Volume 48, Issue 17 e2021GL094301
Research Letter
Free Access

The Mercury Isotopic Composition of Earth's Mantle and the Use of Mass Independently Fractionated Hg to Test for Recycled Crust

Frédéric Moynier

Corresponding Author

Frédéric Moynier

Institut de Physique du Globe de Paris, Université de Paris, CNRS, Paris, France

State Key Laboratory of Geological Processes and Mineral Resources, School of Earth Sciences, China University of Geosciences, Wuhan, China

Correspondence to:

F. Moynier and J. Chen,

[email protected];

[email protected]

Search for more papers by this author
Matthew G. Jackson

Matthew G. Jackson

Department of Earth Science, University of California, Santa Barbara, CA, USA

Search for more papers by this author
Ke Zhang

Ke Zhang

Institute of Surface-Earth System Science, Tianjin University, Tianjin, China

Search for more papers by this author
Hongming Cai

Hongming Cai

Institute of Surface-Earth System Science, Tianjin University, Tianjin, China

Search for more papers by this author
Sæmundur Ari Halldórsson

Sæmundur Ari Halldórsson

Nordic Volcanological Center, Institute of Earth Sciences, University of Iceland, Reykjavík, Iceland

Search for more papers by this author
Raphael Pik

Raphael Pik

Centre de Recherche Pétrographique et Géochimique, CNRS, Université de Lorraine, Nancy, France

Search for more papers by this author
James M. D. Day

James M. D. Day

Scripps Institution of Oceanography, University of California San Diego, La Jolla, CA, USA

Search for more papers by this author
Jiubin Chen

Corresponding Author

Jiubin Chen

State Key Laboratory of Geological Processes and Mineral Resources, School of Earth Sciences, China University of Geosciences, Wuhan, China

Institute of Surface-Earth System Science, Tianjin University, Tianjin, China

Correspondence to:

F. Moynier and J. Chen,

[email protected];

[email protected]

Search for more papers by this author
First published: 25 August 2021
Citations: 18

Abstract

The element mercury (Hg) can develop large mass-independent fractionation (MIF) (Δ199Hg) due to photo-chemical reactions at Earth's surface. This results in globally negative Δ199Hg for terrestrial sub-aerially-derived materials and positive Δ199Hg for sub-aqueously-derived marine sediments. The mantle composition least affected by crustal recycling is estimated from high-3He/4He lavas from Samoa and Iceland, providing an average of Δ199Hg = 0.00 ± 0.10, Δ201Hg = −0.02 ± 0.0.09, δ202Hg = −1.7 ± 1.2; 2SD, N = 11. By comparison, a HIMU-type lava from Tubuai exhibits positive Δ199Hg, consistent with altered oceanic crust in its mantle source. A Samoan (EM2) lava has negative Δ199Hg reflecting incorporation of continental crust materials into its source. Three Pitcairn lavas exhibit positive Δ199Hg which correlate with 87Sr/86Sr, consistent with variable proportions of continental (low Δ199Hg and high 87Sr/86Sr) and oceanic (high Δ199Hg and low 87Sr/86Sr) crustal material in their mantle sources. These observations indicate that MIF signatures offer a powerful tool for examining atmosphere-deep Earth interactions.

Key Points

  • The Hg isotopic composition of the primitive mantle was determined by analyzing lavas from the Samoa and Iceland hotspots

  • Key samples from the canonical mantle end member were analyzed to track crustal recycling in the mantle

  • We demonstrate the presence of recycled oceanic and continental materials in the source of ocean island basalts

Plain Language Summary

While Earth's mantle is continuously chemically and isotopically stirred by convection, some ocean island lavas preserve isotopic anomalies. Their most likely origin is the recycling of crustal material into Earth's mantle by subduction. A question is then whether these crustal materials originate from the ocean or the continents. By using mercury stable isotopic compositions, which have specific signatures in ocean and continent materials, we identify whether these anomalies are due to continental or oceanic crustal material in various ocean island basalts.

