samples covering the 1960 to 2006 period: 16 kg of snow from F4 snow tank, 15 kg from F6 tank and 4 sub-samples ranging from 1 to 3 kg each from F1 snow tank. Each snow sample was weighed with a 1 g-precision scale before melting. This mass, divided by the average snow accumulation rate of 2.7 g.cm -2 .a -1 , allows us to infer an exposure parameter S (m 2 .a), a key metric to derive an ET flux estimate (Farley et al., 1996; Rojas et al., 2021). The detailed characteristics of the snow samples analyzed in this study are reported in Appendix A. 2.2. Extraction pressurized system and granulometric sorting of ET particles Previous extraction systems used water pumping vacuum or gravitational setting (Yada et al., 2004; Brook et al., 2009) to concentrate MMs. Here, we built a dedicated stainless-steel system to extract cosmic particles from melted snow in a clean room. This extraction system was designed by adapting a “pressure cooker” connected to a nitrogen gas flow high pressure regulator (up to 2 bars). Below this water reservoir, we used three steel filters of 62, 25 and 5 µm mesh sizes. Finally, the melted snow that flowed through the 5 µm filter and carried the smallest particles was evaporated in aluminum capsules of ∼20 mL on hot-plates under a laminar flow hood. This experimental step was designed to tentatively collect and analyze particles smaller than those analyzed in previous studies (e.g., Engrand and Maurette, 1998; Duprat et al., 2007; Brook et al., 2009; Rojas et al., 2021). More information about this extraction system is available in Appendix B. We observed microplastic contaminants of various sizes, mostly on the 5 µm filters, that probably originate from firn tanks (described in Appendix C). This protocol did not permit to check that the largest ET particles were not fragmented during the sieving, nor that smallest particles were not encased in larger terrestrial dusts or plastic contaminants. 2.3. Measurement of Helium-3 and 4 abundances Helium isotopes were measured at CRPG (Centre de Recherches P´ etrographiques et G´ eochimiques, Vandoeuvre-l` es-Nancy) following the procedure described in Blard et al. (2023) and Blard (2021). After heating the filters at 1500 ◦C for 15 min under vacuum in a stainless-steel induction furnace (Zimmermann et al., 2018), extracted gases were purified using three successive charcoal traps cooled in liquid N 2 and four Ti-sponge getters. Once purified, helium was concentrated on a cryogenic head at 12 K, before being released at 75 K into a GV Instrument Helix Split Flight Tube mass spectrometer (Blard et al., 2015). The Helix SFT (Thermo Fisher Helix SFT, Split Flight Tube) was calibrated with a daily analysis of the HESJ gas standard (He Standard of Japan, R =20.63 Ra; Matsuda et al., 2002), yielding 4 He and 3 He sensitivities of 7.04 ×10 13 ±0.5 % mV.mol -1 and 4.16 ×10 18 ±0.9 % cps. mol -1 , respectively. In each carousel, we measured CRONUS-P pyroxene standards that yielded 3 He concentrations in agreement with published values (Blard et al., 2015; Schaefer et al., 2016). Helium measurements were corrected for furnace blanks ((3.10±0.19)x10 9 at of 4 He and below detection limit for 3 He). These blanks accounted for 0.1 to 16 % of the measured 4 He abundance in filtered samples. We also measured and subtracted the helium blank contributions from the steel filters and Al capsules, whose masses are reported in Appendix D. Steel filters contained (3.41±0.27)x10 10 atoms of 4 He.g_steel -1 , while 3 He contents were below the detection limit. These 4 He blank corrections represented from 1 % to 69 % of the 4 He measured in the