1 Introduction

The composition of Earth's mantle is in part known through the chemical and isotopic analyses of lavas from different tectonic settings such as mid-ocean ridge basalts and ocean island basalts (OIB) (e.g., Hofmann, 2013). The variable isotopic compositions of OIB are usually interpreted to reflect mantle heterogeneity formed by recycling of surface material back into the mantle through subduction, the contribution of Earth's core into the deep source of certain lavas, or the survival of early formed heterogeneities in the mantle (e.g., Hauri & Hart, 1993; Hofmann, 1997; Hofmann & White, 1982; Mukhopadhyay & Parai, 2019; Mundl-Petermeier et al., 2020; Rizo et al., 2019). Identifying the geological origin of such mantle sources is paramount to understand mantle dynamics through time.

Mantle heterogeneities have been defined and traditionally traced by using radiogenic isotopes such as Sr, Nd, Hf, and Pb (e.g., Blichert-Toft et al., 19992003; Chauvel et al., 1992; Hofmann, 19881997; White & Hofmann, 1982; Zindler & Hart, 1986), and several end-members have been defined: EM-1 (Enriched-Mantle 1, intermediate 87Sr/86Sr and low 206Pb/204Pb), EM-2 (Enriched-Mantle 2, high 87Sr/86Sr and intermediate 206Pb/204Pb), and HIMU (defined by low 87Sr/86Sr and high 206Pb/204Pb). There is a general consensus emerging for the origin of EM-2 from continental crust materials (e.g., Jackson et al., 2007; White & Hofmann, 1982) and HIMU as oceanic crust and lithosphere or marine carbonates (e.g., Cabral et al., 2013; Chauvel et al., 1992; Hofmann & White, 1982), but the origin of EM-1 is debated (e.g., continental crust, pelagic sediments, delaminated sub-continental lithosphere or a common origin with EM-2 as terrestrial sediments; Castillo, 2017; Delavault et al., 2016; Eisele et al., 2012; Garapic et al., 2015) and understanding its origin represents an important challenge. Stable isotope systems can give complementary information to radiogenic systems, but the potential sources of isotopic fractionation are diverse (e.g., magmatic differentiation, surficial processes), and their interpretation can be somewhat ambiguous.

A few stable isotope systems, the most notable being S, O, and Hg, exhibit mass-independent isotopic fractionations (MIF) in naturally occurring mantle-derived samples that are not strongly modified by high temperature processes (e.g., Cabral et al., 2013; Delavault et al., 2016; Moynier et al., 2020). For example, the presence of MIF sulfur in Mangaia (Cook Islands, Polynesia) and Pitcairn lavas demonstrated the presence of an Archean atmosphere-derived sulfur component in their mantle source (Cabral et al., 2013; Delavault et al., 2016). Mercury is the most volatile among the moderately volatile elements (Lodders, 2003) and exhibits large mass-dependent isotopic fractionations (MDF) and MIF. The MIF-Hg signatures of odd isotopes (odd-MIF) are due to magnetic isotope effects (e.g., Bergquist & Blum, 2007), or a nuclear field shift effect (Estrade et al., 2009) in surficial environments (e.g., Bergquist & Blum, 2007; Blum et al., 2013; Chen et al., 2012; Estrade et al., 2010; Foucher & Hintelmann, 2006; Sherman et al., 2010; Sonke et al., 2010), producing >‰ effects. High temperature processes such as magmatic degassing produce limited (<10 ppm) Hg-MIF (Moynier et al., 2020).

Mercury isotopes are mass-independently fractionated in surface samples and exhibit specific signatures related to different environments, showing potential application to study the nature of recycled materials in the mantle sources of lavas. This is due to the distinct MIF signatures of terrestrial surface environments whereby sub-aerially formed terrestrial reservoirs having generally negative odd-isotopes MIF (e.g., Blum et al., 2014) while the oceanic environment, including sediments and seawater, shows positive odd-isotope MIF (Meng et al., 2019,2020; Yin et al., 2015). In particular, it may be possible to test whether EM1 lavas sample a component of deeply recycled terrestrial sediments (Castillo, 2017; Eisele et al., 2012; Garapic et al., 2015).

A recent landmark study showed that Hg from gold deposits associated with arc magmatism had positive odd-isotope MIF associated with recycling of marine Hg from the subduction zone into the mantle source of arc lavas (Deng et al., 2020). This highlights the utility for Hg isotopes to serve as a tracer of deep recycling processes, from subduction zones to hotspots. Here, we test the origin of the material present in various end-member signatures (EM1, EM2, and HIMU) of OIB using Hg isotopes.

Due to the high volatility of mercury, constraints on its isotopic composition within Earth's mantle are critical for evaluating the origin of terrestrial volatile elements by means of comparison with the isotopic composition of primitive meteorites (Meier et al., 2016; Moynier et al., 2020). However, the Hg isotopic composition of the mantle has been estimated from only four basaltic rocks (Geng et al., 2018; Moynier et al., 2020), and these samples may not be ideally suited for evaluating the composition of the primitive mantle. A primary hurdle for Hg isotopic characterization of mantle-derived lavas is their low Hg contents (a few ppb). We have recently developed a method that yields high precision data on less than 5 ng of Hg, which enables isotope characterization with just a few grams of material (Moynier et al., 2020). This allows targeting of a key set of mantle-derived lavas to evaluate the Hg isotopic composition of the mantle least impacted by crustal recycling, and place key constraints on the origin of the volatile element Hg on Earth.

To evaluate the Hg isotopic composition of the mantle least affected by crustal recycling, we selected lavas with least degassed signatures reflected by their high-3He/4He (>13 atmospheric ratio, RA), including four samples from Samoa (Jackson et al., 2007), and seven from Iceland (Füri et al., 2010 and Halldórsson, Hilton, et al., 2016). To further characterize the composition of the mantle we analyzed three additional Icelandic samples (Óskarsson et al., 1982) and six basalts from the Afar (Ethiopia) (Deng et al., 2018; Pik et al., 2006). To search for deep recycling of surface-derived Hg, samples from the three canonical mantle endmembers thought to reflect deep recycling of crustal materials: EM1 (Pitcairn, N = 3), HIMU (Tubuai, N = 1), EM2 (Samoa, N = 1) were selected.

2 Materials and Methods

2.1 Sample Descriptions

We report isotopic data on international rock standards to assess data quality and intra-laboratory data comparisons: USGS samples: BCR-2 (basalt, USA), RGM-2 (Rhyolite, USA), and STM-2 (syenite, USA) and three Icelandic samples used as internal standards at the University of Iceland (I-ICE, A-ALK, B-ALK, Óskarsson et al., 1982), all of which are available upon request. The Hg isotopic composition of BCR-2 has been reported previously (Geng et al., 2018). One trachyandesite from the Samoa hotspot (ALIA-115-21) (Adams et al., 2021), three basalt samples from Pitcairn (Garapic et al., 2015), and one basalt sample from Tubuai (Hauri & Hart, 1993) were selected to represent the EM2, EM1, and HIMU mantle end-members, respectively. Four samples from Ofu island (Samoa [Jackson et al., 2007]) and seven samples from several different regions (Eastern Rift Zone, Reykjanes Peninsula, Western Rift Zone, and Northern Rift Zone) of Iceland (Halldórsson, Barnes, et al., 2016; Halldórsson, Hilton, et al., 2016; Jackson et al., 2020; Macpherson et al., 2005; Rasmussen et al., 2019) were taken to represent the high-3He/4He mantle (>13RA). Six basalt samples from the Stratoid Series in the Afar hotspot (Deng et al., 2018; Pik et al., 2006) were selected to further estimate the mantle composition. While the 3He/4He of the Afar samples have not been characterized, the composition of samples from this region is ∼10–13 RA (Marty et al., 1996; Medynsky et al., 2013). The UM-Almedén cinnabar standard was also analyzed for inter-laboratory comparison.

2.2 Methods

The method follows the same protocol as in Moynier et al. (2020). The Hg isotopic analysis was carried out using a MC-ICP-MS (Nu-Plasma-3D, Nu-Instruments) (Chen et al., 2010; Huang et al., 2015; Yuan et al., 2018; Zhang et al., 2020). The instrumental mass bias was corrected by an internal NIST-SRM-997-Tl standard using the standard-sample bracketing method. All samples were analyzed in three blocks (33 cycles). A 10-min washout time in between samples ensured blank levels <0.2% of the preceding sample signals. Signal intensities were 2.5 V on 202Hg (1 ppb solution).