samples (average 14 %). The Al capsule contained (6.06±0.54)x10 5 atoms of 3 He.g_Al -1 and (1.11±0.13)x10 10 atoms of 4 He.g_Al -1 . Blank contributions from Al capsules accounted for 2 and 28 % of the measured 3 He, and 36 and 39 % of the measured 4 He signals, respectively. Stainless steel filters (5 µm, 25 µm and 62 µm; n =6) and aluminum evaporation-capsules (n =3) resulting from the filtering of the 3 snow tanks (F1, F4, F6, with F1 snow tank being split into 4 different bags to test the homogeneity of particle concentrations in ∼2 kg of snow) were analyzed for 3 He and 4 He Fig. 1. Map of Antarctica showing the locations where cosmic dust has previously been collected and analyzed for light noble gases. NASA stratospheric collections are given in black. Cosmic dust expeditions are more detailed in Appendix A. G. F´ enisse et al. Earth and Planetary Science Letters 663 (2025) 119396 3 concentrations. 3. Results All results are reported in Appendix D, 3 He and 4 He concentrations from each sample and granulometric fractions on Fig. 2. The 3 He concentrations (at.g -1 ) in snow measured for each granulometric fractions vary over several order of magnitudes, ranging from (1.84±0.52)x10 1 at.g -1 to (1.34±0.02)x10 4 at.g -1 , while 4 He concentrations range from < 10 6 at.g -1 up to (3.76±0.01)x10 8 at.g -1 (Fig. 2). Despite inter-tank variability beyond analytical uncertainties, we find a clear granulometric control on He concentrations: both 3 He and 4 He concentrations are the highest in the 5–25 µm particles, followed by the 25–62 µm fractions. The helium concentrations measured in the smallest (<5 µm) and the largest (>62 µm) fractions are by far the lowest. The 3 He/ 4 He ratios measured in all granulometric fractions range from 0.55±0.08 Ra to 85±1 Ra, with most of the values standing between 20 and 40 Ra. Using an isotopic mixing model considering ET and terrestrial endmembers with 3 He/ 4 He ratios of 200 and 0.02 Ra, respectively (McGee and Mukhopadhyay, 2013), our data indicate that >99 % of the measured 3 He is of ET origin in almost all samples (Appendix D). We also find that terrestrial 4 He represents from 29 % to 100 % of the total 4 He measured in these samples, reflecting the inter-sample variability of the terrestrial dust contributions. Combining helium contributions from all granulometric fractions, the total 3 He ET concentrations measured in the three tanks range from (8.17±0.11)x10 3 at.g -1 to (1.52±0.01)x10 4 at.g -1 , and 4 He from (1.61±0.01)x10 8 at.g -1 to (5.19±0.03)x10 8 at.g -1 (sum of all size fractions±1SD) (Appendix D). 4. Discussions 4.1. The global 3 He ET flux and mass-distribution of 3 He-bearing dust 4.1.1. 3 He/ 4 He and He source deconvolution In the two largest granulometric ranges (>62 µm and 62–25 µm), the averages of 3 He/ 4 He ratios from Concordia samples (Appendix D) are similar to those measured in Vostok, 47±5 Ra and 25±4 Ra in >63 µm and 63–20 µm size fractions, respectively (Brook et al., 2009). However, the average of 3 He/ 4 He ratios of the 25–5 µm size rangesis much lower (43±11 Ra) than the value reported in Vostok core (209±5 Ra). Noticeably, theses values are below that reported in individual stratospheric IDPs, which display 3 He/ 4 He ratios around 219±83 Ra (mean±1SD) (Nier and Schlutter, 1992). The observed variability of 3 He/ 4 He ratios in snow samples most probably reflects variable abundances of ET and terrestrial dust fractions. Finally, it is important to emphasize that these 3 He/ 4 He ratios (from 13.6±0.2 Ra to 55.9±0.7 Ra, sum of all size fractions±1SD), demonstrate that 3 He is nevertheless mostly extraterrestrial (>99 %) in all snow samples. 