The data for Hg-MDF is reported as δxHg:
urn:x-wiley:00948276:media:grl62907:grl62907-math-0001(1)
where x refers to 199, 200, 201, or 202. δ202Hg is used for the discussion of MDF. The MIF data are defined as the deviation of the δxHg from the theoretical value based on kinetic isotopic fractionation (e.g., Blum & Johnson, 2017):
urn:x-wiley:00948276:media:grl62907:grl62907-math-0002(2)
where the mass dependent scaling factor βx is 0.2520, 0.5024, and 0.7520 for 199Hg, 200Hg, and 201Hg, respectively.

Errors are reported as twice the standard deviation from the replicates of the same solution, and are typically 0.08, 0.03, 0.03, 0.05‰ for δ202Hg, Δ199Hg, Δ200Hg, and Δ201Hg, respectively. Long-term measurements of UM-Almadén Hg and CRM-GBW07405 standards give identical values for Hg isotopic composition (both MDF and MIF) to literature (Chen et al., 2010; Moynier et al., 2020; Huang et al., 2019), attesting the quality of Hg isotope analysis.

3 Results

Isotopic data for samples are reported in Table S1 and the Δ199Hg values are reported against δ202Hg values in Figure 1. The isotopic composition of BCR-2 is identical within error to literature values (Geng et al., 2018).

Details are in the caption following the image

Δ199Hg versus δ202Hg for all samples analyzed here. Samples from the Afar, Ofu island, and Iceland have similar Δ199Hg that cluster around zero and variable δ202Hg values. A sample from Tubuai has a positive Δ199Hg value, whereas two Pitcairn samples and one Samoan sample with recycling signatures have negative Δ199Hg.

The terrestrial igneous rock samples show variable δ202Hg values, from −3.23 to −0.01, consistent with literature data (Geng et al., 2018; Moynier et al., 2020) (Figure 1). High 3He/4He localities — Ofu island and Iceland — do not exhibit any MIF, and the Afar samples also lack MIF-Hg signatures. By contrast, the HIMU sample from Tubuai has a positive odd-isotope-MIF (Δ199Hg = +0.23 ± 0.07). The extreme EM2 Samoa ALIA-115-21 sample has a distinct negative odd-isotope-MIF signature (Δ199Hg = −0.14 ± 0.02). In the Pitcairn lava suite, there is a positive correlation between 87Sr/86Sr and MIF-Hg: two Pitcairn lavas with the less extreme EM1 signatures have clear negative odd-isotope-MIF-Hg (Δ199Hg down to −0.45 ± 0.01), while the Pitcairn sample with the most extreme EM1 signature has no resolvable MIF (Figure 2). No samples exhibit even-isotope-MIF signatures.

Details are in the caption following the image

87Sr/86Sr versus Δ199Hg for Pitcairn lavas. The correlation of Δ199Hg values with 87Sr/86Sr suggests a mixture of both recycled oceanic crust (high 87Sr/86Sr and Δ199Hg) and terrigenous materials (low 87Sr/86Sr and Δ199Hg) in the Pitcairn mantle source. 87Sr/86Sr data from Garapic et al. (2015).

4 Discussion

4.1 The Isotopic Composition of the Primitive Mantle

Before discussing the potential effects of recycling of surficial material on the Hg isotopic composition of mantle sources, it is critical to establish the composition of the primitive mantle. OIB with the highest 3He/4He are inferred to sample mantle sources that are least degassed (Craig & Lupton, 1976; Mukhopadhyay & Parai, 2019) and the least modified by crustal recycling (Jackson et al., 2020; White, 2015). Making the assumption that these samples represent materials with Hg isotopic compositions consistent with the most “pristine” mantle domain, we can then establish the impact of crustal subduction and recycling on the Hg systematics of mantle-derived lavas.

The Ofu island samples have high-3He/4He (22RA to 34RA, see Table S1) (Jackson et al., 2007) and are inferred to sample an ancient mantle domain (Mundl-Petermeier et al., 2020) less impacted by crustal recycling than other Samoan lavas (Jackson et al., 2020). Given that the high-3He/4He reservoir is least impacted by recycling, and the discovery that some high-3He/4He lavas preserve early Hadean signatures identified using several different short-lived isotope systems — 129Xe/130Xe (Mukhopadhyay & Parai, 2019), 142Nd/144Nd (Peters et al., 2018), and 182 W/184W (Mundl-Petermeier et al., 2020; Rizo et al., 2019) — this mantle reservoir is most likely to preserve signatures associated with terrestrial accretion. Indeed, arguments have been made that the high-3He/4He reservoir is the oldest domain that has survived in Earth's mantle (Giuliani et al., 2020; Jackson et al., 2010) but alternative views exist (e.g., Mukhopadhyay & Parai, 2019). Therefore, the absence of Hg-MIF in the high-3He/4He samples suggests that Earth's mantle accreted Hg devoid of MIF, and therefore any observed MIF are the consequence of terrestrial fractionation. Combining Hg isotope data from the Ofu (N = 4) with Iceland samples for which 3He/4He are available (N = 7, all of which have 13 < 3He/4He < 34), return an average Hg isotopic composition for the mantle of Δ199Hg = 0.00 ± 0.10 and Δ201Hg = −0.02 ± 0.0.09 (2SD, N = 11). This represents the current best estimate of the terrestrial “primitive mantle” composition. The δ202Hg values are more variable within these samples. Since Hg is highly volatile, degassing during magma ascent can induce isotopic fractionation (e.g., Zambardi et al., 2009), and further isotopic fractionation could also occur during igneous processes. Therefore, compared to MIF-Hg signatures, which should not be significantly impacted by high temperature magmatic and degassing processes, the δ202Hg values are likely more impacted by the processes leading to the formation of the basalts. Nevertheless, the δ202Hg of the 11 high-3He/4He samples from Ofu island and Iceland show limited variation, with an average −1.7 ± 1.2 (2SD, N = 11), which compare well with a previous estimate of the mantle composition based on only a few unrelated igneous rocks (δ202Hg < −2.35, (Moynier et al., 2020), see Figure S1). Furthermore, samples from the Afar (3He/4He ∼ 10–13 Ra, (Marty et al., 1996; Medynsky et al., 2013)) (N = 6) have similar Hg isotopic composition within error, suggesting a large-scale homogeneity of Earth's modern mantle.

Our estimate of the primitive mantle Hg isotopic composition falls within the range defined by chondrites (Figure 3). While the large uncertainties on the Hg concentration of the Earth's mantle prevent an accurate estimate the amount of terrestrial Hg that is stored into the Earth's core, the present chondritic isotopic composition of the Earth's primitive mantle is an argument that late accretion of chondritic material delivered the present mantle Hg (and other volatile elements) (see discussion in Moynier et al., 2020). This conclusion is consistent with observations made from C, N, and noble gas isotopes (Marty, 2012) and the volatile and chalcophile elements S, Se, and Te (Varas-Reus et al., 2019; Wang & Becker, 2013).

Details are in the caption following the image

Mercury isotopic composition of the Earth's mantle (red box, defined by high 3He/4He lavas from Samoa and Iceland) compared to chondrites (CC = carbonaceous chondrites, EC = enstatite chondrites, OC = ordinary chondrites). Mantle-derived lavas exhibit no clear signatures for crustal recycling and fall within the chondritic range.

4.2 Evidence for Hg Recycling?

Considering the Hg isotopic composition of high-3He/4He lavas defined above as representative of the mantle domains the least modified by crustal recycling, we can compare our estimate for the Hg isotopic composition for low-3He/4He mantle endmembers that are suggested to host recycled surface materials — EM1 Pitcairn (EM1), Samoa (EM2), and Tubuai (HIMU) — and then identify Hg isotope signatures associated with recycling in the mantle sources of these lavas.

Three OIB samples examined in this study that represent mantle end-member compositions — TBA-B3 (Tubuai HIMU), ALIA-115-21 (Samoa EM2), and two Pitcairn lavas (Pit-6 and Pit-8) — show clearly resolvable odd-MIF. The correlation between Δ199Hg and Δ201Hg values with a slope of ∼1 (Figure 4), typical of surface photochemical reactions on Hg (Bergquist & Blum, 2007), suggests that the three samples share likely a similar origin for the MIF signatures in surficial environments. The Hg-MIF signatures in OIB therefore provides evidence for recycling of shallow geochemical reservoirs into the deep sources of hotspots. This correlation between Δ199Hg and Δ201Hg values also provides further confidence on the quality of the data as isobaric interference would produce isotopic signatures that lie outside of this range.