4.1.2. Variability of 3 He ET concentrations per granulometric size range Based on the helium abundances of snow tanks (F4, F6 and F1), MMs smaller than 25 µm represent the largest contribution (65 %) of the total 3 He ET , while ~28 % is carried by 25–62 µm particles (Fig. 2). This result is consistent with previous observations (Brook et al., 2009; Mukhopadhyay and Farley, 2006; McGee and Mukhopadhyay, 2013; Torfstein, 2012), supporting the hypothesis that most of the 3 He ET is carried by cosmic dust smaller than 50 µm. Suttle et al. (2023) demonstrated that fossil MM sizes and He signatures are not directly correlated by Fig. 2. (a) 3 He and (b) 4 He concentrations from different filtered samples of Concordia snow. Three snow tanks were processed, F1, F4 and F6. The red spots indicate the arithmetic means of 3 He ET fluxes for each granulometric size range. The right box called ‘All size ranges’ integrates contributions from all size fractions of each snow sample. During the analysis period, we lost one filter of 25 µm mesh size. G. F´ enisse et al. Earth and Planetary Science Letters 663 (2025) 119396 4 investigation of fossil MM abundance across the Miocene 3 He peak (~8.5 Ma). By modeling the size-dependent heating of ET particles during their atmospheric entry, Farley et al. (1996) established that MMs ranging from 3 to 35 µm bring 70 % of the total 3 He ET flux, with a maximum contribution from the ∼7 to 10 µm particles, in agreement with most observations (Mukhopadhyay and Farley, 2006; Brook et al., 2009; Torfstein, 2012; this study). Importantly, this result is observed in a variety of substrates, including Holocene snow and ice (Brook et al., 2009) and pelagic clays having very different deposition ages, from the modern to the Eocene (Mukhopadhyay and Farley, 2006; Torfstein, 2012). Interestingly, the 3 He ET contribution of the <5 µm fraction differs from that reported in Vostok ice: in the Concordia snow, the smallest fraction yields 3 He ET concentration of 41±1 at.g -1 , ∼18 times lower than the value reported by Brook et al. (2009) (752±54 at.g -1 ). This discrepancy might result from a protocol bias. The cause of this mismatch may be explained either (i) by a loss of the smallest particles (<5 µm) during the evaporation step, via airborne escape in the laminar flow hood, or (ii) the fact that the smallest cosmic dusts may have been encased in larger terrestrial aerosols, hampering their recovery in the proper granulometric fraction during the sieving procedure. 4.1.3. Estimating a global 3 He ET flux Taking into account the mass of processed snow and the local accumulation rate (2.7±0.1 g.cm -2 .a -1 at Dome C, Frezzotti et al., 2005; Le Meur et al., 2018), the inferred exposure parameters range from 0.41±0.02 to 1.24±0.05 cm 2 .ka for the 4 sub-samples of tank F1, and 2.83±0.11, 5.87±0.22 and 5.55±0.21 cm 2 .ka (±1SD) for the full snow tanks F1, F4 and F6, respectively. The total 3 He ET fluxes from each four bags of F1 show a significant variability: (1.52±0.05) pcc STP.cm -² .ka -1 , (0.82±0.02) pcc STP.cm -² .ka -1 , (1.24±0.04) pcc STP.cm -² .ka -1 and (0.92±0.03) pcc STP.cm -² .ka -1 (Mean Square Weighted Deviation (MSWD) =74, Figs. 3 and 4a; Ludwig, 2003). This heterogeneity is probably due to the small masses of ice processed (from 1 to 3 kg). With such small masses, the low number of particles can introduce a statistical bias as well as nugget effects (Farley et al., 1996; Rojas et al., 2021; Fig. 5). Taking into account the 3 He ET concentrations measured in each granulometric fraction, it is possible to compute an