Details are in the caption following the image

Δ199Hg versus Δ201Hg for the Earth's mantle (red box, defined by high 3He/4He lavas from Samoa and Iceland), Pitcairn, Tubuai, and Samoa lavas. HIMU (Tubuai), Enriched-Mantle 1 (EM-1) (Pitcairn) and EM-2 (Samoa) deviates from the mantle composition, reflecting a contribution of recycled surface materials in their mantle sources (marine material for Tubuai, continental material for Samoa, and both marine and continental material in the Pitcairn mantle).

Photochemical reduction of aqueous Hg(II) to Hg (0) vapor is the major source of odd-MIF, and is the main pathway of transfer of Hg from the ocean (surface) to the atmosphere (e.g., Bergquist & Blum, 2007). During the photochemical reduction, a positive Δ199Hg [Hg(II)] is retained in the marine environment while the complementary negative Δ199Hg [Hg (0)] is released into the atmosphere (e.g., Bergquist & Blum, 2007). Given the short residence time of Hg (0) in the atmosphere (∼1 year), its quick subsequent deposition confers a globally negative Δ199Hg to the terrestrial reservoirs, while marine environments are characterized by positive Δ199Hg (e.g., Biswas et al., 2008; Blum et al., 2014; Demers et al., 2015; Grasby et al., 2017; Jiskra et al., 2017). How a negative Δ199Hg preserved in terrestrial material is then transported into the ocean through time is unclear at present. Further work is required to test the robustness of this negative signature on long timescales necessary for this material to be recycled within the mantle. This difference between terrestrial and marine settings is well reflected in sediments, with generally negative Δ199Hg in coastal sediments (which originate via erosion of nearby terrestrial materials) and positive Δ199Hg for open sea marine sediments (e.g., Meng et al., 2019; Yin et al., 2015). The odd-MIF of ALIA-115-21 (Samoa EM2), TBA-B3 (Tubuai HIMU), and Pit-6 and Pit-8 (Pitcairn) therefore likely reflect a marine Hg isotopic signature in Tubuai, and continental Hg signatures Samoan EM2 lavas (Figure 4). As discussed below, Pitcairn lavas exhibit Hg isotopic evidence for recycled marine and terrestrial materials.

The presence of marine and continental signatures in the mantle sources of Tubuai HIMU and Samoa EM2 lavas, respectively, is consistent with traditional interpretations based on radiogenic isotope signatures. Tubuai exhibits a HIMU signature, which is considered to result from recycling of ancient oceanic crust into the mantle (Hofmann & White, 1982). The geochemically depleted 87Sr/86Sr and 143Nd/144Nd signatures in HIMU lavas globally precludes a significant sediment (terrigenous or marine) contribution to their mantle sources, but transfer of positive Δ199Hg isotope compositions of seawater to oceanic crust during hydrothermal circulation at a mid-ocean ridge could explain the MIF signatures in Tubuai HIMU OIB. A potential caveat with this model is the low abundance of Hg in seawater compared to fresh basalts that would require a water/rock ratio of ∼1,000 to transfer the Hg isotopic composition of seawater to oceanic crust. However, the absence of data on Hg behavior during seawater alteration of oceanic crust (e.g., Hg concentration measurements on altered oceanic crust, including serpentinites) limits our ability to quantitatively model this scenario. An additional complexity is that a fraction of the slab Hg budget would be lost during subduction dehydration, but relevant data are not yet available to quantify this. By contrast, Samoan EM2 lavas exhibit geochemically enriched (high)87Sr/86Sr and (low) 143Nd/144Nd signatures consistent with the input of recycled terrigenous materials (Jackson et al., 2007; Workman et al., 2008), consistent with the slightly negative Δ199Hg terrestrial signature.

The presence of negative Δ199Hg in the source of Pitcairn EM1 OIB suggests that, like EM2, EM1 contains a fraction of recycled terrigenous material (Castillo, 2017; Delavault et al., 2016; Stracke, 2012). However, there is complexity in the MIF present in the Pitcairn suite that is not explained by our simple model: the correlation between the Δ199Hg and 87Sr/86Sr for the three Pitcairn samples (Figure 2) is puzzling, as we would expect the sample with the highest 87Sr/86Sr to host the largest fraction of continent-derived material and, therefore, the lowest Δ199Hg. Instead, the Pitcairn lava with the weakest EM1 signature has the strongest negative Δ199Hg, and the most extreme EM1 Pitcairn lava has no resolvable Δ199Hg signature. The positive correlation between Δ199Hg and 87Sr/86Sr is unlikely to be explained by seawater alteration, a mechanism that would increase both Δ199Hg and 87Sr/86Sr. This is because (a) Δ199Hg also exhibit a negative correlation with 143Nd/144Nd (Figure S2), a radiogenic isotope system immune to seawater influence, and (b) the selected Pitcairn lavas are young (<1.0 Ma; Duncan et al., 1974) and fresh (i.e., no signs of visible alteration). Therefore, we explore a mantle source origin for the correlation between 87Sr/86Sr and Δ199Hg, acknowledging that this relationship is only for three samples.