average number of cosmic dust particles present in each fraction of the F1 sub-bag samples (e.g., Taylor et al., 2020; Blard et al., 2023). This suggests that there is 0 to 1 particle >62 µm, 1 to 30 particles in the 25–62 µm size range, and 3 to 9 in the 5–25 µm size range. The particle counting statistics are determined by Poisson’s law (sigma =1/  n √) and are compatible with the inter bag heterogeneity observed between the bags of F1 (for masses ranging from 1 to 3 kg). The undersampling effect (coupled with a nugget effect) of the F1 bags, whose He analyses are highly heterogeneous, results in a statistical effect that increases for lower exposure parameters (<10⁻³ m²⋅a), as described by Farley et al. (1996). This statistical effect significantly affects the largest size range. The number of particles present in the ~15 kg of snow from tanks F4 and F6 is much larger (i.e., 2 to 3 particles larger than 62 µm, 20 to 25 particles between 25 and 62 µm, and up to 100 particles in the 5–25 µm fraction), as shown by the agreement between the 3 He ET fluxes determined from these two tanks. Using a weighted mean (Ludwig, 2003) for the four sub-bags of F1, we obtained total 3 He ET fluxes of (1.09±0.13) pcc STP.cm -² .ka -1 for tank F1, (1.27±0.04) pcc STP.cm -² .ka -1 for F4 and (1.25±0.05) pcc STP.cm -² . Fig. 3. 3 He ET flux brought by cosmic dust trapped in the Concordia station (Antarctica). Boxplot compilation of 3 He ET flux (in pcc STP.cm -2 .ka -1 ) are reported for different size ranges, from 5 µm to 62 µm of 6 snow samples. The red spots indicate the arithmetic means of 3 He ET fluxes for each granulometric size range. G. F´ enisse et al. Earth and Planetary Science Letters 663 (2025) 119396 5 ka -1 for F6 tank (±1SD) (Fig. 4b), and a weighted mean 3 He ET flux of (1.25±0.03) pcc STP.cm -² .ka -1 (±1SD, MSWD =0.8). Note that the calculation of the weighted mean approach assumes a gaussian distribution of the three fluxes deduced from these tanks (Appendix E). Fig. 4 summarizes the impact of sample size on the statistical representativity of 3 He ET analysis. This issue is critical for studies using 3 He ET as a proxy of net snow accumulation rate from ice core samples, given that this precious material is only available in limited quantities. If samples of 1 kg are analyzed, then the inter-samples heterogeneity due to the scarcity of cosmic dust implies a relative standard deviation of ~20 % on the estimate of the 3 He ET concentrations. The average of 3 He ET fluxes derived from the three Concordia snow tanks are close to previous literature estimates from snow and ice collected at other Antarctica locations (Fig. 5). To compare these flux Fig. 4. (a) 3 He ET flux results from the 4 sub-bags of the F1 tank (blue), (b) 3 He ET flux from tank F1 (blue, mean of data shown in Fig. 4b), F4 and F6 (yellow) and weighted mean, and (c) 3 He ET flux (pcc STP.cm2 .ka -1 ) vs. exposure parameter (m 2 .a) of each sample. Best fit calculated with the maximum likelihood with overdispersion model of York et al. (2004). Errors are 1 σ . Fig. 5. Comparison of Quaternary 3 He ET fluxes estimated from marine sediments, polar ice cores, and Antarctic snow samples (triangle, square and circle shapes). Blue circle shows our result of 1.25±0.03 pcc STP.cm -² .ka -1 . Horizontal dashed lines show the weighted averages for three different matrices: sediments (0.79±0.06 pcc STP.cm -² .ka -1 ), ice cores (0.78±0.08 pcc STP.cm -² .ka -1 ) and firn snow samples (1.25±0.03 pcc STP.cm -² .ka -1 ). Differences with published value may result from our choice of recalculating weighted mean for all dataset, in a homogeneous way. Colorscale shows exposure parameters in m 2 .a for each study. Sampling stations, locations, and references are available in Appendix E. G. F´ enisse et al. Earth and Planetary Science Letters 663 (2025) 119396 6 results to those from the literature, we recalculated the mean 3 He ET flux of replicates for each study using the weighted mean method (Ludwig, 2003) as shown in Fig. 5. All data and statistical tests are available in Appendix E. Our average flux from the Concordia is notably in agreement with the Vostok ice core and the SPWW firn estimates of 1.11±0.20 and 1.19±0.07 pcc STP.cm -² .ka -1 , respectively (weighted average±1SD) (Brook et al., 2009; Taylor et al., 2020). Interestingly, in Fig. 5, we observe a significant difference between the mean 3 He ET flux recalculated (i.e., using weighted mean from Ludwig, 2003) derived from the firn (1.24±0.03 pcc STP.cm -² .ka -1 ) and those derived from marine sediments (0.79±0.06 pcc STP.cm -² .ka -1 ) and polar ice cores (0.78±0.08 pcc STP.cm -² .ka -1 ). These differences between firn, ice and marine sediments do not seem to result from a latitudinal control (Evatt et al., 2020), but rather from differential conservation of the ET matter. 3 He ET in silicate particles (especially fragile unmelted MMs) could partially escape from old marine sediments. Moreover, the contribution of particles in the >62 µm size fraction to the total 3 He ET flux observed in this Concordia snow (8 %; Appendix D) is lower than the previous estimates (of c.a. 20 %) derived from pelagic sediment analyses by Mukhopadhyay and Farley, 2006. Considering the potential under-sampling factor of 18 for the smallest <5 µm cosmic dusts (Section 4.1.2; Brook et al., 2009), our total 3 He ET flux would increase by only 2 % (1.28±0.03 vs. 1.25±0.03 pcc STP.cm -² .ka -1 ), the contribution of this granulometric class being in any case almost negligible in Antarctic ice (Brook et al., 2009; McGee and Mukhopadhyay, 2013). Fig. 6. (a) 3 He ET concentrations and (b) 3 He ET abundance vs. mass for Antarctica cosmic dust with various degrees of atmospheric (literature database, 14 references in color). For the smallest unweighted particles, calculation assumes mean density of 1 g.cm -3 . The types of particles are reported using different symbol shapes. Regression: log10([ 3 He ET ]) =-(0.94±0.05).log10(Mass) +(15.03±0.68), n =170 and MSWD =30, Dispersion =0.75±0.04. Errors are reported at 1 σ . Liner regressions shown on Fig. 6a are derived from bootstrap aggregating and weighted by ³He analytical errors. G. F´ enisse et al. Earth and Planetary Science Letters 663 (2025) 119396 7 Using a Monte-Carlo modeling approach (described in Appendix E), we computed an average number of cosmic particles per size fraction in the total amount of snow we analyzed here (∼38 kg): 4 particles (m 2 .a 1 ) −1 for 62–250 µm, 35 particles (m 2 .a 1 ) −1 for 25–62 µm and 140 particles (m 2 .a 1 ) −1 for 5–25 µm. No estimate of the number of dust particles <5 µm is provided due to under-constrained He concentrations. 4.2. Cosmic dust database of light noble gas concentrations Helium isotopic abundances have previously been reported for ET particles from several Antarctic locations (Fig. 1) to constrain both the sources and thermal histories of these ET materials (e.g., Stuart et al., 1999; Osawa et al., 2000, 2003 ; Osawa and Nagao, 2002; Baecker et al., 2018, 2022). Here, we compile and exploit an exhaustive dataset of He concentrations to (i) determine an empirical relationship between 3 He ET concentrations and ET particle sizes, and (ii) combine this relationship with the