Unlike Samoan EM2 (which is a melt of a peridotite mantle source that hosts recycled terrigenous material) (Jackson & Shirey, 2011) and Tubuai HIMU (which is a melt of a mantle source with recycled oceanic crust) (Hauri & Hart, 1993), extreme Pitcairn EM1 is suggested to host both recycled oceanic crust and terrigenous materials (Eisele et al., 2003). A mixture of both recycled oceanic crust and terrigenous materials can help to explain the correlation between Δ199Hg and 87Sr/86Sr (Figure 2), where Pitcairn's extreme EM1 endmember (highest 87Sr/86Sr) lacks a clear odd-MIF, and the Pitcairn lava with the weakest EM1 signature (lowest 87Sr/86Sr) has a negative Δ199Hg. Because recycled oceanic crust and terrigenous materials have complementary positive and negative Δ199Hg signatures, respectively, it is possible for their mixture to generate in a near-zero Δ199Hg value like that observed in the extreme Pitcairn EM1 sample. In this case, the presence of terrigenous materials explains the high 87Sr/86Sr in the extreme EM1 Pitcairn lava. In contrast, the Pitcairn lava with the lowest 87Sr/86Sr (weakest EM1 signature) is explained by a smaller fraction of recycled terrigenous material, and the very negative Δ199Hg in this lava can explained by the absence (or near absence) of oceanic crust (which would have a positive Δ199Hg values): the remaining contribution of recycled terrigenous material has strongly negative Δ199Hg, and explains the negative Δ199Hg values determined in lavas. In this way, the relative proportions and absolute abundances of recycled terrigenous and oceanic crust materials can be modulated to explain the magnitude and sign of the Δ199Hg values, and its relationship with radiogenic isotopes.

5 Conclusions

To estimate the Hg isotopic composition of the mantle that is least affected by crustal recycling, we analyzed high-3He/4He lavas from Samoa and Iceland. We obtained an average Δ199Hg = 0.00 ± 0.10, Δ201Hg = −0.02 ± 0.0.09 and δ202Hg = −1.7 ± 1.2 (2SD, N = 11), which is the best estimate for the composition of the primitive mantle.

HIMU lavas from Tubuai exhibit positive Hg-MIF, likely reflecting a transfer of the isotope compositions of seawater to oceanic crust during hydrothermal circulation at a mid-ocean ridge; positive Hg-MIF therefore signals the presence of altered oceanic crust in its mantle source. The negative MIF observed for EM2 lava from Samoa is consistent with assimilation of continental crust material. The negative MIF for the Pitcairn lavas, and the positive correlation of Δ199Hg values with 87Sr/86Sr, provides an independent argument in favor of mixture between terrestrial and marine materials in the mantle source of EM1.

Acknowledgments

Runsheng Yin and Michael Bizimis are thanked for insightful reviews that have improved this manuscript. This study was supported by the National Natural Science Foundation of China (41625012, 41961144028, U1612442, 41830647) to J. B. Chen and by the China National Space Administration (No. D020205) to ZW. F. Moynier thanks Yongsheng Liu for making the visit at CUG Wuhan possible. M. G. Jackson thanks IPGP for providing a stimulating academic home during sabbatical, and acknowledges partial support from NSF EAR-1900652 and OCE-1928970. Frédéric Moynier acknowledge funding from the UnivEarthS Labex program (numbers: ANR-10-LABX-0023 and ANR-11-IDEX-0005-02) and parts of this work were supported by IPGP multidisciplinary program PARI, and by Region île-de-France SESAME Grant (no. 12015908).

    Data Availability Statement

    All the data are provided in Table S1 as well as on Mendeley.