new helium data to derive the mass fluxes of ET material for each granulometric fraction. In practice, we generated an exhaustive database containing all published light noble gas data ( 3 He, 4 He and 20 Ne) measured in IDPs and MMs collected in the stratosphere and Antarctica (Figs. 1 and 6, Appendix F). >40 % of this database is composed of IDPs (stratospheric particles) (e.g., Kehm et al., 2006). In these publications, 65 % of cosmic dust were directly weighed (> 0.1 µg) (e.g., Bajo et al., 2011). For the remaining particles, we rely on their published dimensions to compute their individual weights, as follows: the observed longest (a) and shortest (b) lengths of cosmic dust were used in Equ.1. to compute the equivalent diameter (D eq , in cm): Deq = a.b2 3 √(1) Then, we considered their equivalent spherical volume and assigned an average bulk density ρ of 1 g.cm -3 to all particles (Love et al., 1994). We then calculated individual masses (m, in g) using Equ.2: m=4 3. π .(Deq 2)3 . ρ (2) 4.2.1. He concentration pattern Noble gas concentrations measured in CSs and MMs range over eight orders of magnitude, from 10 8 – 10 17 at.g -1 of 3 He ET . In Fig. 6a, the data reveal a clear anticorrelation between 3 He ET concentrations and sizes of cosmic dust (log10([ 3 He ET ]) =-(0.94±0.05).log10(Mass) +(15.03 ±0.68), n =170 and MSWD =30, Dispersion =0.75±0.04) (Table of Appendix D). This negative correlation was previously suggested by Eberhardt et al. (1965), Suttle et al. (2023) from a smaller dataset. This relationship most likely results from surface-dependance effects, as the helium originating from solar wind implementation is expected to be located at the surface of the ET particle, this effect being amplified by atmospheric entry heating, leading larger particles to be more degassed. The smallest particles are indeed characterized by both the highest surface-to-volume ratios and the highest helium concentrations (e.g., Eberhardt et al., 1965; Stuart et al., 1999). Interestingly in Fig. 6b, we observe that the 3 He ET abundances (from 10 4 to 10 8 at) are greater in the unmelted or partially melted cosmic dust (i.e., uMMs and SCs, Scoriaceous), likely because of lower degrees of atmospheric heating (Farley et al., 1996; Füri et al., 2013). From our 3 He ET database, we note that the heated particles (CSs and SCs) are less abundant than unmelted particles and represent only 32 % of the total (Appendix F). It is worth noting that this proportion of melted particles is similar to the one reported by optical counting from the Concordia firn (39 %, from Table 1 extracted from Rojas et al., 2021), suggesting that there is no sampling bias that would favor undegassed ET particles. Finally, we also note that this database includes a larger proportion of > 100 µm particles than expected from the repartition established by optical counting (Rojas et al., 2021). 4.2.2. He and Ne isotopic ratios and origin of these noble gases in cosmic dusts In Fig. 7, the high 3 He/ 4 He ratios point to a dominant contribution from solar-derived components, with data points plotting near the fractionated solar wind (FSW) endmember ( 3 He/ 4 He ratio FSW =(2.17 ±0.05) ×10 –4 , Ozima and Podosek, 2002). This relative excess of 3 He above the SW (Solar Wind) endmember may indicate an additional contribution of cosmogenic 3 He in these particles. Hence, for the majority of the samples, the distribution 3 He/ 4 He ratios primarily reflects a mixing between SW, FSW and limited atmospheric (EA) and cosmogenic contributions. In a Ne-three isotope plot (Appendix G), most data appear consistent with a dominant solar wind-derived contribution. They show a minor contribution of cosmogenic Ne, in line with previous observations on Antarctic MMs by Baecker et al. (2022). These samples also correspond to the ones that exhibit 3 He/ 4 He ratios well above SW-derived values (Fig. 7), consistent with significant contributions of cosmogenic 3 He and 21 Ne. The lack of large cosmogenic noble gas excesses in the majority of IDPs and MMs suggests rather short lifetimes of cosmic dust particles in interplanetary environments (Pepin et al., 2000), likely due to the Poynting-Robertson effect which decreases cosmic-ray exposure ages for small particles. Based on infrared and lidar observations (LIght Detection And Ranging, Gardner et al., 2014), the CABMOD-ZoDy model combines a chemical ablation model (CABMOD, Carrillo-S´ anchez et al., 2020; Carrillo-S´ anchez, 2020) with a Zodiacal Cloud Model (ZoDy), before atmospheric entry, to predict the ET mass contributions of three dust sources (70 %, 21 % and 9 % from short-period (JFCs), longer-period Halley-type comets (HTC) and asteroid belt (AST), respectively). Similarly, the composition of the flux predicted after atmospheric entry from CABMOD-ZoDy modelling is 87 %, 10 % and 3 % for JFCs, HTCs and AST, respectively. 4.3. The mass flux-distribution of cosmic dust on the filters We use the measured 3 He ET flux observed in the Concordia firn (at. cm -² .ka -1 , shown on Fig. 3) to calculate mean ET mass fluxes for several granulometric fractions (<5 µm, 5–25 µm, 25–62 µm and 62 µm to 100, 250 and 500 µm) (g.m -2 .a -1 , Figs. 9 and 10). For this, we divide the 3 He ET Table 1 Computed 3 He ET concentrations (at.g -1 ) from the two methods (linear regression and weighted mean by the particle masses) considering three different densities. Errors are 1 σ . Size ranges ( μ m) Density <5 5–25 25–62 62–100 62–250 62–500 Weighted mean 0.8 –(7.3±0.5) ×10 14 (4.8±0.4) ×10 13 (9.7±0.7) ×10 12 (4.9±0.4) ×10 12 (1.2±0.1) ×10 12 1–(5.9±0.5) ×10 14 (3.9±0.3) ×10 13 (9.1±0.7) ×10 12 (4.6±0.4) ×10 12 (1.1±0.1) ×10 12 2.2 –(2.7±0.2) ×10 14 (2.4±0.2) ×10 13 (7.6±0.6) ×10 12 (3.9±0.4) ×10 12 (8.2±0.9) ×10 11 Regression 0.8 (3.9±0.7) ×10 18 (1.5±0.4) ×10 15 (4.2±1.1) ×10 13 (5.9±2.0) ×10 12 (1.6±0.6) ×10 12 (6.0±2.2) ×10 11 1 (3.8±0.6) ×10 18 (1.5±0.3) ×10 15 (4.2±1.1) ×10 13 (5.9±1.7) ×10 12 (1.6±0.5) ×10 12 (6.0±2.0) ×10 11 2.2 (3.7±0.7) ×10 18 (1.5±0.5) ×10 15 (4.1±1.5) ×10 13 (5.7±2.4) ×10 12 (1.6±0.7) ×10 12 (5.8±2.7) ×10 11 G. F´ enisse et al. Earth and Planetary Science Letters 663 (2025) 119396 8 flux by the average 3 He concentrations measured in these cosmic dusts (at.g -1 ), as shown in Equ.3: ET MassFluxa−b= 3HeET flux [3HeET]b a ,(3) where [3HeET]b a is the average 3 He ET concentration of each fraction belonging to the size range a-b (<5 µm, 5–25 µm, 25–62 µm and 62 µm to 100, 250 and 500 µm). Since the 3 He ET concentration in cosmic dust is strongly dependent on the dust size (Fig. 6a, Appendix F), we consider different Fig. 7. 3 He/ 4 He isotopic ratios of cosmic dust. Plot of 3 He/ 4 He ratios vs. particle size in log-log scale encompassing the SW, FSW and EA values. Isotopic endmembers: 3 He/ 4 He (SW) =(4.645±0.008) ×10 –4 (Huss and Lewis, 1994); 3 He/ 4 He =(2.17±0.05) ×10 –4 (Ozima and Podosek, 2002) and 3 He/ 4 He (atmo) =1.384 ×10 –6 (Kurz, 1986). Errors are reported as 1 σ . Fig. 8. ET mass fluxes calculated with by 3 different densities (0.8, 1 and 2.2 g.cm -3 ) from the (a) weighted mean and (b) regression methods. Error bars, reported as 1 σ , integrate the density variability. G. F´ enisse et al. Earth and Planetary Science Letters 663 (2025) 119396 9